Cell Cycle Regulation of Myosin-V by Calcium/Calmodulin-Dependent Protein Kinase II

See allHide authors and affiliations

Science  17 Aug 2001:
Vol. 293, Issue 5533, pp. 1317-1320
DOI: 10.1126/science.1061086


Organelle transport by myosin-V is down-regulated during mitosis, presumably by myosin-V phosphorylation. We used mass spectrometry phosphopeptide mapping to show that the tail of myosin-V was phosphorylated in mitotic Xenopus egg extract on a single serine residue localized in the carboxyl-terminal organelle-binding domain. Phosphorylation resulted in the release of the motor from the organelle. The phosphorylation site matched the consensus sequence of calcium/calmodulin–dependent protein kinase II (CaMKII), and inhibitors of CaMKII prevented myosin-V release. The modulation of cargo binding by phosphorylation is likely to represent a general mechanism regulating organelle transport by myosin-V.

The partitioning and transport of organelles in eukaryotic cells is achieved by the regulated action of motor proteins that bind to organelles and transport them along microtubules and actin filaments to ensure proper localization in the cell. During mitosis, movement of cytoplasmic organelles usually ceases, preventing the interference of organelles with components of the cell-division machinery and ensuring stochastic partitioning of multicopy organelles between daughter cells (1).Xenopus melanophores provide a convenient system to study the regulation of organelle transport. The concerted efforts of two microtubule motors, kinesin-II and cytoplasmic dynein, and an actin motor, myosin-V, achieve movement of pigment organelles (melanosomes) in melanophores (2–4). Because melanosomes do not respond to hormonal signals during mitosis, activity of these motors may be down-regulated during cell division (5). Whereas the regulation of microtubule motors in the cell cycle has been documented in other cell types (6,7), the melanophore system can be used for analysis of the regulation of myosin-based organelle transport. Myosin-V–driven transport of melanosomes is inhibited by their incubation in mitotic, but not interphase, Xenopus egg extract because of the dissociation of the motor from the surface of melanosomes (8). Myosin-V could be regulated in the cell cycle through the modification of its COOH-terminal tail domain, a part of the motor that is involved in cargo binding (9, 10).

To test whether the COOH-terminal globular domain of myosin-V tail (MGT) was sufficient for melanosome binding, we transfected melanophores with a construct encoding myc epitope–tagged MGT (amino acids 1443 to 1853 of mouse dilute myosin-Va) in the plasmid pcDNA3 (11). Melanosomes from transfected cultures were purified and probed for bound myc-MGT. Myc-MGT indeed bound to melanosomes (Fig. 1A). To determine whether the binding of myc-MGT was cell cycle–regulated in the same way as the binding of the endogenous motor, we treated melanosomes from transfected cultures withXenopus egg extracts as previously described (8). Both endogenous, full-length myosin-V and mouse myc-MGT were released from melanosomes after treatment with mitotic but not with interphase extract (Fig. 1, A and B). Thus, myosin-V release was determined by the modification of its COOH-terminal globular domain and did not require the presence of other parts of the molecule. In addition, mouse MGT accurately reproduced the behavior of endogenous, full-length myosin-V from Xenopus and could be used to study myosin-V regulation.

Figure 1

Mouse MGT binds to Xenopusmelanosomes and dissociates in mitotic egg extract. (A) MGT is dissociated from melanosomes in mitotic extract. Western blotting was done with the antibody to myc 1-9E10.2. MGT bound to melanosomes (C) and was retained after treatment with interphase egg extract (I) but was released after treatment with mitotic extract (M). (B) Myosin-V is dissociated from melanosomes in mitotic extract. Western blotting was done with polyclonal antibody DIL-2 against myosin-V (8). Melanosomes (C), melanosomes treated with mitotic egg extract (M), and melanosomes treated with interphase egg extract (I) are shown.

The release of myosin-V from melanosomes by mitotic extract correlates with its phosphorylation (8); therefore, MGT should contain the phosphorylation site. To test this hypothesis, we expressed MGT in Escherichia colias a fusion protein with glutathione S-transferase (GST). The recombinant protein was bound to glutathione-agarose beads (12); beads with fusion protein were incubated in extracts supplemented with [γ-32P]adenosine 5′-triphosphate (ATP) (13). After incubation, MGT was cleaved from GST with thrombin and analyzed by SDS–polyacrylamide gel electrophoresis (PAGE) and autoradiography. MGT was heavily phosphorylated in mitotic but not interphase extracts (Fig. 2A). Thus, the phosphorylation of myosin-V tail likely induced its dissociation from organelles.

