During nitric oxide signaling, type Iα cGMP-dependent protein kinase (PKGIα) activates myosin light chain (MLC) phosphatase through an interaction with the 130-kDa myosin targeting subunit (MYPT1), leading to dephosphorylation of 20-kDa MLC and vasodilatation. It has been suggested that the MYPT1-PKGIα interaction is mediated by the COOH-terminal leucine zipper (LZ) of MYPT1 and the NH2-terminal LZ of PKGIα (HK Surks and ME Mendelsohn. Cell Signal 15: 937–944, 2003; HK Surks et al. Science 286: 1583–1587, 1999), but we previously showed that PKGIα interacts with LZ-positive (LZ+) and LZ-negative (LZ−) MYPT1 isoforms (13). Interestingly, PKGIα is known to preferentially bind to RR and RK motifs (WR Dostmann et al. Proc Natl Acad Sci USA 97: 14772–14777, 2000), and there is an RK motif within the aa 888–928 sequence of MYPT1 in LZ+ and LZ− isoforms. Thus, to localize the domain of MYPT1 important for the MYPT1-PKGIα interaction, we designed four MYPT1 fragments that contained both the aa 888–928 sequence and the downstream LZ domain (MYPT1FL), lacked both the aa 888–928 sequence and the LZ domain (MYPT1TR), lacked only the aa 888–928 sequence (MYPT1SO), or lacked only the LZ domain (MYPT1TR2). Using coimmunoprecipitation, we found that only the fragments containing the aa 888–928 sequence (MYPT1FL and MYPT1TR2) were able to form a complex with PKGIα in avian smooth muscle tissue lysates. Furthermore, mutations of the RK motif at aa 916–917 (R916K917) to AA decreased binding of MYPT1 to PKGIα in chicken gizzard lysates; these mutations had no effect on binding in chicken aorta lysates. However, mutation of R916K917 to E916E917 eliminated binding, suggesting that one factor important for the PKGIα-MYPT1 interaction is the charge at aa 916–917. These results suggest that, during cGMP-mediated signaling, aa 888–928 of MYPT1 mediate the PKGIα-MYPT1 interaction.
- myosin light chain phosphatase
- nitric oxide
- smooth muscle
- calcium desensitization
- cGMP-dependent protein kinase
steady-state force during smooth muscle contraction is dependent on the level of phosphorylation of myosin regulatory light chain [20-kDa myosin light chain (MLC20)], which is determined by the activities of myosin light chain (MLC) kinase and MLC phosphatase (11). Ca2+-calmodulin activation of MLC kinase has been proposed to be the primary event for regulation of MLC20 phosphorylation, and MLC phosphatase was thought to be constitutively active (12, 22). However, recent studies have shown that MLC phosphatase is an important target for the physiological regulation of vascular tone; MLC phosphatase activity can be stimulated to produce Ca2+ desensitization and inhibited to produce Ca2+ sensitization (reviewed in Refs. 12 and 22).
MLC phosphatase is a trimeric enzyme consisting of a 130-kDa myosin targeting subunit (MYPT1), a 37-kDa catalytic subunit (PP1cδ), and a 20-kDa subunit of unknown function (M20) (12). MYPT1, the primary site for regulation of phosphatase activity, has multiple binding and phosphorylation sites, which allow its function to be tightly controlled. Multiple pathways feed into this regulation, including G protein-coupled signaling and nitric oxide (NO) (reviewed in Refs. 12 and 22).
NO-mediated vasodilatation is a fundamental response of the vasculature and is the classical model for Ca2+ desensitization (9, 10). NO diffuses into the smooth muscle cells and activates guanylate cyclase, thereby increasing the production of cGMP. This second messenger can then bind to and activate type Iα cGMP-dependent protein kinase (PKGIα), which interacts with multiple targets within the smooth muscle cell, including the L-type Ca2+ channel (8), maxi-K+ channels (1), sarcoplasmic reticulum channels (20), telokin (4, 25), and MYPT1 (13, 23, 24), all of which have been demonstrated to produce smooth muscle relaxation.
