Phosphorylation of myosin light chain kinase by p21-activated kinase PAK2

Phosphorylation of myosin II regulatory light chains (RLC) by Ca(2+)/calmodulin-dependent myosin light chain kinase (MLCK) is a critical step in the initiation of smooth muscle and non-muscle cell contraction. Post-translational modifications to MLCK down-regulate enzyme activity, suppressing RLC phosphorylation, myosin II activation, and tension development. Here we report that PAK2, a member of the Rho family of GTPase-dependent kinases, regulates isometric tension development and myosin II RLC phosphorylation in saponin permeabilized endothelial monolayers. PAK2 blunts tension development by 75% while inhibiting diphosphorylation of myosin II RLC. Cdc42-activated placenta and recombinant, constitutively active PAK2 phosphorylate MLCK in vitro with a stoichiometry of 1.71 +/- 0. 21 mol of PO(4)/mol of MLCK. This phosphorylation inhibits MLCK phosphorylation of myosin II RLC. PAK2 catalyzes MLCK phosphorylation on serine residues 439 and 991. Binding calmodulin to MLCK blocks phosphorylation of Ser-991 by PAK2. These results demonstrate that PAK2 can directly phosphorylate MLCK, inhibiting its activity and limiting the development of isometric tension.     

Phosphorylation of smooth muscle myosin at two distinct sites by myosin light chain kinase

The 20,000-dalton light chain of turkey gizzard myosin is phosphorylated at two sites. Dual phosphorylation is observed when both intact myosin and isolated light chains are used as substrates. Phosphorylation of the second site is not observed at higher ionic strength (e.g. 0.35 M KCl). The first phosphorylation site (serine 19) is phosphorylated preferentially to the second site. The latter is phosphorylated more slowly than the first site, and its phosphorylation requires relatively high concentrations of myosin light chain kinase. It is suggested that myosin light chain kinase catalyzes the phosphorylation of both sites on the light chain, and several reasons are cited that make it unlikely that a contaminant kinase is involved. The second phosphorylation site is a threonine residue. Based on the results of limited proteolysis of the light chain, it is concluded that the threonine residue is close to serine 19, and possible locations are threonines 9, 10, and 18. At all concentrations of MgCl2, phosphorylation of the second site markedly increases the actin-activated ATPase activity of myosin and accelerates the superprecipitation response of myosin plus actin.     

Phosphorylation of a putative myosin light chain inChara by calcium-dependent protein kinase

Summary Ca²⁺-dependent protein kinase (CDPK) has been proposed to mediate inhibition by Ca²⁺ of cytoplasmic streaming in the green algaChara. We have identified the in vivo substrate(s) of CDPK inChara by using vacuolar perfusion of individual internodal cells with [-³²P]ATP. Phosphorylation of several polypeptides is enhanced when perfusions are performed at 10^–4M free Ca²⁺ compared to <10^–9M free Ca²⁺. The Ca²⁺-stimulated phosphorylation of these proteins is inhibited by the presence of a monoclonal antibody to soybean CDPK. One of these proteins is 16 to 18kDa and is recognized by an antibody against gizzard myosin light chains. These results demonstrate that inChara, several polypeptides are phophorylated by CDPK and one of these proteins has been tentatively identified as a myosin light chain. These observations support the hypothesis that Ca²⁺-regulated phosphorylation of myosin is involved in the regulation of cytoplasmic streaming.Abbreviations CDPK calcium-dependent protein kinase - mAb monoclonal antibody     

Phosphorylation of a second site for myosin light chain kinase on platelet myosin

The 20,000-dalton light chain of bovine platelet myosin is phosphorylated at two sites by myosin light chain kinase. The first and second phosphorylation sites are at a serine and a threonine residue, respectively. The location of the phosphorylation sites was determined by using limited proteolysis. The N-terminal sequence of the 17,000-dalton tryptic fragment of platelet myosin 20,000-dalton light chain was found to be identical with that of gizzard 20,000-dalton light chain from Ala-17 to Phe-33. On the basis of these results and the distribution of 32P among the proteolytic fragments, it was concluded that serine-19 and threonine-18 were the two phosphorylation sites. Phosphorylation at the threonine residue markedly increases the actin-activated ATPase activity of myosin. It was found that platelet myosin forms 10S and 6S conformations and its Mg2+-ATPase activity parallels the transition from the 6S to the 10S conformation. The conformational transition was influenced by phosphorylation at both sites, and the phosphorylation at the threonine residue further shifted the equilibrium toward the 6S conformation. The phosphorylation at the threonine residue also induced thick filament formation in the presence of ATP. These results suggest that the phosphorylation at the threonine residue as well as at the serine residue may play an important role in the contractility of nonmuscle cells.     