Figure 2

Mitotic phosphorylation of Ser1650 of myosin-V and phosphopeptide analysis. (A) Phosphorylation of recombinant MGT inXenopus egg extracts. MGT purified on glutathione agarose beads (12) was phosphorylated in mitotic extract (M) but not in interphase egg extract (I). (B) Phosphopeptide-selective liquid chromatography electrospray–MS analysis (14) of MGT treated with mitotic (top) or interphase (bottom) egg extract (31). SDS-PAGE bands were excised and digested in situ with trypsin. Tryptic digests of each sample were fractionated by reverse phase HPLC. The column flow was split 10:1, with 3.6 μl/min being collected as fractions and 0.4 μl/min sent to the mass spectrometer, which was operated in the negative ion mode and optimized to detect the phosphate-specific marker ion, m/z 79 (PO3 ), produced by collision-induced dissociation (CID) in the ion source of the mass spectrometer. Fractions 7 to 9 were analyzed by MALDI-PSD and were found to each contain a phosphopeptide of mass 1920.9. Under PSD conditions, phosphopeptides lose H3PO4 from phospho-serine and -threonine side chains and HPO3 from phosphotyrosine side chains to produce intense–98 daltons and [M+H] –80 daltons fragment ions. Thus, phosphopeptides can be easily distinguished from nonphosphorylated peptides, which do not fragment to produce these ions (16). (C) CID product ion spectrum of the 1920.9-dalton phosphopeptide from fraction 8. The yn-ion series shows that residues 1651 to 1664 are not phosphorylated. A weak bn series is present, showing a b2 ion corresponding to unmodified Thr1648 and Ser1649. All subsequent bn ions (b3 to b10) are shifted in mass by 80 daltons. For the sake of clarity, not all ions are labeled on the spectrum. The actual sequence coverage is indicated on the peptide sequence. Nomenclature is after Biemann (34). (D) Localization of the phosphorylation site by mutagenesis in positions 1648 through 1650. The Ser or Thr residues at positions 1648 through 1650 were mutated to Ala. Amino acid sequences in positions 1648 to 1650 corresponding to wild-type, single, double, or triple alanine substitutions (17) are shown above each lane.

We used recombinant MGT phosphorylated by mitotic extracts to map the site of phosphorylation with a multidimensional mass spectrometry–based strategy (14). Tryptic digests of MGT treated with mitotic and interphase egg extracts were analyzed for phosphopeptide content by online liquid chromatography–electrospray mass spectrometry, monitoring for the highly diagnostic phosphopeptide marker ion mass-to-charge ratio (m/z) 79 (PO3 ) (15). A comparison of the marker ion trace for the mitotic and interphase MGT samples (Fig. 2B) showed an increase in the level of phosphorylation present in a cluster of three chromatographic peaks (labeled 7 to 9 in Fig. 2B). We used matrix-assisted laser desorption/ionization–post source decay (MALDI-PSD) (16) to show that these peaks contained a phosphopeptide of mass 1920.9 daltons, which corresponded to an MGT tryptic peptide, TSSIADEGTYTLDSILR (17), plus 1 mol of phosphate. To define the location of the modified residue from among the seven potential phosphorylation sites, we sequenced the 1920.9-dalton peptide in each of the three fractions (18) using nanoelectrospray tandem mass spectrometry (MS/MS). Sequence data from all three fractions showed that the phosphorylation site is located within the first three residues of each peptide (corresponding to Thr1648, Ser1649, and Ser1650 in full-length mouse myosin-Va); however, we were unable to assign the specific site of modification for any of them. The MS/MS spectrum of the peptide from the most abundant fraction (fraction 8) contained some evidence to suggest that the major site of phosphorylation is Ser1650 (Fig. 2C). To rule out the possibility that we had overlooked additional phosphorylated sequences in fractions 7 to 9, we combined the fractions and analyzed them by nanoelectrospray MS with a precursor ion scan of m/z 79 (18). No other sequences were detected, which confirmed that a single peptide was selectively phosphorylated in mitotic extracts.

To unambiguously localize the site(s) of phosphorylation, we generated seven mutant versions of MGT with all possible single, double, and triple substitutions of serine and threonine in positions 1648 to 1650 to alanine, and we compared the phosphorylation of mutant and wild-type proteins in mitotic extracts (11). Only the mutation in position 1650 affected phosphorylation (Fig. 2D). Hence, a protein kinase(s) in mitotic extract selectively phosphorylated MGT at Ser1650.

Mapping of the mitotic phosphorylation site allowed us to test whether phosphorylation indeed releases myosin-V from melanosomes. If phosphorylation regulates binding, the substitution in MGT (MGT-Ser1650Ala) should create a protein that binds to melanosomes but cannot be phosphorylated and released by mitotic extracts. On the other hand, replacement with negatively charged glutamic acid should mimic the phosphorylated state, and MGT carrying this substitution should be unable to bind to melanosomes.

Constructs encoding the mutant or wild-type forms of myc-MGT were expressed in melanophores, melanosomes were purified and incubated in mitotic extracts, and binding to melanosomes and release was analyzed by Western blotting (8). The results of this experiment demonstrate that both wild-type and Ser1650Ala forms of MGT bound to melanosomes. However, unlike the wild-type protein, MGT-Ser1650Ala did not dissociate from organelles in mitotic extracts (Fig. 3A). Conversely, MGT-Ser1650Glu could not bind to melanosomes.

