Activation of protein phosphatase 1. Formation of a metalloenzyme.

The recombinant catalytic subunit of protein phosphatase 1 is produced as an inactive enzyme which can be activated by Mn2+ (Zhang, Z., Bai, G., Deans-Zirattu, S., Browner, M. F., and Lee, E. Y. C.(1992) J. Biol. Chem. 267, 1484-1490). In this report, we have investigated the effects of divalent cations on the activity of recombinant catalytic subunit of protein phosphatase 1. Latent phosphatase 1 can be activated by Co2+ or Mn2+, whereas other metal ions tested including Fe2+, Zn2+, Mg2+, Ca2+, Cu2+, or Ni2+ were not effective or were only weakly effective in activating the enzyme. The Mn2+-stimulated activity was susceptible to inactivation by EDTA; however, the Co2+-activated phosphatase was stable after dilution and chelation of the Co2+ with excess EDTA. After stable activation of phosphatase 1 using 57Co2+, a stoichiometric amount of 57Co2+ was shown to be tightly bound to phosphatase 1. These findings demonstrate for the first time the generation of a stable metalloenzyme form of phosphatase 1. Fe2+ reversibly deactivated the Co2+-stimulated activity, but did not displace the bound Co2+. Interestingly, treatment of the enzyme with a combination of Fe2+ and Zn2+ (but not the individual metal ions) significantly activated phosphatase 1. These results suggest that at least two metal binding sites exist on the enzyme and that protein phosphatase 1 may be an iron/zinc metalloprotein in vivo.

Protein phosphatase 1 (phosphorylase phosphatase), one of the four major Ser/Thr protein phosphatases, has been studied mainly in relation to its central role in the regulation of glycogen metabolism (for reviews, see Bollen and Stalmans (1992), Shenolikar and Nairn (1991), and Shenolikar (1994)). The enzymology of the enzyme is complex and involves multiple forms of the enzyme generated by combinations of a 37-kDa catalytic subunit (PP1) 1 with different regulatory proteins that may also provide for molecular targeting of the enzyme. Several regulatory subunits have been well characterized, including inhibitor 2, the glycogen binding subunit, a nuclear inhibitory subunit, and myofibril binding subunits (Bollen and Stalmans, 1992;Shimizu et al., 1994;Chen et al., 1994). Most of the previous studies of the isolated catalytic subunit have been of an active enzyme that is independent of metal ions for its activity. However, it has been clear for a number of years that there exists a metal ion dependent form or forms of PP1. In the ATP/Mg-dependent enzyme, which is a 1:1 complex of PP1 with inhibitor 2, PP1 is present as an inactive or latent enzyme that is reversibly stimulated by Mn 2ϩ (Villa-Moruzzi et al., 1984). All recombinant forms of PP1 expressed in Escherichia coli, including the four known isoforms, are dependent on Mn 2ϩ for activity (Zhang et al., 1992(Zhang et al., , 1993aAlessi et al., 1993). Zhang et al. (1993b) have suggested that the recombinant enzyme represents the conformer that is present in the PP1-inhibitor 2 complex. We have recently isolated a form of PP1 catalytic subunit from cardiac muscle which is inactive, but can be converted to a stable active form by exposure to Co 2ϩ (Chu et al., 1994).
