M20, the small subunit of PP1M, binds to microtubules

Norio Takizawa, David J. Schmidt, Katsuhide Mabuchi, Emma Villa-Moruzzi, Richard A. Tuft, Mitsuo Ikebe


Myosin light chain phosphatase (PP1M) is composed of three subunits, i.e., M20, MBS, and a catalytic subunit. Whereas MBS is assigned as a myosin binding subunit, the function of M20 is unknown. In the present study, we found that M20 binds to microtubules. The binding activity was revealed by cosedimentation of M20 with microtubules and binding of tubulin to M20 affinity resin. Green fluorescent protein (GFP)-tagged M20 (M20-GFP) was expressed in chicken primary smooth muscle cells and COS-7 cells and was used as a probe for studying the association between M20 and microtubules in living cells. M20-GFP was localized on filamentous structures in both cell types. Colocalization analysis revealed that M20-GFP colocalized with tubulin. Treatment with nocodazole, but not cytochalasin B, abolished the filamentous structure of M20-GFP. These results indicate that M20-GFP associates with microtubules in cells. Microinjection of rhodamine-tubulin into the M20-expressing cells revealed that incorporation of rhodamine-tubulin into microtubules was significantly facilitated by microtubule-associated M20. Consistent with this result, M20 enhanced the rate of tubulin polymerization in vitro and produced elongated microtubules. These results suggest that M20 has a microtubule binding activity and plays a role in regulating microtubule dynamics.

  • M20
  • tubulin
  • myosin light chain phosphatase
  • protein phosphatase
  • green fluorescent protein

myosin light chain phosphatase (PP1M) is composed of three subunits when isolated from smooth muscle tissue, i.e., a 37-kDa catalytic subunit, a 110-kDa myosin binding subunit (MBS), and a 20-kDa small subunit (M20) (1, 27). The catalytic subunit is classified as a type 1 serine/threonine phosphatase (PP1) on the basis of its substrate specificity and its sensitivity to inhibitors (7) and has been identified as the δ-isoform with the use of isoform-specific antibody probes (30). It has been postulated that the regulation of serine/threonine protein phosphatases is governed by their noncatalytic subunits. This concept was originally proposed by Cohen and colleagues (7, 8) for the glycogen-binding subunit associated with type 1 phosphatase. For PP1M, it has been reported that the holoenzyme has a strong affinity for myosin and shows higher phosphatase activity than the isolated catalytic subunit (1, 27), indicating that the binding of the regulatory subunits increases enzyme activity.

Several splicing variants have been reported for MBS (6, 9,26), but the overall structural motifs are present in these variants. The NH2-terminal third of the molecule is composed of eight repeated sequences corresponding to the sequence motif for the “ankyrin repeat” that is found in proteins involved in tissue differentiation, cell cycle regulation, or regulation of the cytoskeleton (24, 29, 33). Shimizu et al. (26) suggested that the large subunit has myosin-binding affinity, because proteolyzed PP1M containing a 58-kDa fragment of the large subunit and the 37-kDa catalytic subunit (but not the 20-kDa small subunit) retained myosin binding activity, albeit with reduced affinity. Ichikawa et al. (17) showed that the recombinant NH2-terminal two-thirds of the large subunit contains a myosin binding site, because phosphorylated myosin HMM bound to an affinity column made with this recombinant fragment. On the other hand, Johnson et al. (18) reported that the COOH-terminal 291 residues of the large subunit, not the NH2-terminal fragment, bind to myosin. Thus, although the precise identity of the myosin-binding domain of PP1M remains controversial, it is generally agreed that the large subunit is the myosin-targeting subunit.

The domains involved in the interaction between the subunits of PP1M were described using two different approaches. Johnson et al. (18) expressed various fragments of PP1M subunits in anEscherichia coli expression system, and the interaction of the subunits was examined by using a gel overlay technique. The 21/20-kDa subunit did not interact with the catalytic subunit but interacted with the COOH-terminal 72 residues of the large subunit. Both large and small (21/20 kDa) subunits bound to myosin. On the other hand, Hirano et al. (15) used the yeast two-hybrid system to identify subunit interactions and concluded that the catalytic subunit binds to the large subunit at two sites, a relatively strong site in the NH2-terminal 38 residues and a weak site in the ankyrin repeat (residues 39–295). A binding site for phosphorylated light chain is also assigned to the ankyrin repeat. The small subunit binding site is identified as the COOH-terminal fragment of the large subunit.

MBS plays an important role in regulating PP1M, and it is known that Rho kinase phosphorylates MBS that inhibits the phosphatase activity (20). In smooth muscle, it has been known that the activation of G proteins increases Ca2+-independent myosin phosphorylation and that this is due to the inhibition of PP1M (28), and it has been postulated that Rho/Rho kinase-dependent phosphorylation of MBS plays a role in this process (16). On the other hand, nothing is known about the function of the small subunit of PP1M, M20.

In the present study, we found that M20 binds to microtubules in vitro. To examine the binding of M20 to mictotubules in cells, green fluorescent protein (GFP)-tagged M20 (M20-GFP) was expressed in cells and used as a probe for studying M20-microtubule binding in living cells. The results suggested that M20 can interact with microtubules and influences the incorporation of tubulin into microtubules.



