Phosphorylation of a novel myosin binding subunit of protein phosphatase 1 reveals a conserved mechanism in the regulation of actin cytoskeleton.

The myotonic dystrophy kinase-related kinases RhoA binding kinase and myotonic dystrophy kinase-related Cdc42 binding kinase (MRCK) are effectors of RhoA and Cdc42, respectively, for actin reorganization. Using substrate screening in various tissues, we uncovered two major substrates, p130 and p85, for MRCKalpha-kinase. p130 is identified as myosin binding subunit p130, whereas p85 is a novel related protein. p85 contains N-terminal ankyrin repeats, an alpha-helical C terminus with leucine repeats, and a centrally located conserved motif with the MRCKalpha-kinase phosphorylation site. Like MBS130, p85 is specifically associated with protein phosphatase 1delta (PP1delta), and this requires the N terminus, including the ankyrin repeats. This association is required for the regulation of both the catalytic activities and the assembly of actin cytoskeleton. The N terminus, in association with PP1delta, is essential for actin depolymerization, whereas the C terminus antagonizes this action. The C-terminal effects consist of two independent events that involved both the conserved phosphorylation inhibitory motif and the alpha-helical leucine repeats. The former was able to interact with PP1delta only in the phosphorylated state and result in inactivation of PP1delta activity. This provides further evidence that phosphorylation of a myosin binding subunit protein by specific kinases confers conformational changes in a highly conserved region that plays an essential role in the regulation of its catalytic subunit activities.

The Rho subfamily of GTPases are biological regulators of actin cytoskeleton. In adherent cells, RhoA induces stress fiber formation, Rac-1 generates lamellipodia, and Cdc42 produces filopodia and actin microspikes (1). A variety of effectors of these cytoskeletal switches has been isolated and characterized (see Refs. 2 and 3 for reviews), some of which are directly involved in regulation of actin dynamics. We and others have reported serine/threonine kinases related to the myotonic dystrophy kinase that play effector roles for the perspective GTPase in cytoskeletal reorganization (4 -7). ROKs or Rho kinases are downstream effectors of RhoA in organizing stress fibers (4 -6), whereas MRCKs 1 play an important role in Cdc42 functions in regulating actinomyosin dynamics in cultured cells (7). Precisely how these occur is not clear, although a number of proteins are known to be effective substrates for these kinases. These include the non-muscle myosin light chain 2 (MLC2), whose phosphorylation state is crucial for actinomyosin contractility and polymerization (7,8), the myosin binding subunit p130 (9 -11), Ezrin, Radixin, and Moesin (ERM) family proteins (12), adducin (13), and intermediate filament proteins (14 -17), which are directly or indirectly linked to the actin cytoskeleton.
In particular, the myosin binding subunit MBS130 appears to play a unique role in the regulation of the activity of the associated PP1 catalytic subunit. The specific binding of MBS130 to RhoA may link this regulatory subunit to Rho-dependent events (9). Indeed chicken gizzard MBS130 is found to be effectively and specifically phosphorylated by ROK at threonine 695 and serine 854 (10,11). Phosphorylation at threonine 695 resulted in inhibition of the intrinsic phosphatase activity. Other proteins that can interact with and phosphorylate MBS130 include the cGMP-dependent protein kinase 1␣ and an unidentified mitotic kinase, but in contrast, such phosphorylation resulted in activation of protein phosphatase activity (18,19). MBS130 therefore appears to serve as a scaffold for multiple protein interactions as well as phosphorylation regulation. Indeed, a recent report has indicated that a number of proteins involve in the Rho signaling pathways including Ezrin, Radixin and Moesin family proteins, adducin, and MBS130 can be found to colocalize at cell periphery upon stimulation. It is possible that MBS130 may provide a bridge for various Rho-dependent components to function in a coordinated manner (20). Since the myotonic dystrophy kinase-related Cdc42-binding kinases (MRCKs) also consist of similar catalytic domains with differential cellular localization and cellular functions (7,21), it is of interest to investigate if these kinases could share similar and different sets of substrates for their cellular activities.
