Crystal Structure and Mutagenesis of a Protein Phosphatase-1:Calcineurin Hybrid Elucidate the Role of the β12-β13 Loop in Inhibitor Binding*

Protein phosphatase-1 and protein phosphatase-2B (calcineurin) are eukaryotic serine/threonine phosphatases that share 40% sequence identity in their catalytic subunits. Despite the similarities in sequence, these phosphatases are widely divergent when it comes to inhibition by natural product toxins, such as microcystin-LR and okadaic acid. The most prominent region of non-conserved sequence between these phosphatases corresponds to the β12-β13 loop of protein phosphatase-1, and the L7 loop of toxin-resistant calcineurin. In the present study, mutagenesis of residues 273-277 of the β12-β13 loop of the protein phosphatase-1 catalytic subunit (PP-1c) to the corresponding residues in calcineurin (312-316), resulted in a chimeric mutant that showed a decrease in sensitivity to microcystin-LR, okadaic acid, and the endogenous PP-1c inhibitor protein inhibitor-2. A crystal structure of the chimeric mutant in complex with okadaic acid was determined to 2.0-Å resolution. The β12-β13 loop region of the mutant superimposes closely with that of wild-type PP-1c bound to okadaic acid. Systematic mutation of each residue in the β12-β13 loop of PP-1c showed that a single amino acid change (C273L) was the most influential in mediating sensitivity of PP-1c to toxins. Taken together, these data indicate that it is an individual amino acid residue substitution and not a change in the overall β12-β13 loop conformation of protein phosphatase-1 that contributes to disrupting important interactions with inhibitors such as microcystin-LR and okadaic acid.

* This work was supported by Grant MGP-37770 from the Canadian Institutes of Health Research (to M. N. G. J.) and by equipment funds from the Alberta Heritage Foundation for Medical Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Before the structure of OA-bound to PP-1c had been determined, it had been hypothesized that the different conformations of the L7 loop in calcineurin and the corresponding ␤12-␤13 loop in PP-1c were the reason for the resistance of calcineurin to toxins (17,20). However, upon superimposition of the PP-1c-OA and calcineurin A structures, it was noted that the backbone atoms of the two loops superimpose. This led to our hypothesis that the resistance of calcineurin to toxins was mainly because of primary sequence differences in the C-terminal end of the L7 loop of calcineurin and the ␤12-␤13 loop of PP-1c/PP-2Ac (19). This solidifies the idea that the differences seen in the MCLR-bound PP1-c structure are not typical and are caused by the covalent reaction (14). To test our hypothesis concerning the importance of the primary sequence on inhibitor potential, we created a chimeric mutant of PP-1c in which the ␤12-␤13 loop residues of PP-1c ( 273 CGEFD 277 ) were substituted with the L7 loop residues of calcineurin A ( 312 LDVYN 316 ). We analyzed this PP-1c-CaN mutant via crystallographic and kinetic studies to probe further the reason for PP1-c sensitivity and the resistance to toxins and inhibitor proteins for calcineurin.

EXPERIMENTAL PROCEDURES
Expression and Purification of Recombinant PP-1c-Mutants of PP-1c␥ were expressed and purified to homogeneity as in Refs. 20 and 21. The catalytic subunit of the human wild type and mutant PP-1c was expressed in Escherichia coli DH5␣ strain using the plasmid pCW and subsequently purified to homogeneity using heparin affinity, Mono Q, and Superdex 75 gel filtration chromatography as previously described (20,22,23) with the following modifications. A single colony of E. coli DH5␣ cells transformed with plasmid pCW-HPP-1␥ was used to inoculate 100 ml of Luria-Bertani media containing 1 mM MnCl 2 and ampicillin (54 M). After overnight growth at 37°C, this culture was used to inoculate 1 liter of Luria-Bertani media containing 1 mM MnCl 2 and ampicillin (54 M) and grown to optical density 0.5 at 600 nm. Expression was then induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for up to 18 h. Cells were harvested by centrifugation for 45 min at 4,000 ϫ g and either frozen at Ϫ70°C or used immediately. Bacterial cells from two 1-liter cultures were resuspended in 80 ml of buffer A (50 mM imidazole, pH 7.5, 0.5 mM EDTA, 0.5 mM EGTA, 2.0 mM MnCl 2 , 0.5 mM phenylmethylsulfonyl fluoride, 2.0 mM benzamidine, 3.0 mM dithiothreitol, and 10% glycerol (v/v)) containing 100 mM NaCl, sonicated, and centrifuged for 60 min at 13,000 ϫ g. The