Interactions of the subunits of smooth muscle myosin phosphatase.

Myosin phosphatase from smooth muscle consists of a catalytic subunit (PP1c) and two non-catalytic subunits, M130 and M20. Interactions among PP1c, M20, and various mutants of M130 were investigated. Using the yeast two-hybrid system, PP1c was shown to bind to the NH2-terminal sequence of M130, 1-511. Other interactions were detected, i.e. PP1c to PP1c, M20 to the COOH-terminal fragment of M130, and dimerization of the COOH-terminal fragment of M130. Mutants of M130 were constructed to localize the PP1c and light chain binding regions. Results from the two-hybrid system indicated two binding sites for PP1c on M130: one site in the NH2-terminal 38 residues and a weaker site(s) in the ankyrin repeats region. Inhibition of PP1c activity with phosphorylase a by the M130 mutants also was consistent with the assignment of these two sites. Overlay assays showed binding of phosphorylated light chain to the ankyrin repeats, probably in the COOH-terminal repeats. Activation of PP1c with phosphorylated light chain required binding sites for PP1c and substrate, plus an additional sequence COOH-terminal to the ankyrin repeats. Thus, activation of phosphatase and binding of PP1c and substrate are properties of the NH2-terminal one-third of M130.

Phosphorylation of the two regulatory light chains of myosin is an important mechanism in controlling the contractile activity of smooth muscle (1). The extent of phosphorylation depends on the balance of activities of two enzymes: the Ca 2ϩ / calmodulin-dependent myosin light chain kinase and a myosin phosphatase (MP). 1 Under some conditions, for example stimulation with certain agonists, the extent of phosphorylation at suboptimal Ca 2ϩ concentrations can increase, and this has been traced to an inhibition of MP activity (2). The pathway from receptor occupancy to phosphatase inhibition is not established but is thought to involve trimeric (2) and monomeric (2-6) G proteins. In addition it has been suggested that MP inhibition is caused by dissociation of the subunits by arachidonic acid (2, 7) or by phosphorylation of the large MP subunit (8,9).
A classification of protein phosphatases which is used widely was based on substrate preferences and inhibition by inhibitors 1 and 2 (10). Two classes were identified, 1 (PP1) and 2 (PP2) with the latter divided into three subgroups, A, B, and C. Several years ago it was suggested that the smooth muscle MP was a type 1 enzyme (11). Subsequently three laboratories have reported that MP from gizzard (12,13) and pig bladder (14) consists of three subunits: a catalytic subunit of about 38 kDa; a large subunit of 110 -130 kDa (based on motilities on SDS-PAGE), termed M130; and a subunit of about 20 kDa, M20. The cDNAs for each subunit have been sequenced. The PP1c from gizzard is the ␦ isoform (13), also referred to as the ␤ isoform (12). In gizzard, two isoforms of the large subunit exist, M130 and M133 (13); these are similar to, but not identical with, the large subunit from rat aorta, termed M110 (15). Another isoform of this subunit was detected from a rat kidney cDNA library and represented the NH 2 -terminal 72.5-kDa fragment (16). The derived sequence of M20 from gizzard (15) indicates several leucine zipper sequences at the COOH-terminal end. This structure is similar to the COOH terminus of rat aorta M110 (15) but different from the two gizzard isoforms (13) which lack leucine zippers.
A possible regulation mechanism for phosphatases, and kinases, involves targeting (12) or anchoring subunits (17). These bind to the catalytic subunit and to the substrate or other specific target. In addition to localizing the phosphatases they may also modify the activity of the catalytic subunit. Thus the two non-catalytic subunits of MP could act as targeting subunits (12). In support of this idea it is known that the holoenzyme has enhanced activity toward myosin, compared with PP1c (12,14), and also that the holoenzyme binds to myosin (13). The sites for interactions among subunits are not defined, but it has been shown that the NH 2 -terminal portion of M130 probably contains the PP1c site(s) and myosin binding site(s) (16,18). An interesting feature of the M130 molecule is that it contains several ankyrin repeat sequences (13,15) close to the NH 2 terminus which could provide a platform for multiple interactions.