As a first step toward identification of new substrates for kinases such as MRCKs, we derived a filter assay to screen for potential candidates. This assay allows renatured proteins on the filter to be phosphorylated by specific recombinant kinases. Using this assay for MRCK␣ kinase, we identified two potential substrate proteins, one of which was MBS130; the other was a novel related protein.
Cell Culture, Transfection, and Cell Staining-COS-7 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, and HeLa cells were cultivated in modified Eagle's medium with 10% fetal bovine serum (Hyclone). Subconfluent HeLa cells plated on coverslips for 48 h were transfected with various HA-or FLAGtagged DNA constructs (1 g/ml) with LipofectAMINE (Life Technologies, Inc.) according to recommended protocol. 16 h after transfection, cells were fixed with 4% paraformaldehyde and stained with the combination of various primary antibodies: anti-HA (12CA5; Roche Molecular Biochemicals) or anti-FLAG (M2; Sigma). Stained cells were analyzed with an MRC 600 confocal imager adapted to a ZEISS Axioplan microscope. For localization studies, transfected HeLa cells were serum-starved for 4 -6 h before treatment with lysophosphatidic acid (300 ng/ml; Sigma) or phorbol myristic acetate (100 ng/ml; Sigma). COS-7 cells grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum were similarly transfected with various constructs. 24 h after transfection, cell extracts were obtained with lysis buffer ((25 mM HEPES, pH 7.7, 0.15 M NaCl , 1.5 mM MgCl 2, 0.2 mM EDTA, 1 mM sodium vanadate, 20 mM ␤-glycerol phosphate, 5% glycerol, 0.1% Triton X-100 and 1ϫ inhibitor mix (Roche Molecular Biochemicals)), separated on a SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with anti-HA or anti-FLAG antibodies for expression.
Enzymatic Measurements-Kinase assays were carried out in kinase buffer containing 20 mM Tris-HCl pH 7.5, 75 mM NaCl, and 10 mM MgCl 2 as previously described (5). For labeling of GST-MLC2, purified fusion protein (10 g) on glutathione beads was subjected to phosphorylation using 20 g/ml GST-MRCK-␣-CAT and 1 M [␥-33 P]ATP for 1 h at 30°C. After an extensive wash with GST purification buffer, the phosphorylated protein was eluted with 5 mM reduced glutathione. Phosphatase assays were carried out at 30°C using 33 P-GST-MLC2 as substrate. To show phophorylation-mediated PP1 inhibition, the immunoprecipitated FLAG-p85⅐HA-PP1␦ complex was first incubated in 20 l of kinase buffer with 0.15 mM ATP␥S in the presence or absence of 0.5 g of GST-MRCK␣-CAT for 30 min at 30°C. Phosphatase assays were initiated by the addition of 5 M 33 P-GST-MLC2 in 30 mM Tris-HCl, PH 7.5, 0.1 M KCl, 2 mM MgCl 2 , and 0. 1 mg/ml bovine serum albumin. The reactions were stopped by adding an equal volume of SDS sample buffer at each time point indicated and boiling for 5 min before gel loading. To show PP1 inhibition by phosphorylated GST-PIM50, 10 g of the fusion protein was first incubated in kinase buffer (as above using ATP␥S) with and without GST-MRCK␣-CAT for 30 min at 30°C. These phosphorylated and nonphosphorylated GST-PIM50 proteins were then separately mixed with the immunoprecipitated FLAG-p85⅐HA-PP1␦ complex and preincubated for 1 min before the start of phosphatase assays. Phosphatase activities were quantified using the Molecular Dynamics PhosphorImager System.
Immunoprecipitation and in Vitro Binding Assays-COS-7 cells coexpressing various FLAG-p85 and HA-PP1␦ constructs were lysed in buffer containing 25 mM HEPES, pH 7.3, 0.15 M NaCl, 0.5 mM MgCl 2 , 0.2 mM EDTA, 20 mM ␤-glycerol phosphate, 1 mM sodium vanadate, 5% glycerol, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 0.1% Triton, 1 g/ml each aprotinin, leupeptin, and pepstatin A, and 1ϫ protease inhibitor mixture and incubated with anti-FLAG-conjugated agarose beads (Sigma) for 1 h at 4°C. After an extensive wash, the immunoprecipitates were resolved by SDS-polyacrylamide gel electrophoresis and subjected to immunoblot analysis using anti-FLAG or anti-HA antibodies as mentioned. To detect any interaction of these complex with the phosphorylated GST-PIM50, an in vitro assay containing either the unphosphorylated or phophorylated GST-PIM50 was added to COS-7 cell lysates co-expressing FLAG-p85 AA and HA-PP1␦ and followed by immunoprecipitation using anti-FLAG-conjugated agarose beads. Immunoblot analyses were carried out as before.