supernatant was loaded onto a heparin-Sepharose CL-6B column (Amersham Biosciences). PP-1c was eluted in buffer A by a linear 400-ml gradient of 0.1-0.6 M NaCl. Activity of fractions was determined by the colorimetric paranitrophenol phosphate assay (Sigma). Active fractions were pooled and diluted to 3ϫ volume in buffer B (50 mM imidazole, pH 7.2, 0.5 mM EDTA, 0.5 mM EGTA, 3.0 mM MnCl 2 , 2.0 mM phenylmethylsulfonyl fluoride, 2.0 mM benzamidine, 3.0 mM dithiothreitol, and 10% glycerol (v/v)). The fractions were then loaded onto a Mono Q HR 10/10 column (Amersham Biosciences), and eluted in buffer B by a 160-ml linear gradient of 50 -400 mM NaCl. Active fractions were pooled and concentrated to a 2-ml volume in a Centriprep-10 concentrator (Amicon) before loading onto a HiLoad Superdex 75 HR 26/60 column (Amersham Biosciences) equilibrated in buffer C (50 mM imidazole, pH 7.2, 0.1 mM EDTA, 0.1 mM EGTA, 2.0 mM MnCl 2 , 2.0 mM benzamidine, 3.0 mM dithiothreitol, 10% glycerol (v/v)) plus 0.3 M NaCl). The PP-1c was eluted by an isocratic 180-ml gradient of buffer C plus 0.3 M NaCl.
Active fractions, adjudged homogeneously pure by SDS-PAGE, were pooled and concentrated to ϳ0.2-1.0 mg/ml in a Centriprep-10 concentrator. An equal volume of glycerol was added and the solution stored at Ϫ20°C. The specific activities of the wild type and mutant enzymes were measured using the phosphorylase a assay (described below), and found to range between 2 and 11 units/mg.
Phosphatase Assays-32 P-Labeled phosphorylase a was used as a phosphatase substrate to determine dose-response curves with microcystin-LR, okadaic acid, and inhibitor-2. Purified PP-1c stocks were diluted in phosphatase assay buffer (50 mM Tris-HCl, pH 7.0, 0.1 mM EDTA, 1 mg/ml bovine serum albumin, 0.8 mM MnCl 2 , and 0.2% ␤-mercaptoethanol). Phosphatase was diluted until a 10-l aliquot of diluted enzyme solution caused 15% 32 P release in a 30-l assay (20,21). Duplicate assays were performed for each toxin and inhibitor protein. Typically, less than 5% variation was seen between duplicates. Microcystin-LR was obtained from Health Canada, okadaic acid from Moana Bioproducts, and recombinant inhibitor-2 was purified from E. coli. The loop mutant protein phosphatase-1 exhibited normal activity toward phosphorylase a as a substrate.
Crystallization of PP1-loop⅐OA Complex-Crystals were obtained by the hanging drop vapor diffusion method at room temperature. The enzyme and okadaic acid were mixed in a 1:2 molar ratio with the concentration of protein being ϳ0.4 mM. The PP1-c⅐OA complex was then mixed with an equivalent volume of mother liquor, which consisted of 2 M lithium sulfate, 100 mM Tris (pH 8.0), 2% polyethylene glycol 400, and 10 mM ␤-mercaptoethanol. The complex crystallized in the space group P4 2 2 1 2 with cell dimensions a ϭ b ϭ 98.8 Å, c ϭ 62.2 Å, with one complex per asymmetric unit.
Data Collection, Structure Determination, and Refinement-A data set to 2.0 Å was collected at 100 K on a Rigaku RU-H3R rotating anode generator equipped with a Rigaku R-AXIS IVϩϩ image plate detector. The data were processed with the HKL suite of programs (24). This initial structure was solved by molecular replacement with the program AMoRe (25), using the PP-1c-OA structure (with OA removed) as a search model (Protein Data Bank number 1JK7). Electron density for both the protein and the inhibitor were clear from the initial map generated from the molecular replacement solution. OA was fitted to  30 a Data in parentheses correspond to highest resolution shell. b R sym ϭ ⌺ I Ϫ ͗I͘ /⌺I, where I is the observed intensity, and ͗I͘ is the average intensity obtained from multiple observations of symmetry related reflections.
and ͉F c ͉ are the observed and calculated structure factor amplitudes, respectively. d R free was calculated as for R cryst with 5% of the data omitted from structural refinement. the difference density using the structure of OA bound to wild type PP-1c as a starting model. The protein-inhibitor model was subjected to rigid body refinement in CNS prior to manually fitting the model using the program XtalView (26,27). The model was then subjected to iterative rounds of macromolecular refinement using CNS with a maximum likelihood target. The crystallographic data are listed in Table I. The final model consisted of density for residues 6 -298 and was checked for validity using WHATCHECK and PROCHECK (28,29). PROCHECK showed that 97% of residues were in allowed Ramachandran plot ranges with an overall G factor of 0.2.