To investigate interactions of the subunits of MP the yeast two-hybrid system was employed. This powerful technique to study protein-protein interactions was developed by Fields and Song (19) and subsequently has been used for many systems, including screening of a gizzard cDNA library for PP1c-binding proteins (20,21). In this present application various subunits, or fragments thereof, were inserted as bait or prey with the objectives of obtaining an outline of the subunit interactions of the holoenzyme. Where applicable the interactions were also monitored by enzyme assays or by overlay techniques.

EXPERIMENTAL PROCEDURES
Materials-Oligonucleotides were synthesized at the Macromolecular Structure Facility at the University of Arizona and by National Biosciences Inc. Plymouth, MN. The antibody to the hexahistidine tag ( MRGS His) was from Qiagen. Enzymes and media for bacterial and yeast cultures and radionucleotides were as listed previously (20). * This work was supported by National Institutes of Health Grants HL23615 and HL20984 (to D. J. H.) and HL07249 (to B. C. P.). 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.
Bacteria and Yeast Strains-Escherichia coli DH5␣ was used as the transformation recipient for plasmid constructions in the two-hybrid system. Saccharomyces cerevisiae strain Y190 (20) and Y187 (Clontech) were used for the two-hybrid assay, as a recipient of bait and prey plasmid, respectively. E. coli JM109 and M15[pREP4] were used as transformation recipient and as expression host for the pQE expression system (Qiagen), respectively.
Construction of Truncation Mutants-Various truncation mutants of M130 were constructed for use in the two-hybrid system and for expression as hexahistidine-tagged proteins (18). All cDNAs for the truncation mutants were obtained by PCR amplification with pACT2-M130 1-633 as the template DNA. The sense and antisense primers were designed to contain BamHI and SalI sites for ligation to the vector plasmid, respectively. Pfu DNA polymerase (Stratagene) was used for the PCR. The BamHI and SalI-digested PCR products were ligated to the BamHI-and XhoI-digested pACT2 for the prey constructs in the two-hybrid system and to BamHI-and SalI-digested pQE32 for expression as hexahistidine-tagged proteins. Truncation mutants obtained were The Two-hybrid Assay Using Yeast Mating-The principle of the two-hybrid assay for protein-protein interaction was described elsewhere (22). In the standard two-hybrid assay, haploid yeast cells are cotransformed with two plasmids, bait and prey. This standard assay was used to examine interactions between the phosphatase subunits with pAS1 and pGAD424 as bait and prey vectors, respectively. However, the sensitivity of this assay was low. To increase sensitivity the two-hybrid assay was carried out using a different prey vector, pACT2, and also by mating two types of yeast strains expressing either the bait or prey hybrid proteins. The advantages of this approach were first, that pACT2 gives higher levels of expression than pGAD424 (see Clontech literature); and second, that the mated diploid cells (Y190 and Y187) contain two copies of the lacZ reporter gene, and this should lead to a higher level of ␤-galactosidase expression compared with a single copy. The bait and prey plasmids were introduced into S. cerevisiae strain Y190 (MATa) and Y187 (MAT␣), respectively, using the lithium acetate method (23). The two strains were mixed and incubated overnight in YPD medium. The mated cells were selected by plating on SC-Try-Leu agar, and this plate was used for lacZ reporter assay. The expression of all bait and prey protein was confirmed by Western blot analysis of the yeast homogenate with anti-hemagglutinin antibody (20), anti-PP1␦ (20), or anti-M130 (8).
lacZ Reporter Assay-The filter lift assay for ␤-galactosidase was as described (20). To quantitate ␤-galactosidase activity all colonies were collected after the filter lift assay and cultured in SC-Try-Leu broth. Activities were determined using chlorophenol red-␤-D-galactopyranoside as substrate and expressed as units defined by Bartel and Fields (22).
HIS3 Reporter Assay-The GAL1-HIS3 reporter gene utilized in Y190 had residual HIS3 expression sufficient to allow growth without exogenous histidine, even in the absence of GAL4 (24). The stringency of the His selection can be modulated by varying the concentrations of 3-aminotriazole. Therefore, the level of HIS3 reporter gene expression was semi-quantitatively evaluated by examining growth of yeast colonies on SC-Try-Leu-His containing various concentrations of 3-ami-notriazole (0, 5, 10, 20, 30, and 50 mM). The mated yeast cells were streaked on SC-Try-Leu-His ϩ 3-aminotriazole and allowed to grow for 5-10 days. The extent of growth for each concentration of 3-aminotriazole was then evaluated.