Thus, there are complex and not completely understood facets of the nature of the differences between these forms of the PP1 catalytic subunit, which are revealed by the effects of divalent cations on its activity. The question of how metals affect phosphorylase phosphatase is an old one. It has long been known that divalent cations, in particular Mn 2ϩ and Co 2ϩ , can activate certain phosphorylase phosphatase preparations (Merlevede and Riley, 1966;Kato and Bishop, 1972;Kato et al., 1975;Ullman and Perlman, 1975;Khatra and Soderling, 1978;Khandelwal and Kasmani, 1980;Brautigan et al., 1980Brautigan et al., , 1982. It has been suggested that the phosphatase present in these preparations is a metalloenzyme (Burchell and Cohen, 1978;Hsiao et al., 1978;Khatra and Soderling, 1978;Defreyn et al., 1979;Mackenzie et al., 1980). However, attempts to demonstrate the presence of bound metal in enzyme preparations have been negative. Metal analysis of a preparation of liver PP1 by atomic absorption showed only small substoichiometric amounts of Ca 2ϩ , Cd 2ϩ , Co 2ϩ , Cu 2ϩ , Fe 2ϩ , Mg 2ϩ , Mn 2ϩ , Ni 2ϩ , Sn 2ϩ , or Zn 2ϩ (Yan and Graves, 1982). Moreover, activation by 54 Mn 2ϩ of the catalytic subunit of ATP/Mg 2ϩ -dependent protein phosphatase (Villa-Moruzzi et al., 1984) or a preparation containing a high molecular weight form of PP1 (Brautigan et al., 1980) did not show significant incorporation of 54 Mn 2ϩ into the enzyme.
Using recombinant PP1, we have explored the issue of whether it may exist as a metalloprotein. In this study, we demonstrate that the activation of PP1 by 57 Co 2ϩ is associated with a stoichiometric incorporation of Co 2ϩ into the enzyme. We also report that PP1 is activated by a combination of Fe 2ϩ / Zn 2ϩ and we suggest that PP1 may be an iron/zinc metalloenzyme in vivo.

EXPERIMENTAL PROCEDURES
Materials-57 CoCl 2 was obtained from ICN. Superose-12 was from Pharmacia Biotech Inc. The recombinant PP1 used in these studies was the PP1␣ isoform expressed in E. coli as described by Zhang et al. (1992). [ 32 P]Phosphorylase a was prepared as described previously (Killilea et al., 1978).
Preparation of Metal-depleted Recombinant PP1-Homogeneous re-* This work was supported by National Institutes of Health Grants HL 36576 (to K. K. S.) and DK 18512 (to E. Y. C. L.). 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.
Assay for Phosphatase Activity-PP1 was preincubated with divalent cations as indicated. The preincubated enzyme was then assayed directly (without chelation of the cation) for phosphatase activity by measuring the release of [ 32 P]P i from [ 32 P]phosphorylase a (Chu et al., 1994). One unit of phosphatase activity was defined as 1 mol of [ 32 P]P i released/min.
Protein Determinations-Protein concentration was determined by the method of Bradford (1976) using BSA as a standard.

Effects of Divalent Cations on Activation of PP1-After metal depletion with EDTA/P i (see "Experimental
Procedures") recombinant PP1 required divalent metal cations for activity (Table I). Of the cations examined individually, only Co 2ϩ and Mn 2ϩ were efficient in activating the enzyme. Treatment of PP1 with 1 mM Ca 2ϩ , Cu 2ϩ , Fe 2ϩ , Ni 2ϩ , or 5 mM Mg 2ϩ did not activate the enzyme, while 1 mM Zn 2ϩ had only a slightly stimulating effect (Table I). Similar results were obtained when metal-dependent PP1 purified from bovine heart myofibrils (Chu et al., 1994) was tested under the same conditions (data not shown). As shown in Table I, PP1 was activated to a greater extent by 1 mM Mn 2ϩ than 1 mM Co 2ϩ when the preincubation was carried out in the presence of 150 mM KCl. It is worth noting that Co 2ϩ and Mn 2ϩ were equally effective in the activation of PP1 when the preincubation was carried out in the presence of 500 mM KCl, indicating that Co 2ϩ -activation was more dependent on ionic strength than Mn 2ϩ (data not shown). Very surprisingly, combined treatment of PP1 with Zn 2ϩ and Fe 2ϩ significantly activated the enzyme (Table I). Preincubation with other metal ion combinations had little or no effect on phosphatase activity. As shown in Fig. 1, maximal activation of the enzyme by Zn 2ϩ in the presence of 0.1 or 1 mM Fe 2ϩ , or by Fe 2ϩ in the presence of 0.1 or 1 mM Zn 2ϩ , was achieved at a metal concentration of 1 mM. The A1 ⁄2 was approximately 0.1 mM for either cation. Activation by the combination of Zn 2ϩ /Fe 2ϩ did not appear to be sensitive to ionic strength (data not shown).