All reagents were obtained from Calbiochem (San Diego, CA) and Sigma (St. Louis, MO). The plasmid vector pEGFP was purchased from Clontec (Palo Alto, CA), pet15b was from Novagen (Darmstadt, Germany), and pcDNA1amp was from Invitrogen (Carlsbad, CA).


The mouse monoclonal IgG1 anti-α-tubulin antibody (DM1A) was purchased from Sigma Immunochemical. The rabbit anti-GFP antibody was obtained from MBL (Nagoya, Japan). Secondary antibodies including horseradish peroxidase-, Texas red-, or Cy5-conjugated affinity-purified goat anti-rabbit IgG, anti-mouse IgG, or anti-mouse IgM antibodies were all obtained from Jackson ImmunoResearch Laboratories (West Grove, PA). A mouse anti-M20 monoclonal antibody was made using hexahistidine-tagged M20 (M20-His) expressed in E. coli as an antigen (Fig.1 A). The antibody production was performed as described previously (14). A rabbit anti-M20 antibody was made using the polypeptide CRSKEFTRNRKSQSDSP conjugated to keyhole limpet hemocyanin as an antigen (see Figs. 7 and10). The rabbit isoform-specific antibodies for PP1 have been previously described (31).

Fig. 1.

Interaction between (M20) and microtubules. A: SDS-PAGE of the purified proteins used for the experiment. Lane 1, hexahistidine-tagged M20 (M20-His); lane 2, microtubule proteins; lane 3, purified tubulin. α and β represent the α- and β-isoforms of tubulin. B: cosedimentation of M20 with microtubules. The association of M20 with microtubules was examined by a cosedimentation assay as described inmaterials and methods. S and P represent supernatant and pellets, respectively. Lane 1, microtubules + M20-His;lane 2, microtubules alone; lane 3, M20-His alone. Note that the apparent molecular mass of M20-His is larger than 20 kDa due to the NH2-terminal tag + linker sequence.C: SDS-PAGE analysis of an M20 affinity chromatography. Affinity columns were made as described in materials and methods. Flow-through fractions (lanes 1 and4), buffer-washing fractions (lanes 2 and5), and the elution fractions with 0.25 M imidazole (lanes 3 and 6) are shown from a control NiNTA column (lanes 1–3) or NiNTA column with bound M20-His (lanes 4–6). Tubulin is indicated by barbed arrowheads, and M20 is indicated by flat arrowhead.

Protein purification.

Production and purification of M20-His was according to Johnson et al. (19). Briefly, cDNA of M20 from chicken was cloned into pet15b. The construct was expressed in E. coli strain BL21(DE3)plysS. Cultures were grown at 37°C in Luria-Bertani medium in the presence of 100 μg/ml ampicillin to an optical density (wavelength 600 nm) of 0.6, and protein expression was induced with 0.2 mM isopropylthiogalactoside for 12 h at 30°C. The cells were harvested, and the packed cells were suspended with 20 ml ofbuffer A containing 50 mM Tris · HCl, pH 8.0, 0.1 M NaCl, 1 mM EDTA, 0.1% 2-mercaptoethanol (2-ME), 0.2 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mM benzamidine and then frozen at −80°C. After thawing, 1 mg/ml sodium deoxylcholate, 8 mM MgSO4, and 10 μg/ml DNase I were added. The extract was then incubated until it was no longer viscous. Next, 6 mM EDTA, 1 mM benzamidine, and 0.2 mM PMSF were added, and the extracts were centrifuged for 10 min at 10,000 g. The pellet was dissolved in buffer A containing 6 M urea and centrifuged for 10 min at 10,000 g. The supernatant was purified with DE52 (Pharmacia) followed by NiNTA agarose chromatography (Qiagen). Porcine brain microtubule proteins and tubulin were purified according to Williams and Lee (32).


The association of M20 with microtubules was examined by a cosedimentation assay. Microtubule proteins (0.80 mg/ml) were preincubated at 37°C for 20 min in an assay buffer containing 10 mM PIPES, pH 6.8, 80 mM imidazole-HCl, pH 6.8, 50 mM KCl, 4 mM MgCl2, 0.1 mM EGTA, 1.6 mM GTP, and 10 μg/ml taxol. The mixture was then incubated with or without M20 (0.13 mg/ml) for 20 min at 37°C. The mixture (60 μl) was centrifuged at 100,000g for 10 min at 37°C, and the resulting supernatant and pellet were analyzed by SDS-PAGE. The pellets were resuspended with the assay buffer so as to directly compare the fraction of protein in the pellets with that in the supernatant.

M20 affinity chromatography.

M20-His (40 μg) was loaded onto a NiNTA agarose column (Qiagen). After a wash with buffer B (10 mM phosphate buffer, pH 7.5, 60 mM NaCl, 2% glycerol, and 0.01% 2-ME), purified tubulin (200 μg) was applied to these columns. The columns were washed with 2 ml ofbuffer B with 50 mM imidazole and then eluted with 200 μl of buffer B containing 0.25 M imidazole. The proteins in each fraction were analyzed by SDS-PAGE.