Expression and Purification of Hexahistidine-tagged Truncation Mutants-E. coli cells M15[pREP4]
, containing the pQE expression plasmid, were cultured in LB broth supplemented with 100 g/mg ampicillin and 25 g/ml kanamycin at 37°C overnight. A 100-fold dilution of the overnight culture was grown at 37°C until the absorbance at 600 nm reached between 0.7 and 0.9. Expression then was induced by addition of isopropyl-␤-D-thiogalactopyanoside to 0.4 mM. Cells were allowed to grow for 3-4 h and collected by centrifugation at 15,000 ϫ g for 15 min at 4°C. After washing with 40 mM Tris-HCl (pH 8.0), 1 mM EDTA and centrifugation, the cell pellets were stored at Ϫ80°C. After thawing at 4°C, cells were homogenized in 6 M guanidine HCl, 50 mM sodium phosphate, 10 mM Tris-HCl, 100 mM NaCl (pH 8.0; buffer A). The homogenate was clarified by centrifugation at 70,000 ϫ g for 20 min at 4°C, and the supernatant was mixed for 20 min with metal affinity resin (Talon, Clontech) equilibrated in buffer A. The resin was loaded into a column and washed extensively with 6 M guanidine HCl, 50 mM sodium phosphate, 100 mM NaCl (pH 7.0). The bound protein was eluted by 100 mM imidazole, 6 M guanidine HCl, 20 mM Tris-HCl (pH 8.0), 100 mM NaCl. The eluate was dialyzed against 30 mM Tris-HCl (pH 7.5), 30 mM NaCl, 0.5 mM dithiothreitol. The dialysate was clarified by centrifugation at 12,000 ϫ g for 10 min at 4°C, and diisopropyl fluorophosphate and leupeptin were added to 0.5 mM and 10 g/ml, respectively. The yield of soluble protein varied with the mutant. Some examples, given in mg of protein/liter of culture, were 0.5 for M130 1-38 and 12 for M130  . In general, the purity was estimated by SDS-PAGE to be 70 -95%.
Phosphatase Assays-32 P-Labeled myosin light chain (11) or phosphorylase a (25) was used as a substrate at 5 M. The assay mixtures contained approximately 1 nM PP1c prepared from turkey gizzard and various concentrations of mutants in 30 mM Tris-HCl (pH 7.5), 30 mM NaCl, 0.5 mM dithiothreitol, and 0.4 mg/ml bovine serum albumin. Caffeine, 3 mM, was added for assays with phosphorylase a. After a 3-min incubation at 30°C reactions were started by the addition of substrate and terminated by the addition of trichloroacetic acid and bovine serum albumin to final concentrations of 7% and 4 mg/ml, respectively. Aliquots were cooled to 0°C and the precipitated protein sedimented at 15,000 ϫ g for 2 min. Radioactivity (i.e. released 32 P) of the supernatant was determined by Cerenkov counting.
Other Procedures-The type 1 catalytic subunit was isolated from frozen turkey gizzard (31). This is predominantly a 38-kDa species. Nucleotide sequences were determined with a Sequenase version 2.0 DNA sequencing kit (U. S. Biochemical Corp.) and ␣-35 S-dATP (Du-Pont NEN). Other methods as given previously (20).