MYPT1 (Fig. 1A) contains two putative COOH-terminal coiled-coil (CC) domains at aa 715–746 and aa 929–969, which have 60% and 81% probabilities of CC formation, respectively (16), and a third domain at aa 1013–1039, which is known as the leucine zipper (LZ) (12). These sequences are numbered according to the M133 MYPT1 isoform sequence reported by Shimizu et al. (21), which is central insert (CI) positive (CI+, aa 512–553), whereas the M130 isoform, which is expressed in chicken gizzard, is CI negative (CI−); therefore, numbering of the sequences is displaced toward the NH2 terminus by 41 amino acids (5, 21). Studies with glutathione S-transferase (GST)-fusion peptides have demonstrated that the NH2-terminal LZ of PKGIα (24) and COOH-terminal LZ of MYPT1 (23) are important for the PKGIα-MYPT1 interaction. However, using full-length MYPT1, we have demonstrated that although the LZ domain is important for cGMP-mediated activation of MLC phosphatase, it may not mediate the binding of PKGIα to MYPT1 (13). Therefore, the MYPT1-PKGIα interaction may depend on the presence of an upstream domain.
MYPT1 (23) and PKGI (17, 19) have been shown to exist as homodimers and may undergo CC exchange to form a heterotetrameric complex that enables the kinase to bind to and activate the MLC phosphatase. However, other studies have demonstrated that PKGIα preferentially binds to its targets at RR and RK motifs (6, 7). Although neither the MYPT1 CC domain at aa 929–969 nor the COOH-terminal LZ contains either amino acid combination, there is an RK motif within a sequence (aa 888–928) just upstream of the CC domain. Thus we hypothesized that the MYPT1-PKGIα interaction may be mediated by R916K917. In this study, we determined whether the MYPT1 aa 888–928 sequence is necessary for the MYPT1-PKGIα interaction.
MATERIALS AND METHODS
Cloning of MYPT1 fragments.
Four MYPT1 fragments (Fig. 1A) were subcloned from the full-length chicken aorta MYPT1 cDNA (13, 23, 24). The production of four endogenous MYPT1 isoforms depends on the presence or absence of a CI and a downstream LZ (12). PCR was performed to generate the COOH-terminal sequence (aa 500 to COOH terminus) of the four naturally occurring MYPT1 isoforms. These isoforms and the corresponding fragments are denoted MYPT1 CI+/LZ+, MYPT1 CI+/LZ−, MYPT1 CI−/LZ+, and MYPT1 CI−/LZ−. These clones were sequenced on both strands, and antibodies to endogenous MYPT1 were able to recognize all four protein fragments. The MYPT1 CI+/LZ+ cDNA was used as a template to clone a second set of MYPT1 fragments, leading to the production of four MYPT1 protein fragments denoted MYPT1FL, MYPT1TR, MYPT1SO, and MYPT1TR2. For MYPT1FL, PCR was performed using a primer pair that amplified MYPT1 cDNA encoding aa 500 to the COOH terminus (aa 1039). This MYPT1 fragment contained the CI identified by Shimizu et al. (21), the RK motif at aa 888–928, and the LZ domain. The PCR product was then cloned into the pAED4 expression plasmid vector and purified using the QIAprep 8 Miniprep kit (Qiagen). The MYPT1FL cDNA was then used as a template to clone MYPT1TR, MYPT1SO, and MYPT1TR2. Various mutagenesis primers were used to 1) truncate MYPT1FL at aa 890 to obtain MYPT1TR (lacking aa 888–928, the putative CC domain, and the LZ domain), 2) modify MYPT1FL to remove the aa 888–928 sequence to obtain MYPT1SO (lacking aa 888–928 but containing the CC and LZ domains), and 3) truncate MYPT1FL at aa 1010 to obtain MYPT1TR2 (containing aa 888–928 and the CC domain but lacking the LZ domain). MYPT1FL cDNA was used as the template to change R916 and K917 to alanine (MYPT1A) or glutamic acid (MYPT1E) by site-directed mutagenesis using the QuickChange II kit (Stratagene). All subclones were sequenced in full to verify the mutagenesis protocols. Sequences of the MYPT1 fragments used in this study are numbered according to the M133 MYPT1, or CI+, isoform (21). The adult chicken gizzard MYPT1 isoform (M130) lacks the 41-amino acid CI at aa 512–553 (5); thus the comparable sequences would be displaced upstream by 41 amino acids.