Effect of EDTA and P i on Co 2ϩ -or Mn 2ϩ -activated PP1-The reversibility of the activation of recombinant PP1 was investigated by studying the effects of EDTA and P i on Mn 2ϩ -or Co 2ϩ -activated PP1. More than 80% of the Mn 2ϩ -stimulated activity was reversed by chelation of the Mn 2ϩ with 5 mM EDTA ( Fig. 2A). It was noted that much of the Mn 2ϩ -stimulated activity was lost simply by dilution of the Mn 2ϩ (data not shown). Mn 2ϩ -activated PP1 was totally inactivated within 5 min by 100 mM P i . Thus, the effect of Mn 2ϩ on PP1 was properly characterized as a stimulation of the enzyme. Activation by Fe 2ϩ /Zn 2ϩ was also reversed by chelation of the metal ions (data not shown). Co 2ϩ -activated PP1 was resistant to EDTA or P i treatment (Fig. 2B). Incubation at 30°C for up to 2 h with 5 mM EDTA did not reverse Co 2ϩ activation. Inorganic phosphate (100 mM) or a combination of 100 mM P i and 5 mM EDTA only partially reversed the Co 2ϩ -activated PP1 even after 2 h incubation. These results indicate the effect of Co 2ϩ on PP1 is an activation, as we previously observed with the latent cardiac PP1 (Chu et al., 1994).
Evidence for a Stable Co 2ϩ ⅐PP1 Complex-Although Co 2ϩactivated PP1 is stable, it is not clear whether the Co 2ϩ activation involved formation of a stable Co 2ϩ ⅐PP1 complex or a transient binding of Co 2ϩ which results in the induction of a stable active conformation. Therefore, we examined the Co 2ϩbinding properties of PP1 by direct binding studies using 57 Co 2ϩ . For analysis of Co 2ϩ binding, the sample of recombinant PP1␣ was depleted of endogenous Mn 2ϩ by EDTA/P i treatment (see "Experimental Procedures"). The enzyme was incubated with 57 CoCl 2 and passed through a Superose-12 gel filtration column to separate free and bound Co 2ϩ . After chromatography, PP1 was fully activated, and 57 Co 2ϩ was associated with the enzyme (Fig. 3). The fact that the bound 57 Co 2ϩ was not removed by chromatography in the presence of EDTA indicates the formation of a stable metalloenzyme complex. This is supported by a calculated stoichiometry of 0.93 mol of 57 Co 2ϩ bound/mol of PP1 for the fractions in which the enzyme activity and radioactivity co-eluted. These results imply that Co 2ϩ is incorporated into PP1 during the enzyme activation The diluted PP1␣ was incubated with indicated concentrations of divalent cations in the presence of 50 mM MOPS, pH 7.0, 150 mM KCl, 0.5 mg of BSA/ml, 0.5 mM DTT at 30°C for 15 min. Five l of treated enzyme were added to 45 l of [ 32 P]phosphorylase a (2.22 M in 50 mM imidazole, pH 7.4, 1.2 mg of theophylline/ml, 0.5 mg of BSA/ml, 1 mM DTT). After 10 min, the reaction was stopped by adding 60 l of ice-cold 20% trichloroacetic acid, and released 32 P i was determined as previously described (Chu et al., 1994). The Mn 2ϩ -activated phosphatase activity was taken as 100%. Note: Fresh FeCl 2 solution was used.
Effect of Fe 2ϩ on Co 2ϩ -activated PP1-Fe 2ϩ itself had very little effect on activation of PP1. However, as shown in the insert of Fig. 4, Fe 2ϩ could inactivate the Co 2ϩ -activated PP1. The possibility that displacement of 57 Co 2ϩ by Fe 2ϩ is responsible for the enzyme inactivation was examined by determining the 57 Co 2ϩ content of the enzyme before and after treatment with Fe 2ϩ . From Fig. 4, it can be seen that 57 Co 2ϩ remained bound, even though phosphatase activity was almost completely lost. The enzyme activity was fully recovered when the Fe 2ϩ was chelated by EDTA indicating that the Fe 2ϩ effect was reversible (data not shown). The fact that 57 Co 2ϩ was not displaced by Fe 2ϩ after the enzyme inactivation suggested that Co 2ϩ and Fe 2ϩ do not share a common binding site.