RESULTS
Two-hybrid Screen Using All Subunits-To assess the binding among the three subunits of the holoenzyme, various bait and prey plasmids were constructed and inserted into the yeast two-hybrid system to detect interactions. The colorimetric procedure monitoring ␤-galactosidase was used. The bait plasmids utilized cDNAs from PP1␦, the M20 subunit minus the leucine zipper sequences, an NH 2 -terminal part of M130 from residues 1 through 511 (M130 1-511 ), and the COOH-terminal part of M130 (M130 512-963 ). The prey plasmids included PP1␦ and M130 1-511 plus an NH 2 -terminal fragment of PP1␦ (PP1␦ 1-108 ), a larger NH 2 -terminal fragment of M130 (M130 1-633 ), and a COOH-terminal portion of M130 (M130 514 -963 ). Results from the two-hybrid assays are shown in Table I. With PP1␦ as bait, the interacting bait-prey pairs were PP1␦-PP1␦, PP1␦-M130 1-511 , and PP1␦-M130  . The interaction between the catalytic subunit and the NH 2 -terminal fragment of M130 was bidirectional and was shown when M130 1-511 was used as bait and PP1␦ as prey (Table I). A PP1␦ mutant containing the NH 2 -terminal third of PP1␦ (PP1␦ 1-108 ) did not show this interaction (Table I). Other interactions also were detected and were: binding of M20 (as bait) and the COOH-terminal fragment of M130 (M130 512-963 ); interaction among the COOHterminal fragments of M130; and a weaker interaction between PP1␦ and the COOH-terminal fragment of M130.
The key point from the results shown in Table I was that PP1c bound to the NH 2 -terminal fragment of M130. However, the smallest NH 2 -terminal fragment used was still relatively large, i.e. 511 residues, and it was important to define more precisely the PP1c binding site(s). To achieve this a variety of mutants was constructed as shown in Fig. 1. These include progressive COOH-terminal deletions from Asn-511 and NH 2terminal deletions with Asn-511 fixed as the COOH terminus. These mutants were used in the two-hybrid system as prey proteins with PP1␦ as bait and were also expressed as hexahistidine-tagged fusion proteins.
The results of the two-hybrid assays using both the filter lift assays and ␤-galactosidase measurements are shown in Table  II. All of the mutants that contained sequences initiated at Met-1 bound PP1␦. The shortest NH 2 -terminal segment was Met-1 to Phe-38, and this also bound to PP1␦. This sequence precedes the ankyrin repeats that for the gizzard M130/133 isoforms begin at Asp-39 (13). Although the interaction between M130 1-38 and PP1␦ appeared weaker than for longer segments of M130 (Table II), it represented a critical interaction since those mutants lacking the NH 2 -terminal 39 residues did not bind PP1␦. This is shown in Table II for the mutant M130 40 -511 . Other mutants lacking longer NH 2 -terminal sequences also did not bind PP1␦ (Table II).
The 3-aminotriazole resistance (an assay of the HIS3 reporter gene) of these bait-prey pairs also was estimated. In general, the same pattern was observed as shown in Table II. The strongest interactions, i.e. PP1␦-M130 1-633 and PP1␦-M130 1-69 , showed vigorous growth at 30 mM 3-aminotriazole and positive but reduced growth at 50 mM. The weaker interactions had a reduced tolerance of 3-aminotriazole (data not shown).
The Effect of M130 Mutants on Phosphatase Activity-It was shown previously that the MP holoenzyme is more active with phosphorylated myosin or P-LC20 than the isolated catalytic subunit (12,14,18) and that this activation is carried by an NH 2 -terminal fragment, i.e. M133 1-674 (18). It is assumed that activation of phosphatase activity requires the presence of at least two sites on M130, a binding site for PP1␦ and one for the substrate, P-LC20. Thus the M130 mutants, expressed as hexahistidine fusion proteins, were assayed for their effect on PP1c activity using 32 P-LC20 as substrate (Fig. 2). Three of the mutants containing the NH 2 -terminal region of M130 activated PP1c activity. These were M130 1-633 , M130 1-511 , and M130 1-374. However, the smallest mutant was less effective, and activation required higher concentrations. Further truncation of the NH 2 -terminal segment caused loss of activation, as shown for M130  . The M130 fragment in which the first 39 residues were deleted, M130 40 -511 , also did not activate phosphatase activity (Fig. 2).
One possibility is that the interaction of PP1c and M130 induced a conformational change in PP1c which resulted in activation of phosphatase activity. It would be predicted that this effect is independent of substrate. To test if the activation of PP1c is dependent only on interaction with M130 and is not specific with respect to substrate, the effect of various mutants on phosphatase activity was assayed using phosphorylase a as substrate. The effect of three representative M130 mutants is shown in Fig. 3. Each mutant inhibited phosphatase activity, but the potency of inhibition varied markedly. The most effective inhibitor was M130 1-511 , whereas M130 40 -511 was considerably less potent. To categorize this effect each of the mutants was assayed for its effect on the PP1c-phosphorylase a system, and IC 50 values (i.e. the concentration of mutant required for 50% inhibition) were calculated. These are given in Table III. The relative value for each mutant compared with M130 1-633 also is given. These values can be divided into three groups. For the mutants including the sequence from 1-633 down to 1-296 (top four mutants of Table III) Table III).