Light scattering.

A PC1 spectrofluorometer (ISS, Champaign, IL) was used to monitor light scattering at a 90° angle. The incident wavelength and emission wavelength were set at 400 nm. The slit spectral widths were set to 1 nm for both beams. The solution was held in a high-UV transparent 2 × 10-mm rectangular cuvette. The polymerization of tubulin was monitored after either 0.20 mg/ml microtubule proteins or 0.16 mg/ml tubulin (250 μl, in 10 mM PIPES, pH 6.8, 50 mM KCl, 4 mM MgCl2, 0.1 mM EGTA, and 1.6 mM GTP) was incubated at 37°C. Various concentrations of M20 were added after 15 min when the light scattering showed a gradual linear increase.

Electron microscopy.

Purified tubulins (0.16 mg/ml) were preincubated in 10 mM PIPES, pH 6.8, 50 mM KCl, 4 mM MgCl2, 0.1 mM EGTA, 1.6 mM GTP, and 10 μg/ml taxol for 10 min at 37°C. The mixtures were incubated without or with M20 (48 μg/ml) for 20 min at 37°C. Microtubules were absorbed to mica-coated grids and negatively stained with 1% aqueous uranyl acetate. The samples were observed with an electron microscope (model EM300; Phillips Electronic Instruments, Mahwah, NJ) at 60 kV.

Construction of the expression vectors of M20.

An M20-GFP fusion construct was made as follows. A cDNA encoding EGFP (Clontec) flanked with BamH1 and EcoR1 at the 5′- and 3′-side, respectively, was subcloned into the pcDNA1amp mammalian expression vector at the polylinker region using theBamH1 and EcoR1 sites. A chicken M20 cDNA (6) that was flanked with BamH1 sites at both ends and had its stop codon eliminated was inserted into the vector at the BamH1 site. After the orientation was confirmed by sequencing, the construct was used as a GFP-tagged M20 expression vector. The construct contains a linker sequence of SGSLHACRSTLEDPRVPVAT between M20 and GFP to aid in segregating the two domains. To produce a yellow fluorescent protein (YFP)-tagged MBS construct, we created a Sal1 site at the initiation codon of MBS in a pBluescript KS vector. MBS cDNA was excised from the vector with Sal1 digestion and subcloned into pEYFPC1 at the polylinker region. A linker sequence of SGLRSRAQASNSAV was present between YFP and the MBS sequence. All vectors for mammalian cell transfection were prepared and purified by using the CsCl2 method.

Cell culture.

COS-7 cells were grown as monolayers in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS). Embryonic gizzard and aorta cells were prepared as described previously (12). In brief, fertilized White Leghorn chicken eggs were incubated for 10–15 days at 38°C in humidified air. Embryos fromdays 13–15 were removed from the eggs and decapitated, after which the gizzards and thoracic aorta were removed and placed on ice in separate tubes containing growth medium (DMEM-Ham's F-12 1:1 mixed with 10% FCS). The cells that were used for the studies were resuspended in DMEM-Ham's F-12 with 0.5% FCS. All cells were maintained in a humid incubator at 5% CO2 and 37°C.


The cells were transfected by electroporation using a Gene pulser II (Bio-Rad, Hercules CA). COS-7 cells (2 × 105) were suspended in 250 μl of ice-cold PBS containing 10 mM HEPES, pH 7.2, and transferred into a 2-mm cuvette. The cells were subjected to electroporation at 200 V, 950 μF with 2–10 μg of DNA. The cells were incubated on ice for 5 min and then suspended in culture medium. Gizzard cells (5 × 105) were suspended in 200 μl of ice-cold buffer containing 27 mM sodium phosphate, pH 7.5, and 150 mM sucrose. The cells were transferred into a 2-mm cuvette and electroporated with 2–10 μg of DNA at 180 V, 100% modulation, 40-kHz radiofrequency, 3-ms burst duration for 10 bursts with a 1-s burst interval. After the transfection, ice-cold culture medium was added immediately to the cells. The cells were cultured in a CO2 incubator with 5% CO2 at 37°C. A liposome-mediated gene transfer method using Lipofectamine (Bio-Rad, Hercules CA) yielded similar results.

Western blotting.

The cells were treated with 10% trichloroacetic acid and centrifuged at 10,000 g for 10 min at 4°C. The pellets were dissolved with 0.5 M NaHCO3 and Tris-base to adjust pH. The samples were subjected to 7.5–20% gradient SDS-PAGE using the method of Laemmli (21). Proteins were transferred to a polyvinylidene difluoride membrane that was then treated with 5% skim milk for blocking. After washing with PBS containing 0.1% Triton X-100 and incubation for 1 h with primary antibodies at room temperature, blots were incubated for 1 h with horseradish peroxidase-conjugated secondary antibodies. Immunoreactive bands were detected with enhanced chemiluminescence (ECL; Amersham, Arlington Heights, IL).