DISCUSSION
As noted in the Introduction, the issue of whether the catalytic subunit of protein phosphatase 1 is a metalloenzyme is an old issue that has not been satisfactorily resolved to date, although the effects of Mn 2ϩ on the enzyme activity have been well documented (see Bollen and Stalmans (1992) for review). Attempts to show the binding of Mn 2ϩ to PP1 have been negative (Brautigan et al., 1980;Villa-Moruzzi et al., 1984). Our data provide an explanation for this in that Mn 2ϩ binding is reversible, whereas Co 2ϩ binding results in the formation of a stable metalloprotein complex. Thus, the use of Co 2ϩ rather than Mn 2ϩ has proven in this study to be more revealing. The data presented here provides the first direct evidence that PP1 is a metalloenzyme by demonstration of a stable 1:1 complex of enzyme and Co 2ϩ . Previously, direct metal analysis of liver PP1 preparations showed the presence of only substoichiometric amounts of metal (Yan and Graves, 1982). While the formation of a cobalt metalloenzyme form is probably not physiological (see below), the results establish that PP1 has the ability to bind a metal ion in a stable and stoichiometric manner. Our studies show that cobalt ion will be a useful tool for the study of the role of metal ions on PP1 activity.
Although Co 2ϩ can bind to and activate PP1 in vitro, it seems unlikely that PP1 is a Co 2ϩ -bound protein in vivo, because the concentration of Co 2ϩ in tissues is less than 1 M (Iyengar and Woittiez, 1988). The observations that PP1 purified from rabbit skeletal muscle is susceptible to P i inhibition 2 and that Co 2ϩactivated PP1 is relatively resistant to P i inhibition (see "Re-FIG. 2. Effect of EDTA and P i on Mn 2؉ -or Co 2؉ -activated PP1. The PP1␣ was diluted with 50 mM MOPS, pH 7.0, 0.5 M KCl, 0.2 mM EDTA, 1 mM DTT, 0.5 mg BSA/ml. The enzyme was preincubated with 1 mM MnCl 2 (A) or 1 mM CoCl 2 (B) at 30°C for 15 min. Further incubations were carried out in the absence (control, E) and presence of 5 mM EDTA (q), 100 mM KPO 4 ( ), 5 mM EDTA plus 100 mM KPO 4 (Ⅺ). At indicated intervals, an aliquot of the reaction was diluted with 50 mM imidazole-HCl, pH 7.4, 1.2 mg of theophylline/ml, 1 mM DTT, 0.5 mg of BSA/ml, and assayed for phosphatase activity. FIG. 4. Deactivation of the Co 2؉ -activated PP1 by Fe 2 ؉. The PP1␣ (5 g in 45 l) was incubated with 1 mM 57 Co 2ϩ in 50 mM MOPS, pH 7.0, 0.5 M KCl, 0.2 mM EDTA, 25% glycerol for 15 min at 30°C. Co 2ϩ was chelated with 2 mM EDTA and the enzyme was further incubated with 2 mM FeCl 2 for 15 min at 30°C. The treated enzyme was separated from unbound 57 Co 2ϩ by passing through a Superose-12 gel filtration column (1.0 ϫ 30 cm), which was equilibrated with the buffer containing 50 mM MOPS, pH 7.0, 150 mM KCl, 10% glycerol, and 0.2 mM FeCl 2 . The control was prepared as described in the legend of Fig. 3. Fractions of 0.4 ml were collected, assayed for phosphatase activity (E, q) and radioactivity ( , ). The Fe 2ϩ -treated enzyme is plotted with filled symbols, the control with open symbols. Insert, PP1 was incubated with 1 mM Co 2ϩ for 15 min at 30°C, then was treated with 1 mM Fe 2ϩ for 15 min at 30°C. The sample was directly assayed as described under "Experimental Procedures." sults") also do not support a role for Co 2ϩ . The fact that the amount of Mn 2ϩ in skeletal muscle ranges from 1 to 2 M (Versieck, 1985) and that 54 Mn 2ϩ did not show significant binding to PP1 catalytic subunit (Brautigan et al., 1980) do not favor the idea that PP1 is activated by Mn 2ϩ in vivo. Our findings that a combination of Fe 2ϩ /Zn 2ϩ , but not the individual