Figure 3

Phosphorylation regulates binding and release during mitosis. (A) Melanosome binding of MGT, MGT-Ser1650Ala, and MGT-Ser1650Glu. Melanophores were transfected with the following constructs (35): wild-type myc-MGT, myc-MGT-Ser1650Ala, or myc-MGT-Ser1650Glu. Melanosomes were purified (36), incubated with Xenopus egg extracts, repurified, and subjected to Western blotting. Wild-type and mutant proteins are listed above each pair of lanes, and the treatment with either mitotic (M) or interphase (I) egg extract is listed below. (B) Autoradiography of an in vitro kinase assay. Purified mouse brain CaMKII phosphorylates wild-type MGT but not MGT-Ser1650Ala. (C) Inhibitors of CaMKII prevent the release of endogenous myosin-V from melanosomes in mitotic egg extract. As previously shown, mitotic egg extract (M) dissociated myosin-V from melanosomes, whereas interphase egg extract (I) did not. AIP (10 μM) or 5,5′-dibromo-BAPTA (10 mM) prevented myosin-V release from the melanosome surface when added to mitotic egg extract.

If the phosphorylation of myosin-V governs organelle binding, what kinase or phosphatase is responsible for this regulation? Two obvious candidates for this phosphorylation are p34(Cdc2) and Plx1, two protein kinases that are known to be up-regulated during mitosis (19, 20). Both of these kinases phosphorylate recombinant myc-MGT in vitro. However, they phosphorylate MGT-Ser1650Ala to the same extent as wild-type MGT. Because MGT-Ser1650Ala was not phosphorylated in mitotic egg extracts, neither p34(Cdc2) nor Plx1 are likely to be involved in myosin-V regulation. On the other hand, the mitotic phosphorylation site in MGT (R-T-S-S) matches the consensus sequence for calcium/calmodulin–dependent protein kinase II (CaMKII) (R-X-X-S/T) (17, 21). CaMKII has also been known to copurify with myosin-V during biochemical fractionation and binds to its tail domain (22). Thus, CaMKII may be responsible for the phosphorylation of myosin-V and the regulation of its binding to organelles. To test this, we first determined whether Ser1650 was a site for CaMKII phosphorylation. Wild-type MGT was indeed phosphorylated by purified brain CaMKII, whereas the Ser1650 → Ala replacement abolished the phosphorylation (Fig. 3B). Therefore, Ser1650 is the only site accessible for CaMKII in MGT. Although both Thr1648 and Ser1650 fit the CaMKII consensus site, only Ser1650 was actually phosphorylated by the kinase.

Phosphorylation of the regulatory site on MGT by CaMKII strongly suggested that it may be responsible for myosin-V phosphorylation and release. We tested whether CaMKII inhibition prevented release of myosin-V by mitotic extracts. The treatment of mitotic extracts with CaMKII inhibitors, the calcium chelator 5,5′-dibromo-1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA) or the highly specific peptide inhibitor autocamtide-2–related inhibitory peptide (AIP) (23), strongly inhibited release (Fig. 3C). These inhibitors did not affect H1 kinase activity of mitotic extracts and therefore did not prevent release by converting mitotic extracts to an interphase state. Thus, CaMKII was responsible for the regulation of myosin-V binding to organelles. Consistent with this idea, we found that activity of CaMKII toward a specific peptide substrate (SignaTECT-CaMKII assay system, Promega, Madison, Wisconsin) in mitotic extracts is three to eight times that in interphase extracts, and that 5,5′-dibromo-BAPTA inhibited myosin-V phosphorylation in mitotic extracts. The involvement of CaMKII in the mitotic regulation of myosin-V is also consistent with the fact that its activation is required for mitotic progression in cultured cells (24) and that the level of free Ca2+ in frog-egg extracts is sufficient for the activation of CaMKII (25).

The phosphorylation of myosin-V tail by CaMKII regulates its binding to melanosomes in Xenopus melanophores, explaining cell cycle–dependent inhibition of organelle transport on actin filaments. However, we believe that tail phosphorylation may be a common mechanism regulating organelle transport by myosin-V in vertebrate cells rather than a peculiar feature of Xenopus pigment cells. First, although the sequence of Xenopus myosin-V is not yet known, conservation of the CaMKII site described here in all three forms (a, b, and c) of myosin-V in all vertebrates suggests a universal mechanism of regulation. Second, myosin-V and CaMKII copurify in vitro (22). Third, these two proteins are often found together in vivo. For example, in neurons they are found in postsynaptic densities (26) and on synaptic vesicles (27,28). It is possible that CaMKII regulates myosin-V functions in neurons with the same basic mechanism that is described here for pigment cells.

  • * Present address: Millennium Pharmaceuticals, Cambridge, MA 02139, USA.

  • To whom correspondence should be addressed. E-mail: vgelfand{at}


View Abstract

Navigate This Article