Expression and purification of MYPT1 fragments.
We performed large-scale expression of MYPT1TR in the BL21 Star Escherichia coli strain (Invitrogen). Six liters of 2XTY culture medium grown to 0.4 unit at 600-nm optical density was induced with 0.2 mM isopropylthiogalactoside and grown for an additional 3 h. The cultures were centrifuged for collection of the bacteria, which were lysed by sonication in a phosphate buffer solution (50 mM phosphate buffer, 1 mM EDTA, 1 mM PMSF, and 2 mM DTT). The lysate was centrifuged at 18,000 rpm at 4°C for 20 min, and ammonium sulfate extractions (15%, 35%, and 65%) of the supernatant were performed. The ammonium sulfate fraction containing the MYPT1TR protein was resuspended in 30 ml of 0.1 mM EDTA, 20 mM MES, 2 mM β-mercaptoethanol, and 0.1 mM PMSF (pH 6) and dialyzed thoroughly against the same buffer. After dialysis, urea was added to 6 M, and the sample was filtered through a syringe and loaded onto a similarly equilibrated cation-exchange column (Q Sepharose, GE Healthcare). The sample was eluted using a linear 500 mM NaCl gradient, and fractions containing the protein were collected, dialyzed against ammonium bicarbonate, and lyophilized. The fraction was resuspended in gel filtration buffer [6 M urea, 0.15 M NaCl, 0.2 M KPO4, 0.1 mM EDTA, 0.1 mM PMSF, and 0.2 mM β-mercaptoethanol (pH 7)] and loaded onto a similarly equilibrated gel filtration column (Superdex 200, GE Healthcare). Fractions containing the purified protein of interest were dialyzed against ammonium bicarbonate and lyophilized. A similar protocol was used in the expression and purification of MYPT1FL, MYPT1SO, MYPT1TR2, MYPT1A, and MYPT1E.
Two-dimensional gel electrophoresis.
Two-dimensional gel electrophoresis was performed as previously described (3). A total protein homogenate from chicken gizzard tissue was prepared by extraction with 7 M urea, 2 M thiourea, 4% CHAPS, 0.5% IPG buffer pH 4–7 (GE Healthcare), 1 mM EDTA, and 1× Roche Complete protease inhibitor. Suitable amounts of the homogenate were resolved on IPG gel strips (pH 4–7) in the “face-up” mode on an Ettan IPGPhor II system (GE Healthcare). After completion of the first dimension, proteins were resolved in the second dimension by 29:1 and then stained with silver and subjected to 8% SDS-PAGE or exposed to appropriate PKGI antibodies (Stressgen) and subjected to Western blotting. The two-dimensional profile of PKGI from the total chicken gizzard homogenate was identical to that of the enriched PKGI fraction (data not shown), suggesting that the partial purification of PKGI did not favor selection of one isoform, or phosphorylated form, over another.
Preparation of PKGIα.
A protein fraction enriched in PKGIα was prepared from chicken gizzard by a modification of the method described by Nakazawa and Sano (18). Briefly, chicken gizzards were homogenized in 5 vol of buffer A [10 mM potassium phosphate (pH 7.0), 2 mM EDTA, 10% glycerol, 50 mM β-mercaptoethanol, and 1× Roche Complete protease inhibitor], allowed to stir on ice for 20 min, and centrifuged at 15,000 g for 30 min at 4°C. The supernatant was adjusted to pH 5.2 by dropwise addition of 1 N acetic acid, allowed to stir on ice for 15 min, and centrifuged at 20,000 g for 20 min for collection of the precipitate. For resolubilization of protein kinase G from the precipitate, the supernatant was dissolved into a small volume (∼20 ml) of buffer A, with periodic addition of K2HPO4 to ensure that pH was maintained at 7.0. This soluble fraction was applied to a Sepharose Q column equilibrated with buffer A and developed with a linear gradient of 0–500 mM NaCl. Fractions enriched in cGMP-dependent protein kinase (PKG) and with low levels of contaminants were identified by SDS-PAGE and Western blotting.