RNA and Protein Analysis-Northern blots containing mRNA from various human tissues were obtained from CLONTECH and were hybridized with a full-length 32 P-labeled p85 probe as previously described (5). One-and two-dimensional gel analyses were carried out according to standard protocol, and separated proteins were transferred to PVDF membranes (PerkinElmer Life Sciences) and probed with various antibodies, including an antibody to a phosphorylated threonine 695 of chicken gizzard M130 (which is identical to the flanking sequence around threonine 560 of p85). Analysis of tryptic peptides and phosphopeptides were performed with a QSTAR Mass-spectrometer (PerkinElmer).

Identification of p130 and p85 as Major in Vitro Substrates
for MRCK␣-CAT-To identify and characterize potential substrates for ROK␣ and MRCK␣, we made GST fusion proteins of the catalytic domains of both ROK␣ (1-432) and MRCK␣ . The yield and catalytic activities of MRCK␣-CAT were consistently higher than ROK␣-CAT, and subsequent experiments were thus carried out with GST-MRCK␣-CAT. Here we observed that renatured proteins separated on SDS-polyacrylamide electrophoresis gels and transferred onto PVDF membrane filters were readily phosphorylated by MRCK␣-CAT.
Two proteins of 130 and 85 kDa were prominently and specifically phosphorylated by MRCK␣-CAT ( Fig. 1) but not by ␣-p21-activated kinase or protein kinase A (data not shown). These two proteins are not abundant in tissues such as brain and testis but could easily be enriched by passing through an affinity dye Reactive Brown 10-Sepharose column (Sigma), and this constitutes a simple one-step enrichment of these proteins for further purification (Fig. 1A). Further separation of these two proteins was achieved with two-dimensional gel electrophoresis (Fig. 1B), and the Coomassie Blue-stained spots corresponding to the phosphorylated proteins were excised for peptide sequencing.
P85 Is a Novel Protein Related to Myosin Binding Subunit Protein MBS130 -Both peptide sequencing and subsequent immunoblotting analyses showed that the p130 corresponds to the previously reported MBS130 (data not shown). Peptide sequence analysis also indicated that the p85 is a novel protein ( Fig. 2A). A 2.8-kilobase human EST clone AI825921 contains most of the coding sequence except for the extreme N terminus. The complete match of this cDNA with two overlapping genomic clones, S51329 and AC005782, from human chromosome 19 (Fig. 2C) suggested that the N terminus is confined within the first exon. A full-length cDNA for p85 was thus constructed from the PCR product of the first exon of human p85 genomic DNA and subsequently joined to the truncated cDNA. The amino acid sequence derived from this cDNA indicated that p85 is structurally related to MBS130 (Fig. 2B). The N terminus of p85 consists of a closely related structure with 6 ankyrin repeats and 48% identity to MBS130, which has been reported to have 7-8 repeats (23,24). A putative PP1 binding consensus sequence, RTVRF (25), was also present immediately before the ankyrin repeats. The C terminus contains a conserved ␣-helical structure with leucine zippers at the Cterminal end (␣LZ). This structure can form homodimers or heterodimers (e.g. with M20, which also contains this structural motif (26)). Of most striking similarity (87% identity) is a central motif, which contains the sole phosphorylation site for MRCK␣-CAT (refer to Fig. 3).
The genomic organization of p85 showed that the mRNA is derived from 22 exons (Fig. 2C). Intriguingly, a number of integration hotspots (AAVS1) of adenovirus-associated virus (AAV) was found in the first exon/intron regions of p85 genomic sequence ( Fig. 2D; Refs. 27 and 28). The consequence of these integrations is not known but is expected to disrupt the expression of this gene. Rearrangements and disruption of a nearby troponin gene were also observed upon adenovirus-associated virus integration (29). Northern blot analysis indicated that the 3-kilobase p85 mRNA was ubiquitously expressed and is especially high in the heart (Fig. 2E).