Phosphorylation of myosin light chain by a protease-activated kinase from rabbit skeletal muscle

A protease-activated protein kinase that phosphorylates the P light chain of myosin in the absence of Ca2+ and calmodulin has been isolated from rabbit skeletal muscle. The enzyme has properties similar to protease-activated kinase I from rabbit reticulocytes [S. M. Tahara and J. A. Traugh (1981) J. Biol. Chem. 256, 11588-11564], which has been shown to phosphorylate the P light chain of myosin [P. T. Tuazon, J. T. Stull, and J. A. Traugh (1982) Biochem. Biophys. Res. Commun. 108, 910-917]. The protease-activated kinase from skeletal muscle has been partially purified by chromatography on DEAE-cellulose, phosphocellulose and hydroxyapatite. The enzyme phosphorylates histone as well as the P light chain of myosin following activation by proteolysis. Stoichiometric phosphorylation of myosin light chain was observed with the protease-activated kinase and myosin light chain kinase. The sites phosphorylated by the protease-activated kinase and myosin light chain kinase were examined by two-dimensional peptide mapping following chymotryptic digestion. The phosphopeptides observed with the protease-activated kinase were different from those obtained with the Ca2+-dependent myosin light chain kinase, indicating that the two enzymes phosphorylated different sites on the P light chain of skeletal muscle myosin. When actomyosin from skeletal muscle was examined as substrate, the P light chain was phosphorylated following activation of the protease-activated kinase by limited proteolysis.     

Acanthamoeba myosin I heavy chain kinase is activated by phosphatidylserine-enhanced phosphorylation

The actin-activated Mg2(+)-ATPase activities of myosins I from Acanthamoeba castellanii are fully expressed only when a single amino acid on their heavy chain is phosphorylated by myosin I heavy chain kinase. Here we show that kinase isolated by a procedure designed to minimize its phosphorylation during purification can incorporate up to 7.5 mol of phosphate/mol of enzyme when incubated with ATP, possibly by autophosphorylation. The rate of phosphorylation is enhanced about 20-fold by phosphatidylserine but is unaffected by calcium ions. Phosphorylation increases the rate at which the kinase phosphorylates the regulatory site of myosin I by about 50-fold. These results suggest that (auto?)phosphorylation may regulate the activity of myosin I heavy chain kinase in vivo. The stimulation of kinase phosphorylation by phosphatidylserine (other phospholipids have not yet been tested) is of particular interest because myosin I has been shown to be tightly associated with membranes, especially the plasma membrane.     

Phosphorylation of skeletal muscle myosin light chain kinase by the catalytic subunit of cAMP-dependent protein kinase

Phosphatidylethanolamine (PE) was isolated from membranes of Bacillus megaterium. The organism was grown at 20°C and 55°C. The phase equilibria in PE/water systems were studied by ²H and ³¹P nuclear magnetic resonance, and by polarized light microscopy. PE isolated from B. megaterium grown at 20°C forms a lamellar liquid crystalline phase at the growth temperature, and at low water contents a cubic liquid crystalline phase at 58°C. The ratio iso/ante-iso acyl chains was 0.3 in this lipid. PE isolated from this organism grown at 55°C forms only a lamellar liquid crystalline phase up to at least 65°C. In this lipid the ratio iso/ante-iso acyl chains was 3.2.     