Adult chicken gizzard and aorta were dissected and homogenized in 10 ml of lysis buffer [8 M urea, 10 mM Tris·HCl, 0.1 mM EDTA, and 1× Roche EDTA-free Complete protease inhibitor (pH 8)], rotated for 15 min at 4°C, and spun, and the supernatants were collected. Small (20–40 μl) aliquots of these homogenates were added to 1 ml of immunoprecipitation buffer [50 mM Tris·HCl (pH 8), 7 mM MgCl2, 2 mM EDTA, and 1 mM PMSF], along with 4 μg of MYPT1 fragment protein, rotated at 4°C for 120 min, and centrifuged, and the supernatants were collected. An anti-PKGIα/β antibody (Stressgen) was added to the clarified samples, which were allowed to rotate at 4°C overnight. Protein G-Sepharose beads were then added, and the samples were incubated at 4°C for 120 min. The immune complexes bound to the protein G beads were collected by centrifugation and washed twice with 200 μl of immunoprecipitation buffer. The bound complexes were removed from the beads by heating in 30 μl of SDS sample buffer. The recovered protein samples were resolved by 29:1 and then subjected to 10% SDS-PAGE or to Western blotting using a polyclonal anti-MYPT1 antibody (Upstate Biotechnologies).
To investigate whether the MYPT1-PKGIα interaction is dependent on the MYPT1 LZ domain, an upstream CC domain, or the MYPT1 aa 888–928 sequence, which contains a PKGIα binding motif, we expressed and purified four MYPT1 fragment proteins (MYPT1FL, MYPT1TR, MYPT1SO, and MYPT1TR2; Fig. 1, B and C; see supplemental Fig. 1 in the online version of this article). These proteins have an NH2-terminal truncation at aa 500 and either contain or lack the COOH-terminal LZ domain, the CC domain at aa 929–969, or the aa 888–928 sequence. We chose to focus on this specific CC domain (aa 929–969), because it was shown to have a higher probability of CC formation (16). Specifically, MYPT1FL contains the LZ domain, the CC domain, and the aa 888–929 sequence; MYPT1TR lacks all three domains; MYPT1SO contains the LZ domain and the CC domain but lacks the aa 888–928 sequence; and MYPT1TR2 lacks the LZ domain but contains the CC domain and the aa 888–928 sequence. The characteristics of these protein fragments are shown in Table 1.
To determine the epitopes recognized by the monoclonal antibodies MYPT1 1C2 and MYPT1 4G8, we performed immunoblots of the four MYPT1 fragments that represent the four naturally occurring MYPT1 isoforms. Specifically, alternative splicing of a central and a 3′ exon generates the presence or absence of a CI and the COOH-terminal LZ, respectively. Full details of cloning and expression have been previously described (13). MYPT1 1C2 recognizes all MYPT1 isoforms (Fig. 2A), whereas MYPT1 4G8 recognizes only LZ+ MYPT1 isoforms (Fig. 2C). To further characterize MYPT1 1C2 and MYPT1 4G8, we used MYPT1SO and MYPT1TR fragments. These Western blots show that MYPT1 1C2 binds to neither MYPT1SO nor MYPT1TR, demonstrating that the MYPT1 1C2 antibody recognizes an epitope within aa 888–928 (Fig. 2B), whereas MYPT1 4G8 recognizes MYPT1SO, but not MYPT1TR, showing that the MYPT1 4G8 antibody recognizes an epitope within the LZ domain (Fig. 2D). These results are similar to our previous results, which demonstrated that 1C2 recognizes all four naturally occurring MYPT1 isoforms (MYPT1 CI+/LZ+, MYPT1 CI+/LZ−, MYPT1 CI−/LZ+, and MYPT1 CI−/LZ−), whereas 4G8 recognizes only LZ+ MYPT1 isoforms and the protein fragments containing the LZ domain (14).