P85 protein expressed in serum-starved HeLa cells mainly showed cytoplasmic punctate distribution (Fig. 2F, a) but was readily redistributed to cell peripheral upon treatment with lysophosphatidic acid and phorbol myristic acetate (Fig. 2F, b  and c).
Identification of the Phosphorylation Site of p85 on Threonine 560 by MRCK␣-CAT-To confirm the nature of phosphorylation of p85, we expressed the FLAG-tagged wild-type and deletion mutants of p85 in COS-7 cells (Fig. 3A). The immuno-precipitated proteins were phosphorylated with MRCK␣-CAT to map the phosphorylation site(s). It is clear that the major site(s) is within the central conserved region, as mutants de- leted of this region were not phosphorylated (Fig. 3B, lanes 2  and 4). To confirm this, we expressed GST fusion protein containing the wild-type conserved PIM motif (PIM50; also refer to Fig. 4) and a mutant S559A/T560A (PIM50 AA ) and showed that the mutant was not phosphorylated (Fig. 3B, lane 8), unlike the wild-type control (lane 7). This in vitro phosphorylation site of p85 on threonine 560 by MRCK␣-CAT was further confirmed by mass spectrometry on the tryptic phosphopeptides derived from p85 phosphorylation.
Further evidence was obtained by probing p85 protein expressed from COS-7 cells that were co-transfected with either MRCK␣-CAT or ROK␣-CAT with an antibody that specifically recognizes phosphorylated threonine 560 of p85. Clearly, threonine 560 of p85 can be phosphorylated by both MRCK␣-CAT and ROK␣-CAT in vivo (Fig. 3C).
p85 Is Specifically Associated with PP1␦ and Its Substrate MLC2, and the Phosphorylation by MRCK␣ Regulates the Phosphatase Activity-To see if p85 can associate with PP1, we co-transfected FLAG-tagged p85 and HA-tagged PP1␣, -␥, and -␦ isoforms. Only PP1␦ isoform was immunoprecipitated with p85, indicating a specific interaction between p85 and PP1␦. This was also evident from the co-immunoprecipitation of the endogenous PP1␦ with the overexpressed p85 (data not shown). Constructs with an intact N terminus, including the ankyrin repeats, could effectively interact with PP1␦ (Fig. 4A, lanes  7)) were co-expressed with HAtagged PP1␦ in COS-7 cells. Proteins immunoprecipitated (IP) with mouse anti-FLAG beads were separated on 10% SDS-polyacrylamide electrophoresis gels, transferred to PVDF membranes, and detected with a rabbit anti-FLAG antibody for p85 and a rabbit anti-HA antibody for associated PP1␦. Overexpressed PP1␦ present in the cell lysate was also shown for comparison. WB, Western blot. B, the N terminus of p85 binds MLC-2. FLAG-tagged p85 NT (lane 1) or p85 LZ (lane 2) was co-expressed with HA-tagged MLC-2. Immunoprecipitated proteins with anti-FLAG antibody transferred onto PVDF membrane were detected with anti-FLAG or anti-HA as described in A. C, phosphorylation of p85 by GST-MRCK␣-CAT inhibits associated PP1␦ activities. Immunoprecipitated wild-type p85 (p85 WT ) or a phosphorylation-deficient mutant, p85 S559A/T560A (p85 AA ), coexpressed with PP1␦ were phosphorylated with GST-MRCK␣-CAT in the presence of 0.1 mM ATP␥S. The nonphosphorylated and phosphorylated proteins were used to initiate the dephosphorylation activities of the associate PP1␦ toward 33 P-MLC2 at different time intervals. [1][2][3][4], whereas a deletion mutant of a single ankyrin repeat (lane 5) can dramatically reduce such interaction. Deletion mutants devoid of N terminus are totally ineffective (Fig. 4A,  lanes 6 and 7).
The N terminus of p85 could also interact with MLC2 (Fig.  4B), and the C terminus is totally ineffective. Hence PP1␦ can form a tight complex with p85 and substrate MLC2 through its N terminus.