Coordinates-The atomic coordinates and structure factors have been deposited in the Protein Data Bank with code 1U32.

Inhibition and Substrate Specificity of Wild-type and Loop
Mutant PP-1c-To understand the role of the ␤12-␤13 loop region of PP-1c in inhibitor selectivity further, we examined the inhibitory effects of both natural product inhibitors and an endogenous inhibitor protein (inhibitor-2) on a mutant of PP-1c in the ␤12-␤13 loop region (Table II). This mutant (herein referred to as the "loop mutant") has residues 273-277 of PP-1c substituted with the corresponding residues of calcineurin A (residues 312-316) (Fig. 1). This loop mutant has allowed us to investigate the role of this non-conserved portion of the type 1, 2A, and 2B (calcineurin) phosphatases in inhibitor selectivity.
Wild-type PP-1c was inhibited by okadaic acid with an IC 50 of 30 nM and microcystin-LR with an IC 50 of 0.1 nM. The loop mutant was ϳ4-fold less sensitive to okadaic acid (IC 50 ϭ 110 -120 nM) and at least 6-fold less sensitive to microcystin-LR (IC 50 ϭ Ͼ0.6 nM). Inhibitor-2 inhibited PP-1c with an IC 50 of 1-3 nM. The loop mutant was less sensitive to inhibitor-2 inhibition (IC 50 ϭ 5-6 nM). The resistance of the loop mutant of PP-1c to all inhibitors studied is consistent with the hypothesis that the residues of the ␤12-␤13 loop region of PP-1c play a role in inhibitor selectivity.
Crystal Structure of PP-1c-loop⅐OA Complex-The structure observed of the loop mutant PP-1c bound to okadaic acid is very similar to other PP-1c structures (0.22 Å C ␣ r.m.s. deviation to wild-type PP-1c:OA, 0.55/0.54 Å C ␣ r.m.s. deviation to PP-1c: MCLR (two in the asymmetric unit), 0.51/0.53 Å C␣ r.m.s. deviation to PP-1c:tungstate (two in the asymmetric unit), 0.45/ 0.48 Å C␣ r.m.s. deviation to PP-1c:calyculin (two in the asymmetric unit), all measurements taken over 284 C␣ atoms). The orientation of OA in the active site is indistinguishable from that seen with the wild-type enzyme, with the OA adopting a cyclic conformation via an intramolecular hydrogen bond between the C 1 -acid and the C 24 -hydroxyl (Fig. 2). Similarly, all contacts between the enzyme and the inhibitor are seen in both the wild-type and loop mutant complexes. Namely, this involves the hydrophobic double ring spiroketal moiety of OA binding into the hydrophobic cleft of PP-1c created by residues Ile 130 , Ile 133 , Trp 206 , and Val 223 and hydrogen bonding interactions between Tyr 272 and the C 1 -acid and C 2 -hydroxyl, Arg 221 and the C 24 -hydroxl and Arg 96 (N ⑀ ) and the C 2 -hydroxyl. The conclusion that hydrophobic energy drives complex formation is supported by this structure because there are very few pro- tein-inhibitor interactions (given the size of the inhibitor) and the strength of these interactions is on a par with the interactions seen in the wild-type PP-1c⅐okadaic acid complex (15).