The cells were fixed with 2% formaldehyde in PBS containing 2 mM EGTA and 2 mM MgCl2 for 10 min and then permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature. After blocking the coverslips with 1% BSA-PBS at room temperature for 1 h, the samples were incubated with primary antibodies at 37°C for 1 h. After being washed with PBS three times, the samples were incubated with fluorescence-labeled secondary antibodies at 37°C for 1 h. After excess antibody was washed off with PBS, the specimens were mounted in 3% DABCO mounting medium (90% glycerol-10% 10× PBS). Microtubules were labeled with murine monoclonal anti-α-tubulin antibody (Sigma Imunochemical). F-actin was labeled with Texas red-phalloidin (Molecular Probes).

Three-dimensional microscopy.

Images of labeled cells were acquired with an inverted Nikon Diaphot microscope equipped with a 100-W mercury arc lamp for epifluorescence microscopy. Cells were viewed with a ×40, ×60, or ×100 Nikon Planapo objective (NA 1.3 or 1.4) and ×2.5 or ×5 eyepiece, and images were captured by a Photometrics liquid-cooled charge-coupled device camera (SITE 512 chip). A personal computer was used to control a Pifoch piezo driver attached to the fine focus (z). Additionally, a controlled shutter regulated the duration of illumination (11). Digitized images of labeled cells were obtained at 0.25 μm z-intervals. This through-focus image series was transferred to Silicon Graphics workstations for analysis. In addition, through-focus sets of images of 190-nm-diameter fluorescent beads were acquired. Beads with fluorescence spectra similar to the fluorophores used to label cells were imaged under optical conditions identical to those under which images from labeled cells were acquired to provide an empirical measure of the microscope point spread function.

Image restoration and analysis.

To prepare images for analysis, the dark current and background were subtracted from the images. The result was then flat-field corrected for nonuniformities in illumination intensity and camera sensitivity across the field of view. Three-dimensional data sets were processed by using a constrained deconvolution algorithm based on regularization theory (4, 5, 11) to reverse blurring introduced by the microscope optics. Image pairs were aligned for maximized overlap. To avoid errors in interpreting the data introduced by superimposition of the front and back surfaces, the image of the cell was sectioned in thex-y plane, and only one section of the cell is displayed. In the figures presented, GFP or YFP chimeras are pseudocolored green; blue fluorescent protein (BFP) chimeras are pseudocolored blue, and Texas red- or Cy5-labeled proteins are pseudocolored red. In the superimposed images (2 or 3), voxels containing signals for both or all three proteins, respectively, were pseudocolored white. Because the relative stoichiometries of the two proteins are not a simple function of the ratio of intensities of the two fluorophores, the white intensity information in these images is not a representation of the true stoichiometries but solely symbolizes colocalization. Signal in a cell for a given protein at a specific voxel in an initial image was considered colocalized with signal in a corresponding image if it had the same x, y, andz coordinates. One voxel in the initial image was allowed to colocalize with only one voxel in the corresponding image.

Microinjection of rhodamine-labeled tubulin into COS-7 cells.

Rhodamine-labeled tubulin (10 mg/ml) was purchased from Cytoskeleton (Denver, CO). Microinjection of rhodamine-tubulin into COS-7 cells was performed with an Eppendorf automatic microinjection system (200 psi, 1.0 s). After microinjection, the cells were incubated for 30 min with culture medium at 37°C and then incubated with 10 μM taxol in culture medium for 30 min at 37°C. The cells were then fixed and permeabilized with 100% methanol for 10 min at −20°C.


Binding of M20 to tubulin in vitro.

Binding of M20 to microtubules was demonstrated by two methods. First, M20 was mixed with polymerized microtubule proteins (Fig.1 A) and binding activity was examined by cosedimentation analysis. As shown in Fig. 1 B, M20 cosedimented completely with microtubules. To see whether M20 binds directly to tubulin or microtubule-associated proteins, the binding between M20 and isolated tubulin was tested using M20 affinity chromatography. M20 containing a hexahistidine tag was first applied to a NiNTA agarose resin to produce tight binding of M20 to the resin. Isolated tubulin was then loaded onto the resin. Excess tubulin was recovered in the flow-through fraction, and bound tubulin was eluted from the column by 0.25 M imidazole, which competes with histidine binding to Ni2+and dissociates the M20-His from the resin (Fig. 1 C). Approximately 30 μg of tubulin were eluted from a column containing 50 μg of M20. On the other hand, very little tubulin was bound to an MBS affinity column (not shown). There is virtually no nonspecific binding of tubulin to the resin, because without initial loading of M20-His tubulin did not bind to the column (Fig. 1). The binding between the proteins is often weakened by high ionic strength due to the elimination of the ionic interaction. We also applied NaCl to the affinity column to elute the bound tubulin. The bound tubulin was eluted at 0.3 M NaCl (not shown). These results demonstrate the direct binding of M20 to tubulin in physiological ionic conditions in vitro. The extent of binding of M20 to microtubule was determined by cosedimentation assay as a function of M20 concentration, and the maximum binding of ∼2 moles of M20 per mole of tubulin dimer was obtained (not shown). This is probably due to the formation of M20 dimer in vitro as reported previously (22).