Next, to examine if the phosphorylation of threonine 560 by MRCK␣-CAT can regulate PP1 activity, we measured the time course of dephosphorylation toward 33 P-MLC2. As shown in Fig. 4C, nonphosphorylated wild-type p85 (p85 WT ) or the phosphorylation-defective p85 mutant (p85 AA ) were equally active in MLC2 dephosphorylation. Wild-type p85 but not the mutant p85 AA , when phosphorylated in vitro with MRCK␣-CAT, showed significant reduction in the rate of MLC2 dephosphorylation. These results confirm a similar observation with MBS130 where phosphorylation of a conserved threonine 695 within a highly conserved motif was essential for the inhibition of phosphatase catalytic activity (11). Based on the biochemical and functional similarities between p85 and MBS130, we therefore designate p85 as MBS85.
N and C Termini of MBS85 Show Independent Morphological Effects Reflecting Their Biochemical Activities-Because PP1 activities are essential for regulating the phosphorylation states of myosin, it is likely that the biochemical interaction of MBS85 with PP1␦ may be correlated with morphological effects in cultured cells. Furthermore adherent cells mainly exhibit active Rho phenotype in serum medium, and the interference of endogenous PP1␦ would be expected to affect this actin structure. As shown in Fig. 5A, expression of wild-type p85 alone did not give any obvious morphological consequence. However, truncation of the C terminus led to various degrees of disassembly of actin stress fibers, with the most striking effects when both the PIM and ␣LZ motifs were totally removed (p85 NT ; Fig. 5, A and B). Deletion of the PIM motif (p85 ⌬Nar1 ) or mutation of the phosphorylation site (p85 AA ) was significantly effective in inducing similar morphological effects, although this appears to depend on the levels of expression (Fig. 5A,  bottom panel). Both the PIM and ␣LZ motifs at the C terminus appear to exert opposing effects on the N-terminal function. Truncation of the N terminus resulted in general increases in actin polymerization (p85 CT in Fig. 5A) that was resistant to C3 treatment (data not shown). As ␣LZ alone was also effective, although to a lesser extent (data not shown), it is likely that PIM and ␣LZ domains have independent activities toward actin assembly. Similar trends were also observed when the various constructs were co-expressed with PP1␦ (Fig. 5B). In this case, the morphological effects were more pronounced in the presence of exogenous PP1␦. As expected, both ankyrin repeat mutants p85 ⌬AR1 and p85 ⌬AR2 that were defective in binding to PP1␦ were totally ineffective in inducing morphological changes (Fig. 5B and data not shown). Hence the phenotypic effects of the various MBS85 variants on actin cytoskeleton correlate well with the biochemical data described earlier.
Phosphorylation Inhibitory Motif (PIM50) of MBS85 Binds PP1␦ When Phosphorylated by MRCK␣-CAT and Exerted an Inhibitory Effect on PP1␦ Activity-It is known that phosphorylation of threonine 695 of MBS (which is equivalent to the threonine 560 of p85) is critical for the inhibitory effects on PP1 catalytic function (11). It is likely that a similar mechanism may operate for MBS85. To test this possibility, we derived an assay to examine the effect of phosphorylation on the interaction of the highly conserved GST-PIM with p85/PP1␦ complex. The phosphorylation-deficient mutant p85 AA was used to eliminate possible competition for binding. As shown in Fig. 6A, phosphorylated GST-PIM, but not the nonphosphorylated form, was detected in the p85/PP1␦ immuno-complex. Similarly only glutathione beads with the phosphorylated GST-PIM, but not the nonphosphorylated or phosphorylation-deficient mutant GST-PIM50 AA , were able to pull down the expressed PP1␦ or p85/PP1␦ complex (Fig. 6B), clearly indicating that phosphorylation of threonine 560 of MBS85 is essential for its interaction with PP1␦.
To examine if such interaction is functional, we measured the catalytic activities of the immuno-complex in the presence of in vitro phosphorylated or nonphosphorylated GST-PIM50. Phosphorylated GST-PIM50, but not the nonphosphorylated form, was likewise effective in inhibiting the catalytic activity of p85⅐PP1␦ complex (Fig. 6C). This provides further evidence that the interaction is functional.