To complement the characterization of the MYPT1 fragments, PKGI isoform expression in the chicken gizzard lysates was also determined. Two-dimensional gel electrophoresis of total gizzard protein homogenate revealed five distinct PKGI spots, as confirmed by Western blotting with an antibody that could identify α- and β-isoforms of PKGI. Western blotting of duplicate two-dimensional separations showed that all five forms were variants of PKGIα, inasmuch as none were identified by a PKGIβ-specific antibody (Fig. 3). These data are similar to our previous results, where although we could detect the PKGIβ transcript and protein in cultured smooth muscle cells, the expression of PKGIβ protein could not be detected in embryonic or adult gizzard or aortic smooth muscle tissue (13).
To determine whether the MYPT1 protein fragment MYPT1FL could compete for binding of PKGIα with endogenous MYPT1, competition assays were performed with samples containing adult chicken gizzard tissue lysates (source of endogenous MYPT1) and various concentrations (0, 1.5, 3, and 8 μg) of MYPT1FL. Co-immunoprecipitation assays were performed with these samples using an anti-PKGIα/β antibody, and the MYPT1 proteins were detected by Western blotting (Fig. 4A) and analyzed by densitometry (Fig. 4B). Our results show that increasing the MYPT1FL concentration reduced the amount of endogenous MYPT1 associated with PKGIα and verified that MYPT1FL successfully competed with endogenous MYPT1 for an overlapping PKGIα binding site.
The four purified MYPT1 fragments were next tested for their ability to bind endogenous PKGIα (Fig. 5). Coimmunoprecipitation assays were performed with each of the four MYPT1 fragment proteins with use of an anti-PKGIα/β antibody to pull down endogenous PKGIα from adult chicken gizzard or aorta lysates and a polyclonal anti-MYPT1 antibody to visualize the proteins. The MYPT1 fragment proteins that contained the aa 888–928 sequence (MYPT1FL and MYPT1TR2; Fig. 5, A and B, respectively) were able to bind PKGIα, and, similar to MYPT1FL (Fig. 4), these protein fragments competed with endogenous MYPT1 for binding with PKGIα (see supplemental Fig. 2 in the online version of this article). PKGIα did not interact with proteins that lacked the aa 888–928 sequence (MYPT1SO and MYPT1TR; Fig. 5, C and D, respectively). Also, these results demonstrate that the LZ motif is not necessary for the interaction with PKGIα, since MYPT1TR2 bound, despite the absence of the LZ, and MYPT1SO did not bind, despite preservation of the LZ. Thus, although the LZ domain may be important in activation of the phosphatase (13), it is the aa 888–928 sequence that is necessary for binding of PKGIα to MYPT1.
These experiments were repeated with the inclusion of 8-bromo-cGMP, a nonhydrolyzable cGMP analog, to determine whether stimulation with this upstream effector would increase the binding of the MYPT1 protein fragments to PKGIα. As shown in Fig. 6 and in agreement with previous findings (13, 23, 24), the addition of 8-bromo-cGMP had no effect on the amount of MYPT1FL that was able to bind in the immunoprecipitation reaction. This suggests that the role of cGMP in NO-mediated vasodilatation may be to activate PKGIα, subsequent to the binding of PKGIα and MYPT1.
It has been reported that inhibitory peptides with amino acid sequences of RR, KR, and RK have high selectivity for PKGIα binding and inhibition, with the RK combination yielding the greatest selectivity (6, 7). Within the aa 888–928 sequence, MYPT1 contains an RK motif at aa 916 and 917. To determine whether these amino acids are involved in mediating the MYPT1-PKGIα interaction, we mutated these two amino acid residues to alanine (MYPT1A) or glutamic acid (MYPT1E). These mutations will not only determine whether R916K917 of MYPT1 is involved in PKGIα binding; they will also determine the contributions of size and charge of the amino acids at these positions to the interaction. To confirm that these mutations did not affect the structure of the MYPT1A and MYPT1E fragments, we performed immunoblots with two different antibodies: a polyclonal MYPT1 antibody and the 1C2 antibody, which is specific for an epitope within the aa 888–928 sequence (Fig. 2). The results show that whereas all four MYPT1 fragments are recognized by a polyclonal MYPT1 antiserum (Fig. 7A), only the proteins containing the aa 888–928 sequence are recognized by the MYPT1 1C2 antibody, and that mutation of the RK motif to AA or EE does not affect this recognition (Fig. 7B). Coimmunoprecipitation assays were performed as described previously, and our results showed that although less PKGIα was able to bind in chicken gizzard lysates than in aorta lysates, the MYPT1A mutations did not abolish the protein-protein interaction (Fig. 8A). However, mutation of the MYPT1 RK motif to an EE combination eliminated the interaction between PKGIα and MYPT1E (Fig. 8B), suggesting that the MYPT1 R916K917 plays a role in the binding of PKGIα and MYPT1 and that the interaction is dependent on charge, rather than size.