The residue changes within the ␤12-␤13 loop involve substitution of the sequence 273 CGEFD 277 with LDVYN from cal-cineurin A. Because the orientation of the backbone atoms in both the wild-type and the loop mutant remains the same, this leaves side chain alterations as possible reasons for inhibitor specificity. Although these substitutions may be expected to affect overall negative charge in the ␤12-␤13 loop of the chi- meric mutant, the two most prominent residues that have potential inhibitor interactions are Leu 273 and Tyr 276 . The leucine (Leu 273 ) lies near a hydrophobic area of OA (C 10methyl, Fig. 2), potentially securing a hydrophobic section of the molecule. With respect to the position of Tyr 276 , the equivalent residue in wild-type PP-1c (Phe 276 ) lies near the same hydrophobic section of OA, again potentially securing this area. The electron density for Phe 276 is very well defined in wild-type PP-1c; it has an average side chain thermal factor of 24 Å 2 above the average thermal factor for the molecule (Fig. 3). In the loop mutant the electron density of Tyr 276 is very poorly defined with an average side chain thermal factor of 31 Å 2 above the average thermal factor for the rest of the molecule (in comparison all four structures of calcineurin have values for the equivalent residue (Tyr 315 ) that are maximally 19 Å 2 above the average thermal factor for the rest of the enzyme). The tyrosine side chain in this position would not have the potential for hydrophobic interactions that the phenylalanine in wildtype PP-1c possesses. The conclusion that hydrophobic energy drives inhibitor binding in the phosphoprotein phosphatase M family may involve binding of the hydrophobic tail (that many of the inhibitors contain) and also a hydrophobic binding closer to the active site involving residue 276.
Inhibitor Binding Is Partially Determined by Residues of the ␤12-␤13 Loop-Our kinetic data show that the L7 loop of calcineurin A and ␤12-␤13 loop of PP-1c partially determine inhibitor specificity. Within this loop, there are few candidate residues that could confer this specificity. Tyr 272 has the greatest potential for impact because it forms the most intimate contacts with the inhibitors, hydrogen bonding with the C 1 acid moiety of OA, the phosphate group of calyculin A, and the acidic moiety in MCLR. Tyr 272 is also important in binding of calyculin A and tautomycin because the mutation Y272F causes a 200-and 400-fold decrease in sensitivity to these toxins (11). This residue is, however, conserved as a tyrosine in both PP-1c and calcineurin A. To determine the individual contributions of the remaining residues within the ␤12-␤13 loop toward inhibitor resistance, we undertook to mutate systematically and express the single point mutations of residues 273-277 from PP-1c to their corresponding residues in calcineurin A.

DISCUSSION
Protein phosphatase-1 and protein phosphatase-2B (calcineurin) share 40% sequence identity in their catalytic sub-units. Despite the similarities in sequence, these phosphatases are widely divergent when it comes to inhibition by both endogenous proteins, such as protein phosphatase inhibitor-2, and natural product toxins, such as microcystin-LR and okadaic acid. One of the more prominent regions of non-conserved sequence between these phosphatases corresponds to the ␤12-␤13 loop of protein phosphatase-1, and the L7 loop of calcineurin. In the present study, mutagenesis of residues 273-277 of the ␤12-␤13 loop of protein phosphatase-1 to the corresponding residues in calcineurin (312-316), resulted in a chimeric mutant that showed a decrease in sensitivity to microcystin-LR, okadaic acid, and inhibitor-2.
A crystal structure of the chimeric mutant in complex with okadaic acid was determined to 2.0-Å resolution. The purpose of determining this structure was to observe directly the differences in the ␤12-␤13 loop structure between wild-type PP-1c and the loop mutant PP-1c, both bound to OA. This was examined in the context of the large structural changes seen in the PP-1c⅐MCLR complex that may have implicated structural changes in the loop as being indicative of inhibitor specificity. However, the backbone conformation of the loop (and the entire active site) is virtually unchanged in both the wild-type com- plex and the loop mutant (Fig. 7). When compared in a larger context, the loop has an extremely similar orientation in the tungstate-, OA-, and calyculin-bound forms. More strikingly, the L7 loop of calcineurin also has a virtually identical conformation to all these structures. The only anomaly is the PP-1c: MCLR structure where the covalent linkage to Cys 273 may affect the loop position. This supports the proposition the that primary sequence, rather than the overall ␤12-␤13 loop structure, determines the inhibitor specificity.
A change in IC 50 with chimeras of the ␤12-␤13 loop region was reported before by Connor et al. (7), when they substituted the C terminus of PP-1c (residues 274 onwards) with the corresponding C terminus of PP-2Ac (7). The chimeric protein produced was not inhibited by either NIPP-1 or thiophosphorylated inhibitor-1 (neither of which inhibit PP-2A). Interestingly, the chimeric protein produced had the dose-response characteristics of PP-1 with respect to fostriecin and tautomycin, suggesting that the ␤12-␤13 loop region is less important for inhibition by these natural products.