Next we tested the effects of co-expression of this PIM50 motif on the p85 NT -induced morphological changes. As shown in Fig. 7, A and B, PIM50 expression is sufficient in reversing the effect of p85 NT -induced actin stress fiber losses. The phosphorylation-deficient mutant PIM50 AA was totally inefficient (Fig. 7B). These inhibitory effects became less pronounced when p85 NT was co-expressed with PP1␦, suggesting that PIM50 may well be competing with the catalytic subunit in regulating actin dynamics. We therefore conclude that from both biochemical and morphological data that the central conserved motif of MBS85 can be regulated by phosphorylation, resulting in conformational changes that affect the associated PP1␦ catalytic property and subsequently effects on actin morphology, probably through the eventual effects on myosin phosphorylation.

DISCUSSION
Two candidate proteins from rat brain cytosolic extract were identified in this study on the basis of their in vitro phosphorylation by MRCK␣-CAT. The p130 protein was confirmed to be MBS130 by peptide sequencing and immunoreactive toward specific antibodies. The smaller protein p85 is a novel protein that is structurally related to MBS130. Overall p85 shares low FIG. 6. Phosphorylated PIM50 binds and inhibits PP1␦ activity. A, phosphorylated GST-PIM50 by MRCK␣-CAT and ATP␥S or nonphosphorylated GST-PIM50 control was preincubated with COS-7 cell extract expressing FLAG-tagged p85 AA and HA-tagged PP1␦ before immunoprecipitation (IP) with anti-FLAG beads. Immunoprecipitated proteins on Western blots (WB) were detected with various anti-tag antibodies. GST-PIM50 in total lysate is shown in the bottom panel for comparison. B, GST-PIM50 or GST-PIM50 AA on glutathione beads was phosphorylated with MRCK␣-CAT and ATP␥S. Nonphosphorylated GST-PIM50 was used as the control. A pull-down assay was performed by incubating these beads with extracts expressing PP1␦ alone (left lanes) or PP1␦ together with p85 AA (right lanes). Immunoblots of the bound proteins were detected with the anti-tag antibodies. A nonspecific band recognized by the HA in total extract was marked by an asterisk. C, time course of dephosphorylation of 33 P-MLC2 by p85 AA and PP1␦ immunoprecipitates in the presence of phosphorylated and nonphosphorylated GST-PIM50 was assayed as described under Fig. 4B. similarity to MBS130 (Ͻ40%). The N terminus of p85 contains six ankyrin repeats that are known to be involved in proteinprotein interactions. This motif shares a 48% identity to MBS130, which has 7-8 repeats (23,24). Preceding these repeats is a short stretch (RTVRF) that resembles PP1 binding consensus sequence (25). The C terminus of p85 is also conserved and consists of an ␣-helical structure with four leucine heptad repeats at the C-terminal end. This motif is known to be involved in dimerization and interaction with other proteins. M20, a small subunit protein of PP1, is found to be an integral part of the heterotrimeric complex (28), and it contains a similar helical structure. Indeed, it has been reported that this M20 may be a spliced product from skeletal muscle isoform of MBS130 (30). More recently, it has also been reported that other proteins can interact with the C terminus of MBS130. This includes RhoA and cGMP-dependent protein kinase 1␣ (9,18). The former links the PP1 complex to Rho-dependent regulatory events, and the kinase phosphorylates and activates PP1 activity. The presence of a similar motif in p85 at the C terminus suggests that it may serve similar functional roles. Indeed p85 can readily translocate to peripheral membrane upon treatment with lysophosphatidic acid and phorbol ester, factors that are known to have prominent effects on Rho GT-Pases and actin cytoskeleton (2).
A central conserved motif that contains the sole phosphorylation site for MRCK␣ and ROK␣ exhibits a more striking similarity. Threonine 560 of p85 is equivalent to threonine 695 of MBS130 and is located within a ϳ50-amino acid conservative phosphorylation inhibitory motif (PIM50). Phosphorylation of the threonine 695 by ROK␣ has been shown to inhibit associated PP1 activity (11,30). A second phosphorylation site (serine 854) by ROK␣ has also been described in MBS130, but this is absent in p85 ( Fig. 2A and Ref. 10).