Phosphorylation and partial sequence of pregnant sheep myometrium myosin light chain kinase

The function of the uterine smooth muscle in gestation and parturition is affected by a variety of hormones and biomolecules, some of which alter the intracellular levels of cAMP and Ca²⁺. Since the activity of smooth muscle MLCK has been shown to be modulated by phosphorylation, the effect of this modification of pregnant sheep myometrium (psm) MLCK by the catalytic subunit of cAMP-dependent protein kinase (PKA) and protein kinase C (PKC) was studied. In contrast to other smooth muscle MLCK reported, PKA incorporates 2.0–2.2 moles phosphate into a mole of psm MLCK both in the presence and absence of Ca²⁺-calmodulin. Modification of serine residues inhibited the activity of the enzyme. PKC also incorporated 2.0–2.1 moles of phosphate per mole psmMLCK under both conditions but had no effect on the MLCK activity. Sequential phosphorylation by PKC and PKA incorporated 3.8–4.1 moles phosphate suggesting that the amino acid residues modified by the two kinases are different. Phosphoamino acid analysis of the MLCK revealed that PKC phosphorylated serine and threonine residues. The double reciprocal plots of the enzyme activity and calmodulin concentrations showed that the V_max of the reaction is not altered by phosphorylation by PKA but the calmodulin concentration require for half-maximal activation is increased about 4-fold. Only 10 out of 17 monoclonal antibodies to various regions of the turkey gizzard MLCK cross-reacted with psmMLCK suggesting structural differences between these enzymes. Comparison of the deduced amino acid sequence of the cDNA encoding the C-terminal half of the psmMLCK molecule showed that while cgMLCK and psmMLCK are highly homologous, a number of nonconservative substitutions are present, particularly near the PKA phosphrylation site B (S⁸²⁸).     

Phosphorylation by cyclic adenosine 3':5'-monophosphate-dependent protein kinase regulates myosin light chain kinase

Smooth muscle myosin light chain kinase, purified to homogeneity, has a molecular weight of 130,000 +/- 5,000 in sodium dodecyl sulfate polyacrylamide gel electrophoresis. The purified enzyme has a specific activity under maximal conditions of 30 mumol Pi transferred to myosin light chain/mg kinase/min at 24 C and is totally dependent on calmodulin and calcium for activity. Incubation of myosin kinase with the catalytic subunit of cyclic adenosine 3':5'-monophosphate-dependent protein kinase results in the covalent incorporation of up to one mol of phosphate per mol of myosin kinase in the absence of bound calmodulin. Limited tryptic digestion of the radioactively labeled kinase indicates that all of the label has been incorporated into a single tryptic peptide (mol wt approximately 22,000), suggesting that a single site is being phosphorylated. Phosphorylation of myosin kinase lowers the rate at which the kinase phosphorylates myosin light chain. The lower rate of light chain phosphorylation is due to a weaker binding of calmodulin to the phosphorylated kinase than to the unphosphorylated kinase. Cyclic adenosine 3':5'-monophosphate-dependent phosphorylation of the kinase actin-myosin interaction represents a possible link between hormonal binding to smooth muscle receptors and muscle relaxation. A scheme for this sequence of events is presented.     

Diphosphorylation of platelet myosin by myosin light chain kinase.

Recently, one of the authors (K.I.) and other investigators reported that myosin light chain (MLC) of smooth muscle (gizzard, arterial and tracheal) was diphosphorylated by myosin light chain kinase (MLCK) and that diphosphorylated myosin showed a marked increase in the actin-activated myosin ATPase activity in vitro and ex vivo. In this study, we prepared myosin, actin, tropomyosin (human platelet), MLCK (chicken gizzard) and calmodulin (bovine brain) and demonstrated diphosphorylation of MLC of platelet by MLCK in vitro. Our results are as follows. (1) Platelet MLC was diphosphorylated by a relatively high concentration (greater than 20 micrograms/ml) of MLCK in vitro. As a result of diphosphorylation, the actin-activated myosin ATPase activity was increased 3 to 4-fold as compared to the monophosphorylation. (2) Both di- and monophosphorylation reactions showed similar Ca2+, KCl, MgCl2-dependence. Maximal reaction was seen at [Ca2+] greater than 10(-6) M, 60 mM KCl and 2 mM MgCl2. This condition was physiological in activated platelets. (3) Di- and monophosphorylated myosin showed similar Ca2+, KCl-dependence of ATPase activity but distinct MgCl2-dependence. Diphosphorylated myosin showed maximal ATPase activity at 2 mM MgCl2 and monophosphorylated myosin showed a maximum at 10 mM MgCl2. (4) The addition of tropomyosin stimulated actin-activated ATPase activity in both di- and monophosphorylated myosin to the same degree. (5) ML-9, a relatively specific inhibitor of MLCK, inhibited the aggregation of human platelets induced by thrombin ex vivo in a dose-dependent manner. Moreover, this drug also partially inhibited both di- and monophosphorylation reactions and actin-activated ATPase activity. On the other hand, H-7, a synthetic inhibitor of protein kinase C, had little effect on the aggregation of human platelets induced by thrombin ex vivo. From these results, we conclude that diphosphorylation of platelet myosin by MLCK may play an important role in activated platelets in vivo.     