To examine the effect of M20 binding on microtubule stabilization, we added M20 to purified microtubule or tubulin and monitored microtubule polymerization/oligomerization by measuring increases in light-scattering intensity (Fig. 2). Tubulin dimers initiated polymerization slowly after addition of GTP (Fig.2 A, inset), but the light-scattering intensity of microtubules markedly increased upon addition of M20. This effect was dose dependent, and an M20-dependent fast component of the measured light scattering increased almost linearly up to 15 μg/ml M20. Extrapolation of the line to the maximum change in light scattering suggests that the effect is saturated at 0.67 mole of M20 per 1 mole of tubulin dimer. The results also suggest that the effect is due to direct binding of M20 to tubulin but not to microtubule-associated proteins. To examine whether or not the observed increase in light scattering is due to the formation of microtubule bundles, we observed microtubules before and after addition of M20 by negative staining using electron microscopy. The microtubules are significantly elongated in the presence of M20 compared with the control (Fig. 3). In the absence of M20 there are significantly more background particles showing oval shape that represent the unpolymerized tubulin dimer. There are some microtubule bundles observed, i.e., the lateral association of microtubules, but no significant change in the microtubule bundle formation was observed in the presence and absence of M20 (not shown). These results suggest that M20 facilitates the polymerization of microtubules.

Fig. 2.

Enhancement of light scattering (LS) and microtubule formation by M20. The light scattering during microtubule formation was measured as described in materials and methods.A: effect of M20 on the light scattering of microtubules formed from isolated microtubule proteins. Addition of M20 is indicated by an arrowhead. Inset: expanded scale of line 7. C: effect of M20 on the light scattering of microtubules formed from purified tubulin. B and D show the M20 concentration dependence of the change in light scattering (ΔLS) in A and C, respectively. Lines 1–10represent M20 concentration: 1, 72 μg/ml;2, 36 μg/ml; 3, 14 μg/ml; 4, 7.2 μg/ml; 5, 3.6 μg/ml; 6, 2.4 μg/ml;7, 1.4 μg/ml; 8, 0.72 μg/ml; 9, 0.36 μg/ml; 10, absence.

Fig. 3.

Electron micrographs of microtubules in the presence and absence of M20. Purified tubulins were incubated without (A) or with (B) M20 for 20 min at 37°C. Microtubules were absorbed to mica-coated grids and negatively stained with 1% aqueous uranyl acetate. Bar, 800 nm.

Association of M20 with microtubules in cells.

To examine whether or not M20 could bind to microtubules in cells, we produced a GFP-tagged M20 subunit (M20-GFP) of PP1M as described inmaterials and methods (also see Fig.4) to aid in determining intracellular localization. COS-7 cells were transfected with the expression vectors, and cell extracts were subjected to Western blot analysis to confirm the expression of GFP fusion proteins. As shown in Fig. 4, use of anti-GFP antibodies detected a single band in the extract of the transfected cells. The apparent molecular mass detected was 47 kDa for M20-GFP-transfected cells (Fig. 4 B). The apparent molecular mass was consistent with the sum of the GFP mass and M20, indicating that M20-GFP was successfully expressed in cells and that no degradation of the GFP-tagged protein occurred. This assured that the GFP signal observed in cells is derived from the GFP-tagged M20 construct but not the free GFP or other degradation products containing GFP moiety. To examine the authenticity of function of the GFP-tagged M20, the MBS binding capability of M20-GFP was examined by coimmunoprecipitation method. Purified MBS was added to the cell lysate of M20-GFP-transfected COS-7 cells or GFP-transfected cells, and then the fractions were immunoprecipitated with anti-GFP antibodies. The precipitates were subjected to Western blot analysis using anti-MBS antibodies (Babco). MBS was coimmunoprecipitated with M20-GFP but not GFP (Fig. 4 C). These results suggest that M20-GFP has MBS binding activity, as is known for isolated natural M20 (1,27). To examine whether GFP-M20 can bind to microtubules, we subjected the extracts of M20-GFP-transfected cells or GFP-transfected cells to the microtubule cosedimentation analysis. M20-GFP but not GFP cosedimented with microtubules, indicating that M20-GFP also retains the microtubule binding activity (Fig. 4 D). The binding of M20-GFP to microtubules was not influenced by the presence of MAPs (not shown).

Fig. 4.

Expression of green fluorescence protein (GFP)-tagged M20. A: construction of the expression vectors. Enhanced GFP cDNA was linked toward the 3′-end of the M20 coding sequence. A stop codon in the M20 sequence was deleted and in-frame ligated to the GFP cDNA. This fusion cDNA was cloned into the pcDNA1amp mammalian expression vector. B: Western blots of GFP-tagged M20. COS-7 cells were transfected with M20-GFP, and the total cell extracts were subjected to Western blot analysis using anti-GFP antibodies as probes. Lane 1, nontransfected cells;lane 2, M20-GFP-transfected cells; lane 3, GFP-transfected cells. C: binding of GFP-tagged M20 to myosin binding subunit (MBS). The transfected cells were extracted, extracts were incubated with 10 μg/ml MBS, and the mixtures were immunoprecipitated with anti-GFP antibodies. Top: Western blots of total cell lysates with anti-GFP antibodies.Bottom: Western blot of immunoprecipitated fractions with anti-MBS antibodies. Lane 1, M20-GFP-transfected cells;lane 2, GFP-transfected cells. D: cosedimentation of M20-GFP with microtubules. The transfected cells were lysed with 0.15 ml of extraction buffer by passage through a narrow-gauge needle. The extracts were centrifuged at 800 g for 30 s. The supernatants were centrifuged at 100,000 g for 10 min at 4°C. The resulting supernatants (6 μl) were used for cosedimentation assay with the purified microtubules as described inmaterials and methods. Top: Western blots of anti-GFP antibody (Ab) staining. Bottom: Coomassie brilliant blue (CBB) stain of tubulin. Lane 1, supernatant of GFP-transfected cell extract; lane 2, pellet of GFP-transfected cell extract; lane 3, supernatant of M20-GFP-transfected cell extract; lane 4, pellet of M20-GFP-transfected cell extract.