We undertook to systematically mutate all residues in the ␤12-␤13 loop region of PP-1c to their corresponding residues in calcineurin A. The point mutants created were PP-1c C273L, G274D, G275V, F276W, and D277N. We did not mutate Tyr 272 as this residue is a Tyr in calcineurin A and our structural data showed that it adopted an identical conformation in the loop mutant PP-1c⅐OA complex (Fig. 7). The necessity to create single mutations in the ␤12-␤13 loop region of PP-1c was suggested following the finding of a poorly defined electron density for Tyr 276 in the loop mutant crystal structure. The corresponding residue in PP-2Ac, Cys 269 , has already been found to be important for okadaic acid sensitivity. Mutation of Phe 276 to Cys in PP-1c increases the sensitivity of PP-1c to okadaic acid 40-fold, indicating that an even smaller hydrophobic residue at this position may facilitate tighter binding of okadaic acid (10).
Systematic mutation of each residue in the ␤12-␤13 loop of PP-1c to their corresponding calcineurin A residue showed that a single amino acid change (C273L) was the most influential in mediating sensitivity of PP-1c to the toxin inhibitors microcystin-LR and okadaic acid. Taken together, these data indicate that an individual amino acid residue substitution (C273L, as occurs in calcineurin) and not a change in the overall ␤12-␤13 loop conformation of PP-1c contributes to disrupting important interactions with inhibitors such as microcystin-LR and okadaic acid. MacKintosh et al. (18) first showed that Cys 273 binds covalently to microcystin-LR and determined that abolition of covalent binding (via mutation to Ala, Ser, or Leu) increased the IC 50 for toxin inhibition of PP-1c by 5-20-fold. This is entirely consistent with our mutational analyses of the ␤12-␤13 loop presented here. However, we have also shown that covalent modification of Cys 273 by microcystin-LR occurs very slowly over several hours and only after the phosphatase is fully inhibited by the toxin (17,19). Taken together, the findings of these studies suggest that covalent modification of Cys 273 is adventitious with respect to inhibition of the phosphatase and that this residue clearly plays a role in the initial inhibition of PP-1c by the microcystin toxins (19).
Consistent with its important role in toxin interactions, Cys 273 is conserved in virtually all known members of the PP-1c family. A notable exception being PP-1c from Dictyostelium discoideum (social amoeba) (30), where it is changed to Phe. The sensitivity to microcystin inhibition of this form of PP-1c was reported as ϳ2-fold less than a "mutant" containing Cys 269 (the equivalent residue in this organism). It is possible that Phe 273 is more favorable for toxin inhibition in this position than Leu 273 (present study and MacKintosh et al. (18)), however, these experiments were conducted on unpurified recombinant protein from crude E. coli extracts. PP-1c from several plants, including Brassica napus, Arabidopsis thaliana TOPP 8, and Phaseolus vulgaris, have Cys 273 changed to Gly (31)(32)(33). Crude extract preparations of B. napus PP-1c were reported to be as sensitive to inhibitors as mammalian PP-1c (34, 35), however, it will be interesting to see whether homo-geneous preparations of highly purified PP-1c from these organisms have any resistance to microcystin inhibition.
In contrast to the role of Cys 273 in mediating toxin inhibition of PP-1c, replacement of Cys 273 with Leu did not affect inhibition of PP-1c by inhibitor-2. The only substitution that affected inhibitor-2 inhibition was F276Y, clearly indicating that the mode of inhibition of PP-1c by this endogenous inhibitor protein is distinct from that of the natural product/toxin inhibitors. The effect of substitutions in the ␤12-␤13 loop of PP-1c on inhibition by inhibitor-2 are difficult to interpret because the three-dimensional structure of this protein bound to PP-1c is unknown. The differences in sensitivity to the inhibitor protein 2 between wild-type PP-1c and calcineurin A are much more dramatic than the differences between wild-type PP-1c and loop mutant PP-1c, clearly indicating that the ␤12-␤13 loop of PP-1c is not the only region playing a role in binding of this inhibitor. Taken together, these data are consistent with the most recent models of inhibitor-2 binding (8).
The importance of the ␤12-␤13 loop was emphasized in a recent structure of the complex between PP-1c and a 34-kDa fragment from myosin phosphatase targeting subunit (MYPT-1) (36). This structure showed some active site remodeling with targeting subunit binding to the catalytic subunit but left the ␤12-␤13 loop in a prominent position to influence enzymatic activity and inhibition. Elucidation of the sequencespecific role of the ␤12-␤13 loop of PP-1c in mediating inhibition of this enzyme by toxins will now facilitate rational protein engineering of calcineurin A that is sensitive to toxin inhibition. This would be an essential first step in designing active site-directed inhibitors that would be specific for calcineurin.