Light Chain Overlays-To obtain an independent assessment of P-LC20 binding an overlay procedure was used. Each of the M130 mutants was screened for binding to thiophosphorylated, biotinylated LC20. In Fig. 4 are shown the SDS-PAGE patterns of the mutants and the light chain overlays. For the larger mutants many of the lower molecular mass bands were proteolysis products since they retained the hexahistidine tag (as shown by the MRGS His antibody). For the smaller mutants (M130 1-171 and below) proteolysis apparently was not a problem. Binding of light chain was detected for the NH 2terminal fragments containing a complete ankyrin repeat region M130 1-296 and larger and also for mutants M130 40 -511 and M130 168 -511 . The latter two mutants did not activate PP1c. The M130 168 -511 mutant contains the COOH-terminal half of the ankyrin repeats, and this suggests that this part of the molecule is involved in light chain interaction. I ␤-Galactosidase activity of mated yeast cells expressing subunits and mutants of myosin phosphatase as bait and prey ␤-Galactosidase activity of the yeast cell lysate was determined using chlorophenol red-␤-D-galactopyranoside as substrate. Unit activity is defined as in Bartel and Fields (22). The underlined values showed significant difference (p Ͻ 0.01) from that of the empty prey vector with the same bait protein, as assessed by Student's t test for unpaired data. Data are mean Ϯ S.E. (n Ն 3).

DISCUSSION
From the initial two-hybrid screen a rough plan of the holoenzyme can be obtained. The NH 2 -terminal third of M130 is involved in the interaction with PP1c, and M20 binds to the COOH-terminal part of M130. The function of M20 is not known. The results of the two-hybrid assays indicate that it does not bind to PP1␦. Earlier results showed that the complex of PP1c and an NH 2 -terminal fragment of M130, i.e. the 58-kDa component, was similar to the trimeric MP holoenzyme in terms of myosin dephosphorylation and binding to myosin (33). This complex did not include M20. Thus a role for M20 in activation of PP1c or in binding to myosin is unlikely. It has not been determined if M20 is required for regulation of MP activity. In addition, it is possible that M20 does not influence phosphatase activity but may serve an auxiliary function such as targeting MP to other proteins and indirectly modifying activity or determining cellular localization.
Various interactions with PP1c (PP1␦) were indicated. The initial two-hybrid screen showed that PP1␦ could self-associ-ate. Dimer formation of PP1c has been demonstrated (34), and it was also proposed that myosin light chain phosphatases isolated from gizzard consisted of a tetramer of catalytic subunits (35,36). Thus, the association of PP1c subunits is consistent with earlier results, but the physiological significance of dimer or tetramer formation is not known. It is unlikely that a PP1c dimer or tetramer would have a high affinity for myosin since PP1c in solution does not bind to dephosphorylated or phosphorylated myosin (18).
Interaction of PP1c and M130 forms an important component of the function of the MP holoenzyme. The two-hybrid assays for various M130 mutants as prey and PP1␦ as bait indicated that PP1␦ binds to the NH 2 -terminal part of M130. The surprising observation was that PP1␦ bound to the NH 2terminal segment of 38 residues. This precedes the ankyrin repeats that begin at Asp-39. The binding of PP1␦ to this segment was weaker than for the longer NH 2 -terminal sequences but was an important component of overall binding since the mutants lacking this sequence, e.g. M130 40 -511 , did  Table I). The criteria used for the filter lift assay were: ϩϩϩϩ, turned blue Յ1 h; ϩϩϩ, 1-3 h; ϩϩ, 3-8 h; ϩ, Ͼ8 h; Ϫ, no blue color development after overnight incubation. The significant difference (*, p Ͻ 0.01; **, p Ͻ 0.05) in activity from that of the empty vector was determined by Students's t test. Activity data are mean Ϯ S.E.  not give a positive signal. Recently, Endo et al. (37) have shown that an NH 2 -terminal peptide of inhibitor-1, KIQF, was required for full inhibition by phosphorylated inhibitor-1. One possibility, suggested by these authors, was that the tetrapeptide represented part of a PP1c binding site distinct from the catalytic site occupied by the phosphorylated Thr-35 of inhibitor-1. A similar sequence, KVKF, is found only in one position in M130/133, namely at residues 35-38 (13). This NH 2 -terminal sequence also is present in rat aorta M110 (15) and rat kidney M110 (16). Thus it is possible that this sequence forms at least part of the PP1c binding site present in the sequence 1-38. While this manuscript was in preparation, Johnson et al. (38) also noted the importance of the NH 2 -terminal sequence of M130 in binding PP1c. However, they reported that the sequence 1-38 activated PP1c (38) and facilitated relaxation in skinned fibers (39), although at relatively high concentrations. This was not observed in our studies.