Association of M20 with microtubules was also determined by examining the localization of M20-GFP in the cells. M20-GFP, which was expressed in primary cultures of live smooth muscle cells, was visualized under the digital fluorescence microscope (Fig.5). Both chicken embryonic aorta and gizzard cells showed a filamentous localization of M20-GFP. This filamentous localization of M20 was confirmed with anti-M20 antibody staining (Fig. 5 C). Anti-M20 antibodies preabsorbed by the M20 peptide antigen did not show any antibody staining showing the specificity of the antibodies against M20. To identify the filamentous structure to which M20-GFP localizes, we performed colocalization analysis between M20-GFP and microtubules or F-actin (Fig. 6). The fluorescence signal of M20-GFP expressed in COS-7 cells showed a high degree of colocalization with the Texas red signal of anti-α-tubulin staining. On the other hand, the GFP signal did not show significant colocalization with Texas red-phalloidin bound to F-actin. Similar results were obtained with primary cultured chicken embryonic aorta cells and gizzard cells (not shown). The filamentous localization of the endogenous M20 was also observed by immunostaining (Fig.7 A). Consistent with the localization of the overexpressed GFP-M20, the endogenous M20 localization coincided with the α-tubulin localization (Fig. 7). Because of the relatively low signal intensity of M20, not all of the microtubule structure was superimposed with the M20 signal. To further confirm the identity of the filamentous structure involved in M20-GFP localization, COS-7 cells expressing M20-GFP were treated with nocodazole or cytochalasin B. The filamentous localization of M20-GFP was completely abolished with nocodazole treatment but not with cytochalasin B treatment, indicating that the structure with which M20-GFP localizes is composed of microtubules but not microfilaments (Fig. 8).

Fig. 5.

Filamentous localization of M20-GFP in smooth muscle cells. Embryonic chicken aorta cells (A) or embryonic chicken gizzard cells (B and C) expressing M20-GFP were observed under the digital fluorescence microscope. A: a living aorta cell at 48 h after transfection was observed with excitation at 488 nm and emission at 520 nm (GFP fluorescence in fluorescein filter set). B and C: a gizzard cell at 48 h after transfection was fixed and stained with anti-M20 antibody (red) followed by Texas red-labeled anti-mouse second antibody (C); B is GFP image (green). Bar, 5 μm.

Fig. 6.

Colocalization of M20-GFP with microtubules. COS-7 cells transfected with the M20-GFP expression vector were fixed and visualized. A and D: GFP images (green).B: cells stained with anti-α-tubulin antibodies followed by Texas red-labeled anti-mouse second antibodies (red). E: cells stained with Texas red-phalloidin (red). Colocalization analyses are shown in C for A and B and inF for D and E. White indicates colocalization. Bar, 5 μm.

Fig. 7.

Filamentous localization of endogenous M20 in smooth muscle cells. Embryonic chicken gizzard cells were coimmunostained with rabbit anti M20 antibodies and mouse anti-α-tubulin antibodies. Localization of M20 and tubulin was visualized with Alexa Fluor 594-conjugated goat anti-rabbit and Alexa Fluor 488-conjugated goat anti-mouse secondary antibodies, respectively. A: M20 localization (green). B: tubulin localization (red).C: colocalization analyses for A andB. Bar, 5 μm. D: lanes 1 and 2are Western blots of endogenous M20 of gizzard cell extracts obtained using anti-M20 antibodies and the antibodies preabsorbed with 10 μM peptide antigen, respectively.

Fig. 8.

Effects of nocodazole and cytochalasin B on the localization of M20-GFP. Nocodazole, but not cytochalasin B, diminished the filamentous localization of M20-GFP. COS-7 cells expressing M20-GFP were treated with 10 μg/ml nocodazole (A andB) or 1 μM cytochalasin B (C and D). The cells were observed at room temperature under the digital fluorescence microscope before (A and C) and 60 min after (B and D) addition of the reagents. The filamentous localization of M20-GFP was maintained at room temperature for at least 3 h in the absence of these reagents. Bar, 5 μm.

Effect of M20 on microtubules in cells.