Phosphorylation of smooth muscle myosin light chain kinase by Ca2+-activated, phospholipid-dependent protein kinase

Protein kinase C incorporates phosphate into two sites of myosin light chain kinase (MLC-kinase) in the absence of calmodulin. Phosphorylation is all but abolished in the presence of Ca2+ and calmodulin, suggesting that both sites of phosphorylation are close to the calmodulin binding site. The phosphorylation of MLC-kinase results in an approximately 10-fold increase in the dissociation constant of MLC-kinase for calmodulin. Following phosphorylation (2 mol/mol of enzyme) of MLC-kinase by protein kinase C, an additional 2 mol of phosphate can be incorporated into the MLC-kinase apoenzyme by the cAMP-dependent protein kinase. Different maps of phosphopeptides were obtained by tryptic hydrolysis from MLC-kinase preparations phosphorylated by each kinase. The phosphorylation sites for the cAMP-dependent kinase were located in a fragment of approximately 25,000 daltons. In contrast the phosphorylation sites for protein kinase C are found in a much smaller tryptic peptide. These results suggest that the phosphorylation sites on MLC-kinase are different for protein kinase C and for cAMP-dependent protein kinase. However, phosphorylation in both regions results in a reduced affinity for calmodulin.     

Phosphorylation of smooth myosin light chain kinase by smooth muscle Ca2+/calmodulin-dependent multifunctional protein kinase

Smooth muscle myosin light chain kinase (MLC kinase) was phosphorylated by smooth muscle calmodulin-dependent protein kinase II (CaM protein kinase II). When MLC kinase was free from calmodulin, two sites were phosphorylated. The phosphorylation at the one site was much faster than the other site; however, the phosphorylation at the first site was completely blocked by calmodulin binding to MLC kinase. Phosphorylation of MLC kinase by CaM protein kinase II increased the dissociation constant of MLC kinase for calmodulin about 10 times without changing the Vmax. The location of the phosphorylation sites was identified by isolating and sequencing the tryptic phosphopeptides of MLC kinase. The preferred site was identified as serine 512 and the second site as serine 525. These sites are the same as the sites phosphorylated by cAMP-dependent protein kinase.     

Inhibition of myosin light chain kinase by amiloride

Phosphorylation of regulatory light chain (LC20) by myosin light chain kinase (MLCK) has been thought to play an important role in both smooth muscle contraction and several functions of vertebrate non-muscle cells. Amiloride, a frequently used Na+/H+ exchange inhibitor, potently inhibited phosphorylation of LC20 by MLCK. The inhibition was non-competitive with respect to myosin but competitive with ATP (Ki = 0.95 microM), suggesting that amiloride may act as an ATP analogue. Amiloride also inhibited the tension development of ether-treated gizzard fibers which were lacking in Na+/H+ antiport, even in the presence of ATP regenerating system. Thus, it must be reminded that amiloride cannot be used as a specific inhibitor of Na+/H+ exchange, and that the inhibition of myosin phosphorylation by amiloride should be taken into consideration in studying the role of Na+/H+ antiport in the cellular function.     

Phosphorylation of myosin light chain kinase by the multifunctional calmodulin-dependent protein kinase II in smooth muscle cells.

Stimulation of tracheal smooth muscle cells in culture with ionomycin resulted in a rapid increase in cytosolic free Ca2+ concentration ([Ca2+]i) and an increase in both myosin light chain kinase and myosin light chain phosphorylation. These responses were markedly inhibited in the absence of extracellular Ca2+. Pretreatment of cells with 1-[N-O-bis(5-isoquinolinesulfonyl)-N- methyl-L-tyrosyl]-4-phenylpiperazine (KN-62), a specific inhibitor of the multifunctional calmodulin-dependent protein kinase II (CaM kinase II), did not affect the increase in [Ca2+]i but inhibited ionomycin-induced phosphorylation of myosin light chain kinase at the regulatory site near the calmodulin-binding domain. KN-62 inhibited CaM kinase II activity toward purified myosin light chain kinase. Phosphorylation of myosin light chain kinase decreased its sensitivity to activation by Ca2+ in cell lysates. Pretreatment of cells with KN-62 prevented this desensitization to Ca2+ and potentiated myosin light chain phosphorylation. We propose that the Ca(2+)-dependent phosphorylation of myosin light chain kinase by CaM kinase II decreases the Ca2+ sensitivity of myosin light chain phosphorylation in smooth muscle.     