The inhibition data from the PP1c-phosphorylase a assays can also be used to assess PP1c binding to the M130 mutants. If it is assumed that there is no specific interaction between M130 and phosphorylase a, then inhibition of phosphatase activity by M130 and its mutants would reflect competitive binding of PP1c to M130 and phosphorylase a. In addition, it is required that the binding sites(s) on PP1c for M130 and phosphorylase a is similar, or the sites overlap. This has recently been shown (38). The most effective inhibitors, therefore, would possess a higher affinity for PP1c. These are represented (see Table III) by those mutants possessing the NH 2 -terminal 38 residues plus a longer NH 2 -terminal segment, possibly the ankyrin repeats. Truncation of these mutants at the COOHterminal end reduces the inhibitory potency. Removal of the four COOH-terminal ankyrin repeats (in M130 1-171 ) produces a less effective inhibitor. The predicted PP1c binding site for this second group of mutants is the NH 2 -terminal sequence 1-38. The loss of this NH 2 -terminal sequence generates the third group of inhibitory mutants. Here the inhibitory potency is low, indicating a reduced affinity of binding, and it is difficult to assign the location of the additional PP1c binding site. The very high IC 50 for M130 40 -511 compared with the other two mutants in this group cannot be explained but was a reproducible observation.
The results from the light chain overlays (using the biotinylated thiophosphorylated LC20 as a probe) indicate that the ankyrin repeats are required for light chain binding. The NH 2terminal segment of 39 residues is not involved. Further, it is suggested that the COOH-terminal half of the ankyrin repeats is important, since binding was detected for M130 1-296 but not M130  . The fifth ankyrin repeat shows a considerably lower homology than the other repeat sequences (13) and in fact was  not considered as an ankyrin repeat in the rat M110 (15). Thus it is suggested that repeats 6, 7, and 8 may play a more crucial role.
The smallest NH 2 -terminal fragment that could activate PP1c (using 32 P-LC20 as substrate) was M130  . This mutant contains in addition to the ankyrin repeats another 78 residues, and in this sequence the notable feature is an acidic cluster, residues 326 -372 (13). It is not known if this sequence contains an additional binding site for PP1␦ or P-LC20 or if it is necessary for correct folding or orientation of the ankyrin repeats.
In summary, a tentative plan of the M130 molecule can be assembled from the above data, with the emphasis on the NH 2 -terminal portion. For PP1c at least two sites are indicated: a relatively strong site in the NH 2 -terminal 38 residue sequence and a second weaker site in the ankyrin repeats, possibly in the COOH-terminal half of the repeats. If such is the case then the NH 2 -terminal portion of M130 may wrap around PP1c. The binding of P-LC20 is indicated in the ankyrin repeats, and again the COOH-terminal repeats are suggested to be involved. Activation of PP1c by M130 is assumed to require binding to both PP1c and substrate, P-LC20. Thus, theoretically the sequence 1-296 should be sufficient for activation. However, the situation is more complex, and additional COOH-terminal sequence (297-374) was required for activation. Whether this sequence contains additional sites for interaction or is needed to stabilize the NH 2 -terminal segment is not known. Another possibility is suggested by earlier results (9), namely, that if inhibition of PP1c by phosphorylated M130 results from binding of Thr-654 to the active site of PP1c then the M130 molecule must fold to accommodate this interaction.