In our previous studies using cultured smooth muscle cells, PKGI and MYPT1 isoforms did not show binding by coimmunoprecipitation. This is in contrast to the results obtained from tissue lysates (Figs. 4 and 5), suggesting that an accessory protein may be involved in the interaction but is downregulated in cultured smooth muscle cells. Therefore, we rationalized that, by partial purification and enrichment of PKGIα from chicken gizzard, the putative accessory protein may be separated, leading to a loss of interaction between PKGIα and the purified MYPT1FL fragment. This protein fraction was enriched in PKGIα and had low levels of contaminants as demonstrated by SDS-PAGE and Western blotting (Fig. 9A). Using this fraction in the coimmunoprecipitation experiments, we could not detect a PKGIα-MYPT1FL interaction (Fig. 9B). This suggests that an accessory protein is necessary to mediate or stabilize the binding of PKGIα to MYPT1FL and that this accessory protein did not copurify with PKGIα.
The binding of PKGIα to MYPT1 and the subsequent activation of the myosin phosphatase are critical steps in NO-mediated vasodilatation. The present experiments demonstrate that PKGIα interacts with MYPT1 via a specific sequence at aa 888–928 of MYPT1, independent of the COOH-terminal CC or LZ domains, and that this binding may occur through charge-charge interactions of the MYPT1 RK motif located at aa 916 and 917.
However, others have suggested that the MYPT1 COOH-terminal LZ domain is necessary for the MYPT1-PKGIα interaction (23, 24). These investigators used short GST-fusion proteins containing a 59-amino acid fragment of PKGIα (24) or a COOH-terminal fragment of MYPT1 (23) to demonstrate an LZ-LZ interaction of MYPT1 and PKGIα. The differences between these previous results and our present results may be due to differences in size and/or structure between the GST-fusion peptides and our protein fragments (>450 aa). However, similar to the present study, we previously demonstrated that although all endogenous MYPT1 isoforms (LZ+ and LZ−) interact with PKGIα, the LZ domain was necessary for cGMP-mediated activation of the MLC phosphatase (13). These results suggest that the MYPT1 LZ is necessary for cGMP-mediated activation of MLC phosphatase but that the MYPT1 LZ is not important for binding of PKGIα.
PKGIα is known to activate its targets through phosphorylation, and cGMP stimulation of cultured smooth muscle cells expressing LZ+, but not LZ−, MYPT1 isoforms leads to a dose-dependent decline in MLC20 phosphorylation (13). Furthermore, the relative expression of LZ+ MYPT1 isoforms correlates with the sensitivity to cGMP-mediated smooth muscle relaxation (15). These observations may be due to a conformational change in LZ+ MYPT1 isoforms that exposes a phosphorylation site, which allows for MYPT1 phosphorylation by PKGIα and, subsequently, activation of MLC phosphatase. However, phosphorylation of MYPT1 is relatively slow (23) compared with smooth muscle cell relaxation (15). Additionally, others (26) demonstrated that PKGIα mediates an MYPT1 phosphorylation at Ser695 (human sequence). This phosphorylation does not directly increase phosphatase activity; rather, it prevents phosphorylation at Thr696, which blunts the effect of Rho kinase-mediated inhibition of MLC phosphatase activity (26). These data could suggest that PKGIα mediates activation of MLC phosphatase through an MYPT1 conformational change and that this conformational change is dependent on the presence of LZ+ MYPT1.