Phosphorylation of smooth muscle myosin light chain kinase by protein kinase C. Comparative study of the phosphorylated sites

Smooth muscle myosin light chain kinase is phosphorylated in vitro by protein kinase C purified from human platelets. When myosin light chain kinase which has calmodulin bound is phosphorylated by protein kinase C, 0.8-1.1 mol of phosphate is incorporated per mol of myosin light chain kinase with no effect on its enzyme activity. Phosphorylation of myosin light chain kinase with no calmodulin bound results in the incorporation of 2-2.4 mol of phosphate and significantly decreases the rate of myosin light chain kinase activity. The decrease in myosin light chain kinase activity is due to a 3.3-fold increase in the concentration of calmodulin necessary for the half-maximal activation of myosin light chain kinase. The sites phosphorylated by protein kinase C and the catalytic subunit of cAMP-dependent protein kinase were compared by two-dimensional peptide mapping following extensive tryptic digestion of phosphorylated myosin light chain kinase. The single site phosphorylated by protein kinase C when calmodulin is bound to myosin light chain kinase (site 3) is different from that phosphorylated by the catalytic subunit of cAMP-dependent protein kinase (site 1). The additional site that is phosphorylated by protein kinase C when calmodulin is not bound appears to be the same site phosphorylated by the catalytic subunit of cAMP-dependent protein kinase (site 2). These studies confirm the important role of site 2 in binding calmodulin to myosin light chain kinase. Sequential studies using both protein kinase C and the catalytic subunit of cAMP-dependent protein kinase suggest that the phosphorylation of site 1 also plays a part in decreasing the affinity of myosin light chain kinase for calmodulin.     

Kaempferol inhibits myosin light chain kinase

Kaempferol, 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H-1-benzopyran-4-one, was found to inhibit bovine aorta myosin light chain kinase with a Ki of 0.3-0.5 microM. It was found to be competitive with ATP and non-competitive with isolated myosin light chains. The specificity of this inhibitor was studied relative to protein kinase C and cAMP dependent protein kinase (IC50 = 15 microM and 150 microM, respectively). It appears not to interact strongly with calmodulin binding proteins, such as Ca2+-calmodulin dependent phosphodiesterase (IC50 = 45 microM), and had little effect on actin-activated myosin subfragment-1 ATPase activity (IC50 greater than 100 microM) or smooth muscle phosphatase activities (IC50 greater than 100 microM).     

Phosphorylation of synthetic peptides by skeletal muscle myosin light chain kinases

Substrate determinants for rabbit and chicken skeletal muscle myosin light chain kinases were examined with synthetic peptides. Both skeletal muscle myosin light chain kinases had similar phosphorylation kinetics with synthetic peptide substrates. Average kinetic constants for skeletal muscle myosin light chain heptadecapeptide, (formula; see text) where S(P) is phosphoserine, were Km, 2.3 microM and Vmax, 0.9 mumol/min/mg of enzyme. Km values were 122 and 162 microM for skeletal muscle peptides containing A-A for basic residues at positions 2-3 and 6-7, respectively. Average kinetic constants for smooth muscle myosin light chain peptide, (formula; see text), were Km, 1.4 microM and Vmax 27 mumol/min/mg of enzyme. Average Km values for the smooth muscle peptide, residues 11-23, were 10 microM which increased 6- and 11-fold with substitutions of alanine at residues 12 and 13, respectively. Vmax values decreased and Km values increased markedly by substitution of residue 16 with glutamate in the 11-23 smooth muscle tridecapeptide. Basic residues located 3 and 6-7 residues toward the NH2 terminus from phosphoserine in smooth muscle myosin light chain and 6-8 and 10-11 residues toward the NH2 terminus from phosphoserine in skeletal muscle myosin light chain appear to be important substrate determinants for skeletal muscle myosin light chain kinases. These properties are different from myosin light chain kinase from smooth muscle.     