As shown in Fig. 9, cells displaying association between microtubule and M20-GFP often show elongated and curved microtubular structures that are clearly distinguishable from microtubules in nontransfected cells (Fig. 9) or control GFP-expressing cells (not shown). The same results were obtained for cells expressing non-GFP-tagged M20. As shown in Fig.10, immunostaining of cells expressing M20 with both anti-M20 and anti-tubulin revealed that M20 localized with microtubules regardless of its expression level (Fig. 10, Aand D). However, the cells expressing a high level of M20 (Fig. 10 A) showed more elongated and denser microtubule structure as was seen in GFP-M20-expressing cells, in contrast to the cells expressing a low level of M20 (Fig. 10 D). Also, whereas the microtubules in the control cells were not well maintained after weak fixation (2% formaldehyde for 10 min), in contrast, the filamentous structure of the microtubules was clearly retained in cells expressing M20-GFP (not shown). These results suggest that the binding of M20 to microtubule may affect microtubule dynamics or stability. To examine the role of M20 in microtubule dynamics in cells, we microinjected rhodamine-labeled tubulin and monitored its incorporation into microtubules under the digital fluorescence microscope. Figure11 shows representative images of such an experiment. In M20-GFP-expressing cells, rhodamine-tubulin was incorporated into microtubules in filamentous structures that coincide with GFP fluorescence derived from M20-GFP (Fig. 11). On the other hand, non-M20-GFP-expressing control cells showed a rather diffuse localization of rhodamine tubulin. Some M20-GFP-expressing cells did not show a filamentous localization of M20-GFP. In such cells, the injected rhodamine-tubulin also showed diffuse localization (Fig.11). These results demonstrate that M20 that was localized on microtubules facilitates the incorporation of tubulin into microtubules.

Fig. 9.

Effect of M20-GFP expression on microtubules in cells. COS-7 cells transfected with M20-GFP expression vector were fixed with 3% formaldehyde 48 h after transfection. The cells were stained with an anti-α-tubulin antibody, and microtubule structure was visualized with Texas red-conjugated secondary antibodies. A: M20-GFP image (green). B: α-tubulin distribution (red). Bar, 10 μm. The M20-GFP-expressing cell shows significantly more intense and elongated microtubules.

Fig. 10.

Effects of non-tagged M20 on microtubule structure. COS-7 cells transfected with M20 expression vector were fixed as described in Fig. 6. The cells were coimmunostained with a rabbit anti-M20 antibody and a mouse anti-α-tubulin antibody. Localization of M20 and tubulin was visualized with FITC-conjugated anti-rabbit and Texas red-conjugated anti-mouse secondary antibodies, respectively.A and D: FITC images of M20 distribution (green). The intensity of FITC in A is more than 10 times higher than that in D. B and E: Texas red image of α-tubulin distribution (red). Bar, 10 μm. The M20-expressing cell shows significantly more intense and elongated microtubules, similar to the M20-GFP-expressing cell. C: colocalization analyses forA and B. F: colocalization analyses for D and E. G: Western blot of the total cell extracts of COS-7 cells. Lane 1, nontransfected cells; lane 2, M20-transfected cells; lane 3, M20-transfected cells using preabsorbed antibodies with 10 μM peptide antigen. Anti-M20 antibodies detected a single band in the extract.

Fig. 11.

Microinjection of rhodamine-tubulin into COS-7 cells expressing M20-GFP. Rhodamine-tubulin (10 mg/ml) was microinjected into cells 48 h after transfection. After injection, the cells were incubated with 10 μg/ml taxol for 30 min and fixed with methanol.A: M20-GFP (green). B: rhodamine-tubulin (red).C: colocalization analyses for A andB. Bar, 10 μm.

M20 induces the localization of other PP1M subunits on microtubules.

Because M20 is a subunit of PP1M, another important question is whether the other subunits of PP1M, i.e., MBS and PP1δ, can also localize at microtubules. To address this question, cells were cotransfected with M20-BFP and MBS-YFP expression vectors. The cotransfected cells were fixed and stained with anti-PP1δ antibodies followed by Cy5-conjugated secondary antibodies. As shown in Fig.12, all BFP, YFP, and Cy5 images showed filamentous structures, which largely colocalize with each other when the images are superimposed. There was no obvious filamentous localization of PP1δ in control cells, as was revealed by anti-PP1δ antibody staining (not shown). These results suggest that M20 plays a role as a docking protein to allow the PP1M holoenzyme to localize at microtubules.

Fig. 12.

Filamentous localization of MBS and PP1δ in GFP-M20-expressing cell. COS-7 cells were cotransfected with M20-BFP (A, blue) and MBS-YFP (B, green). The cells were stained with anti-PP1δ antibodies followed by Cy5-conjugated secondary antibodies (C, red). D: colocalization analyses for A, B, and C. White color in D indicates the colocalization among these images. Bar, 10 μm. E: Western blots of the transfected cell lysates.Lane 1, M20-BFP-transfected cell lysate stained with anti-GFP antibodies; lane 2, MBS-YFP-transfected cell lysate stained with anti-GFP antibodies; lane 3, nontransfected cell lysate stained with anti-PP1δ antibodies.


M20 is a subunit of PP1M, which also contains two further subunits, MBS and a catalytic subunit (PP1δ) (1, 27). Whereas the role of MBS has been identified as a myosin binding subunit, the function of M20 is not known. Although it was reported (34) that the addition of exogenous M20 to the permeabilized arterial smooth muscle strip increases the Ca2+ sensitivity of contraction, not much is known for the underlying mechanism. The present study revealed that M20 has tubulin binding activity and can associate with microtubules in cells.