metals, can activate PP1 raises the possibility that PP1 is an iron/zinc metalloenzyme. The activation by Fe 2ϩ /Zn 2ϩ was found to have an A1 ⁄2 of approximately 0.1 mM for both cations. Fe 2ϩ and Zn 2ϩ are present in skeletal muscle in millimolar and near millimolar concentrations respectively (Versieck, 1985;Iyengar and Woittiez, 1988). Interestingly, metal analysis of a purified preparation of the rabbit liver PP1 revealed that although Zn 2ϩ and Fe 2ϩ were present in substoichiometric amounts, there was considerably more Zn 2ϩ and Fe 2ϩ detected than some other metal cations tested (Yan and Graves, 1982). It is likely that the substoichiometric levels of Zn 2ϩ and Fe 2ϩ in the purified enzyme are due to the loss of metal cation during enzyme purification. This speculation is consistent with the fact that the Fe 2ϩ /Zn 2ϩ co-activated PP1 loses phosphatase activity when the cations are removed by chelation.
Fe 2ϩ itself cannot effectively activate PP1 but it can reversibly inactivate the Co 2ϩ -activated enzyme. Even though the Co 2ϩ -activated enzyme was inactivated in the presence of Fe 2ϩ , the bound Co 2ϩ was not removed. These results indicate that the deactivation does not result from the displacement of Co 2ϩ . It may result from Fe 2ϩ binding at another site and/or an Fe 2ϩ -induced conformational change in the enzyme. These results are consistent with two metal binding sites on PP1. Another family of phosphatases, the mammalian purple acid phosphatases, are metalloproteins Averill, 1990a, 1990b). Comparison of the primary structures of purple acid phosphatases and Ser/Thr protein phosphatases have lead Vincent and Averill to speculate that PP1 and phosphatase 2A are iron/zinc metalloenzymes with active sites isostructural with those of the purple acid phosphatase. Our data provide the first experimental evidence to support the postulate that PP1 is an iron/zinc metalloenzyme. It is interesting to note that we recently established that the catalytic subunit of phosphatase 2A can also exist in a divalent cation-dependent form (Cai et al., 1995).
While the suggestion of Vincent and Averill (1990b) that the Ser/Thr protein phosphatase may contain two metal sites was based on weak sequence homologies of the purple acid protein phosphatase with the Ser/Thr protein phosphatases represented by PP1, protein phosphatase 2A, and protein phosphatase 2B (calcineurin), the recent elucidation of the crystal structure of protein phosphatase 2B has now confirmed the existence of iron and zinc in the active site (Griffith et al., 1995). While this manuscript was under review, the crystal structure of recombinant PP1 (␣-isoform) was reported (Goldberg et al., 1995). This structure shows the presence of two metal ions in the catalytic site. Since the enzyme was prepared in the presence of Mn 2ϩ , the ions were presumed to be Mn 2ϩ . These studies confirm our findings that PP1 is a metalloenzyme and strengthen the view that PP1 may have bound zinc and iron ions at the active site. Given the structural similarities of protein phosphatase 2B with PP1 and protein phosphatase 2A, it seems likely that the latter will possess similar metal ion sites. On the other hand, Zhuo et al. (1993Zhuo et al. ( , 1994 reported that Mn 2ϩ or Ni 2ϩ activation of a bacteriophage ␥ Ser/Thr protein phosphatase (␥ PPase) had an apparent single K m for each of the divalent metals. The latter results are consistent with one metal ion binding site involved in the activation of ␥ PPase. Further studies will be necessary to identify the physiologically important metal ions responsible for PP1 activation.