Selective binding of L-thyroxine by myosin light chain kinase

L-Thyroxine selectively inhibited Ca2+-calmodulin-activated myosin light chain kinases (MLC kinase) purified from rabbit skeletal muscle, chicken gizzard smooth muscle, bovine thyroid gland, and human platelet with similar Ki values (Ki = 2.5 microM). A detailed analysis of L-thyroxine inhibition of smooth muscle myosin light chain kinase activation was undertaken in order to determine the effect of L-thyroxine on the stoichiometries of Ca2+, calmodulin, and the enzyme in the activation process. The kinetic data indicated that L-thyroxine does not interact with calmodulin but, instead, through direct association with the enzyme, inhibits the binding of the Ca2+-calmodulin complex to MLC kinase. L-[125I]Thyroxine gel overlay revealed that the 95-kDa fragment of chicken gizzard MLC kinase digested by chymotrypsin and all the fragments of 110, 94, 70, and 43 kDa produced by Staphylococcus aureus V8 protease digestion which contain the calmodulin binding domain retain L-[125I]thyroxine binding activity, whereas smaller peptides were not radioactive. Since MLC kinase is phosphorylated by cAMP-dependent protein kinase (2 mol of phosphate/mol of MLC kinase), the effect of L-thyroxine on the phosphorylation of MLC kinase also was examined. L-Thyroxine binding did not inhibit the phosphorylation of MLC kinase and, moreover, reversed the inhibition of phosphorylation obtained with the calmodulin-enzyme complex. These observations support the suggestion that L-thyroxine binds at or near the calmodulin-binding site of MLC kinase. L-Thyroxine may serve as a different type of pharmacological tool for elucidating the biological significance of MLC kinase-mediated reactions.     

Properties of filament-bound myosin light chain kinase

Myosin light chain kinase binds to actin-containing filaments from cells with a greater affinity than to F-actin. However, it is not known if this binding in cells is regulated by Ca2+/calmodulin as it is with F-actin. Therefore, the binding properties of the kinase to stress fibers were examined in smooth muscle-derived A7r5 cells. Full-length myosin light chain kinase or a truncation mutant lacking residues 2-142 was expressed as chimeras containing green fluorescent protein at the C terminus. In intact cells, the full-length kinase bound to stress fibers, whereas the truncated kinase showed diffuse fluorescence in the cytoplasm. After permeabilization with saponin, the fluorescence from the truncated kinase disappeared, whereas the fluorescence of the full-length kinase was retained on stress fibers. Measurements of fluorescence intensities and fluorescence recovery after photobleaching of the full-length myosin light chain kinase in saponin-permeable cells showed that Ca2+/calmodulin did not dissociate the kinase from these filaments. However, the filament-bound kinase was sufficient for Ca2+-dependent phosphorylation of myosin regulatory light chain and contraction of stress fibers. Thus, dissociation of myosin light chain kinase from actin-containing thin filaments is not necessary for phosphorylation of myosin light chain in thick filaments. We note that the distance between the N terminus and the catalytic core of the kinase is sufficient to span the distance between thin and thick filaments.     

Mitosis-specific phosphorylation of myosin light chain kinase

Cell cytosol preparations from mitotic HeLa cells exhibit a kinase activity that phosphorylates myosin light chain kinase (MLCK). This MLCK kinase activity is apparently distinct from the known MLCK kinases, including cAMP-dependent protein kinase, cGMP-dependent protein kinase, Ca(2+)-activated phospholipid-dependent protein kinase, or Ca(2+)-calmodulin-dependent protein kinase II, based on the following criteria. First, the MLCK kinase activity of mitotic cells does not respond to a variety of characteristic activators or inhibitors of these known kinases. Second, one- and two-dimensional peptide maps have revealed that the site of phosphorylation by the MLCK kinase of mitotic cells differs from those by these known kinases. The mitotic MLCK kinase phosphorylates MLCK at a threonine residue at a ratio of up to 1 mol of phosphate/mol of chicken gizzard MLCK. The MLCK kinase is mitosis-specific because mitotic cell extracts show much higher phosphorylation activity than nonmitotic cell extracts.