We produced GFP-tagged M20 to employ it as an in vivo probe to study the function of M20. Although it is difficult to assess this issue because the function of M20 has not been identified, it seems that GFP-tagged M20 retains the authentic properties of untagged M20 because1) M20-GFP can bind to MBS, 2) M20-GFP binds to microtubules as non-tagged M20 does, and 3) the expressed M20-BFP recruited both MBS and PP1δ into filamentous structures (Fig.12). Because it is known from in vitro experiments that M20 binds MBS, which in turn binds the catalytic subunit, thus producing the holoenzyme of PP1M, this result also indicates that GFP-tagged M20 retains its ability to support formation of holoenzyme in vivo. Thus M20-GFP likely retains proper function as a subunit of PP1M.

GFP-tagged M20 localized to a filamentous structure in living smooth muscle cells as well as in COS-7 cells. This filamentous structure was identified to be microtubules because 1) colocalization analysis revealed that M20-GFP as well as non-tagged M20 highly colocalized with α-tubulin but not with Texas red-phalloidin,2) nocodazole but not cytochalasin B treatment diminished the filamentous localization of M20-GFP, 3) rhodamine-tubulin injected into cells colocalized with filaments containing M20-GFP, 4) M20 cosedimented with microtubules, and 5) isolated tubulin bound to an M20 affinity resin.

PP1M was originally described as a phosphatase that dephosphorylates myosin regulatory light chain at serine-19 and has an affinity for myosin via its myosin binding subunit (MBS) in vitro. Therefore, one would expect that PP1M associates with myosin filaments in cells. However, when using anti-MBS antibodies as probes, it was found that the majority of PP1M is diffuse and cytosolic in cultured arterial smooth muscle cells (25), although some MBS was found on filaments. In the present study, we did not find localization of M20 on myosin filaments in the majority of cells, although some cells did show minor stress fiber localization of M20-GFP (not shown). Because no significant binding activity of isolated M20 to actin and/or myosin filaments was detected (not shown), the observed localization of M20-GFP on stress fibers is likely due to its binding to MBS, which is known to associate with myosin.

Interestingly, the association of M20 seems to stabilize microtubules. This was first suggested by the observation that cells expressing M20-GFP associated with microtubules showed highly elongated and curved microtubule structures. Furthermore, we found that the incorporation of rhodamine-tubulin into microtubules was significantly increased by the presence of M20 on microtubules. It seems that the localization of M20 on microtubules, and not just the expression of M20 itself, is critical to facilitate polymerization of the injected rhodamine-tubulin into microtubules, because cells expressing M20-GFP that was not localized on microtubules failed to show an enhanced incorporation of the injected tubulin into microtubule filaments. There are two possibilities to account for this effect. One is that the binding of M20 stabilizes the microtubule itself, and the other is that the binding of M20 changes the function of other microtubule binding proteins to stabilize the structure of microtubules. Because the addition of M20 induces the polymerization of tubulin in vitro (Fig.2), the observed microtubule elongation in cells expressing M20 is at least partly due to a direct effect of M20 binding to tubulin.

It is also plausible that the binding of M20 recruits other PP1M subunits to microtubules, thus inducing dephosphorylation of microtubule-associated proteins. As a matter of fact, cells expressing M20-BFP on microtubules also showed colocalization of MBS and PP1δ at the microtubules (Fig. 12). It was shown previously that the type 1 and type 2A serine/threonine protein phosphatase inhibitors okadaic acid (OA) or calyculin-A (CL-A) decrease the stability of microtubules (13). A dose-response analysis of effects of OA and CL-A suggested that PP1 but not PP2A is involved in maintaining the stability of microtubules. It is known that microtubule-associated proteins play a role in regulating the stability of microtubules (10) and that they are phosphorylated in cells, and it has been postulated that microtubule-associated protein functions can be regulated by phosphorylation (3). Recently, it was reported (23) that a PP1 phosphatase termed “microtubule-associated PP1” can bind to and dephosphorylate the microtubule-associated protein tau. The localization of an isotype of PP1c, PP1γ1, to microtubules in mitotic cells was also reported (2). Together, these previous findings and the present results raise a possibility that M20 may function as a microtubule-targeting subunit and that M20 recruitment of the phosphatase holoenzyme to microtubules may be involved in regulating microtubule dynamics. Further studies are required to understand the function of M20 in microtubule dynamics.


We thank Dr. P. Cohen (University of Dundee) for generous gifts of M20 and rat1 MBS cDNA.


  • This work was supported by National Heart, Lung, and Blood Institute Grants HL-60831 and HL-61426.

  • Address for reprint requests and other correspondence: M. Ikebe, Dept. of Physiology, Univ. of Massachusetts Medical School, 55 Lake Ave. North, Worcester, MA 01655 (E-mail:mitsuo.ikebe{at}umassmed.edu).

  • 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.

  • First published September 18, 2002;10.1152/ajpcell.00153.2002


View Abstract