In this study, we have demonstrated that R916K917 of MYPT1 plays an important role in the interaction with PKGIα and that this interaction is dependent on the charge of the residues. Additionally, the PKGIα NH2-terminal CC domain contains an EE motif (aa 28/31). These sites are located on the exposed side of the CC domain and may undergo ionic interaction with the oppositely charged MYPT1 RK motif, possibly explaining why the positive charge at aa 916–917 of MYPT1 is necessary for the MYPT1-PKGIα interaction.
Homodimerization of PKGIα has been shown to occur at the NH2-terminal CC domain (17). Other studies have shown that MYPT1 also homodimerizes (23), possibly via the CC domain at aa 929–969. Thus the PKGIα-MYPT1 interaction may form a tetrameric structure, with both MYPT1 and PKGIα homodimerizing through their own CC domains and the subsequent interaction between these homodimers occurring through a charge-charge interaction of the RK motif of MYPT1 (aa 916–917) with the exposed EE motif of PKGIα at aa 28/31. Recently, Casteel et al. (2) demonstrated that dimerization of PKGIβ was not affected by mutation of two negatively charged residues within the NH2-terminal LZ. However, the mutation of these amino acids (D26 and E31) of PKGIβ to the positively charged residues KR within the aligned sequence of PKGIα abolished the interaction of PKGIβ with the general transcription regulator TFII-1 and the inositol 1,4,5-triphosphate receptor-associated PKG substrate. These data suggest that the specificity of the interaction of PKGIβ with TFII-1 and inositol 1,4,5-triphosphate receptor-associated PKG is determined by two oppositely charged amino acids in the LZ of the two isoforms of PKGI. Thus the specificity of PKGIα and MYPT1 binding could similarly be determined by charge differences between two amino acids in the LZ of PKGIα and PKGIβ; Q34/E37 in PKGIβ corresponds to E28/E31 in PKGIα, and the hydrophobic amino acid in PKGIβ could be responsible for the specificity of the PKGIα-MYPT1 interaction.
PKGI has multiple targets throughout the smooth muscle cell, including the L-type Ca2+ channel (8), maxi-K+ channels (1), sarcoplasmic reticulum channels (20), telokin (4, 25), and MYPT1 (13, 23, 24). This diversity in targets would suggest a need for tight regulation of PKGIα activity in the cell. A common mechanism for targeting kinases to their substrates is the presence of accessory binding proteins, and previous studies have shown that PKGIα can be localized to its substrates by a targeting protein (17). Interestingly, coimmunoprecipitation studies with endogenous MYPT1 and PKGIα have shown an increased level of PKGIα binding in aorta vs. gizzard tissue lysates (13), and we could not detect binding of MYPT1FL with PKGIα in the enriched protein fractions (Fig. 9). These data suggest that an accessory protein, which may have a decreased level of expression in the gizzard, could act to increase the stability of the PKGIα-MYPT1 interaction. However, we cannot rule out that an accessory protein acts to mediate an indirect MYPT1-PKGIα interaction, rather than to stabilize a direct interaction. Differential expression of this accessory protein could explain the difference in association of MYPT1A with PKGIα between gizzard and aorta (Fig. 8A), and downregulation of the accessory protein in cultured smooth muscle cells would explain why we were unable to detect an MYPT1-PKGIα interaction in our previous study (13).
In summary, our results show that, during NO-mediated smooth muscle cell relaxation, the PKGIα-MYPT1 interaction occurs through an MYPT1 aa 888–928 sequence, and the presence of positively charged residues at aa 916 and 917 is important for this interaction. The subsequent activation of MLC phosphatase, through phosphorylation or an MYPT1 conformational change, requires the presence of the MYPT1 COOH-terminal LZ domain. Thus the relative expression of LZ+ MYPT1 isoforms would regulate the sensitivity to cGMP-mediated vasodilatation and explain the diversity of tissue sensitivity to NO.
This study was supported by National Heart, Lung, and Blood Institute Grant HL-64137 (to F. V. Brozovich) and HL-078845 (to O. Ogut).
We thank Dr. Frank Soennichsen for help with the predictions of protein structure and Dr. J. P. Jin for his help with the production of the monoclonal antibodies.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society