Here we have also demonstrated that p85 is functionally similar to MBS130. First it is specifically associated with PP1␦, and this depends on the N terminus including the whole of the ankyrin repeats. N-terminal as well as ankyrin-repeat deletion mutants are ineffective in binding PP1␦. In agreement with a previous report for MBS130 (32), we also detected binding of the substrate MLC2 within the N-terminal region. In this respect, the N terminus of p85 alone can therefore act as scaffold for PP1 and myosin and confers specific phosphatase activity on MLC2 substrate (30). p85 (designated here as MBS85) is therefore a genuine myosin binding subunit of PP1␦, similar to MBS130. Furthermore, the phosphorylation of thre-onine 560 of the central motif of MBS85 resulted in inhibition of PP1␦ activity, also suggesting a conservation of phosphorylation mechanism in regulating the catalytic event.
Similar conclusions can be derived from the morphological assays that clearly reflect the biochemical interactions. HeLa cells expressing the N terminus of MBS85 alone exhibits the greatest loss in actin stress fibers, an indicator for activation of PP1 in vivo. Interestingly, the two conserved motifs (PIM and ␣LZ) appear to act independently to counteract this N-terminal function as deletion of either motif attenuates the stress fiber losses. Overexpression of C-terminal ␣LZ motif alone can induce actin stress fibers that are resistant to C3 treatment, suggesting that this activates an event downstream of Rho. The exact mechanism of how this ␣LZ motif works is currently not clear. One possibility is that this motif may directly compete with binding proteins such as cGMP-dependent kinase 1␣, which is known to activate PP1 (18). Furthermore, the ␣LZ motif is known to be able to form homodimers or heterodimers with the small subunit M20, although the biological functions of these complexes are currently not known (data not shown; Refs. 26 and 30). We conclude from both biochemical and morphological analyses that different domains of the MBS85 have opposing activities in regulating actin polymerization and that the phosphorylation of threonine 560 in a central motif appears to fine tune these events.
It is therefore important to demonstrate a direct interaction of the conserved central motif (PIM50) with PP1␦. Such an interaction occurred only when the threonine 560 was phosphorylated. Moreover, not only did such a motif play a role in intramolecular interaction, it is also equally effective both in vitro and in vivo in promoting intermolecular interactions when introduced separately. This leads to the conclusion that phosphorylation of MBS85 by myotonic dystrophy kinase-related kinases such as MRCK␣ and ROK␣ induces a conformational change at the central conserved motif that results in higher affinity toward PP1␦. Such an interaction may change the orientation or catalytic properties of PP1␦ toward the associated myosin (through MLC2; Fig. 6C), resulting in myosin phosphorylation and subsequent cytoskeletal changes. This is depicted as a working model in Fig. 8. It remains unclear as to which GTPase(s) and downstream kinase(s) are regulating this event, as endogenous levels of p85 in cultured cells are much lower than p130MBS, and the specific antibody that recognizes the phosphorylated peptide was unable to detect the endogenous phosphoprotein. Further experiments are therefore required to clarify this issue.
In summary, we have isolated a novel myosin binding subunit that is ubiquitously expressed. Compared with MBS130, the smaller size and simpler arrangement of the regulatory domains of this novel MBS85 allow an easier analysis of structure and function relationships. The identification of an increasing number of these myosin binding subunits, which share a similar regulatory mechanism, should help to understand how each of these are regulated by various diverse signaling pathways in the control of the actin cytoskeleton. FIG. 8. A model for the phosphorylation regulation of threonine 560 of p85MBS on PP1␦ activity. When threonine 560 (T) MBS85 is not phosphorylated, PP1␦ assumes an orientation in contact with its substrate MLC2, resulting in an active conformation for the dephosphorylation of MLC2 and subsequent actin-myosin disassembly. Upon phosphorylation (T with an asterisk), it presents a conformation that has a higher affinity to PP1␦ and disrupts its accessibility or catalytic activity toward MLC2, resulting in a shutdown of MLC2 dephosphorylation that favors myosin phosphorylation and actin-myosin assembly. AR, ankyrin repeats.