Temporal regulation of cdc2 mitotic kinase activity and cyclin degradation in cell-free extracts of Xenopus eggs

During the first few hours following fertilization of Xenopus laevis eggs, the cell division cycle is abbreviated to a rapid succession of S and M phases, without any significant Gj and G2 phases (Newport and Kirschner, 1982; Signoret and Lefresne, 1971). The rapid alternation between two cytoplasmic states, interphase and mitosis, has been interpreted in terms of a ‘master oscillator’ that controls the various structural and metabolic events which characterize S and M phases (Kirschner et al. 1985).

The molecular nature of this oscillator is beginning to be defined. The level of overall protein phosphorylation increases in M phase in all eukaryotes (Ajiro et al. 1981; Bradbury et al. 1973; Capony et al. 1986; Dorée et al. 1983; Karsenti et al. 1987; Mailer et al. 1977; Meijer et al. 1982; Peaucellier et al. 1984). This is correlated with the activation of a major histone Hl kinase and a cytoplasmic activity called MPF for maturation promoting factor (Capony et al. 1986; Cicarelli et al. 1988; Guerrier et al. 1977; Picard et al. 1987; Sano, 1985; Mailer, 1985). Using a cell-free assay developed from frog eggs, in which MPF activity can be followed by the induction of nuclear envelope breakdown and the formation of mitotic chromosomes (Lohka and Maller, 1985; Lohka and Masui, 1984; Miake-Lye and Kirschner, 1985), Lohka et al. (1988) obtained a high degree of MPF purification. Two major polypeptides of 32 and 45K (K= 10³M_r) are present, and a histone Hl kinase activity is associated with MPF. The histone Hl kinase activity is apparently due to the 32K polypeptide that has been identified as the homologue of the S. pombe cdc2⁺ gene product in frogs, starfish and human (Arion et al. 1988; Gautier et al. 1988; Labbé et al. 19886; Lee and Nurse, 1987). It thus appears that the kinase activity associated with the homologue of the cdc2⁺ gene product plays a fundamental role in the ‘initiation of mitosis. Since the level of cdc2⁺ polypeptide does not appear to fluctuate Muring the cell cycle (Draetta and Beach, 1988; Labbé et al. 19886; Simanis and Nurse, 1986), whereas its activity does, the focus of attention turns to the matter of Mow the activity of this enzyme is controlled to give the metastable mitotic state.

It is well established that synthesis of new proteins is required for turning mitotic kinase on at each cell cycle (Karsenti et al. 1987; Miake-Lye et al. 1983; Picard et al. 1985; Wagenaar, 1983), and a family of proteins, the cyclins, are prime candidates to account for the protein synthesis that is required to drive the cell cycle. They are . synthesized continuously, accumulate during interphase and are destroyed abruptly during mitosis (Evans et al. 1983; Pines and Hunt, 1987; Standart et al. 1987; Swenson et al. 1986). Antisense ablation of cyclin mRNA from Xenopus extracts prevents them entering mitosis (Minshull et al. 1989), and conversely, sea urchin cyclin mRNA is able to restore mitotic cycles to such extracts after their mRNA has been destroyed by RNase (Murray and Kirschner, 1989). This shows that cyclin I synthesis is required for activation of the kinase. Genetic analysis in yeast has revealed that regulation of cdc2⁺ kinase activity involves at least two other kinases (Russell and Nurse, 1987). This suggests that the mechanism of kinase activation requires a series of kinases and phosphatases that comprise a cascade of reactions. It is not yet clear what role cyclin plays in this cascade. In order to study the control of mitosis, a cell-free system is needed that mimics as closely as possible the activation and inactivation kinetics of the cdc2⁺ kinase observed in vivo.

In this paper, we report the characterization of cell-free extracts from Xenopus eggs, in which the temporal regulation of the mitotic histone Hl kinase activity can be studied under physiological conditions. We show that histone Hl kinase activity oscillates at least twice in low speed extracts of activated eggs incubated at 22°C. The oscillation can be divided into two phases. The first requires protein synthesis, while in the second one, the kinase activates spontaneously. This is similar to what is observed in vivo. Kinase activity is lost abruptly, coincident with destruction of added sea urchin cyclin at the end of metaphase.

When these extracts are fractionated by successively higher speed centrifugation, it emerges that two regulatory components involved in the activation of the histone kinase are associated with particulate material. One component sediments above 100 000g and seems to inhibit kinase activation; it appears to be inactivated by phosphorylation. Another component sediments only at very high speed (more than 2h at 250 000g). It is part of the activation system of the kinase. Our data suggest, however, that the final activation of the kinase involves protein dephosphorylation.

These results show that the normal physiological mechanism of activation of the mitotic kinase is very complex. It involves several steps, not all of which are controlled by soluble molecules. Further studies of this system should allow us to identify the various elements involved in the control of the mitotic protein kinase.

FemaleXenopus laevis were purchased from the Service d’élevage d’Amphibiens CNRS (France). Eggs were obtained from females injected 3 to 8 days before use with 100 units of pregnant mare serum gonadotropin (PMSG) (Intervet, France) and the day before with 1500 units of human chorionic gonadotropin (HCG) (Sigma). The eggs were collected in 0.1M-NaCl to prevent activation. The jelly coat was removed with 2% cysteine-HCl, pH 7.8 and the eggs were then washed extensively with modified Ringer’s solution MMR/4 (25 mM-NaCl, 0.4mM-KCl, 0.25 IHM- MgSCh, O.SmM-CaCb, 1.25 mM-Hepes, 25μM-EDTA, pH7.2).

Eggs were activated by an electric shock (Karsenti et al. 1984), incubated for 55-60 min at 22°C and then transferred to cold acetate buffer (100 mM-K-acetate, 2.5 mM-Mg-acetate, 60mM-EGTA, 5μgml^-1 cytochalasin D (Sigma), ImM-DTT, 250 mM-sucrose, pH7.2). Excess acetate buffer was removed prior to centrifugation. The 2000 g supernatants were obtained by crushing the eggs by gentle pipetting with a 1 ml Gilson pipette and centrifuging the homogenate at 2000 g’for 10 min at 4 °C. This removed the yolk, but left most of the cytoplasmic components in the supernatant. The 10 000g and high-speed supernatants were obtained from eggs crushed by centrifugation at 10 000g for 10 min at 4°C in an L5-65 Beckman centrifuge with maximum acceleration rate. In both cases, the cytoplasmic material between the upper lipid layer and the yolk pellet was collected and an ATP regenerating system was added. We used a final concentration of 10 mM creatine phosphate (Boehringer, from a 0.5M stock in H₂O), 80μgml^-1 creatine kinase (Boehringer, 4 mg ml^-1 stock in 50 % v/v glycerol/), 1 mM-ATP (disodium salt, Boehringer, 50 mM stock in H₂O, pH 7). The 2000 and 10 000 g supernatants were kept on ice and used the same day. Highspeed extracts were obtained by further centrifugation at 100 000g for 1 h, or 250 000g for 2h at 4°C in the SW50.1 or SW60 rotors with adaptors for 0.6 ml tubes. These supernatants were frozen in 50 μl aliquots in liquid nitrogen. The protein concentration (determined by the Bradford procedure using bovine serum albumin as a standard) was 35-45 mg ml^-1 in the 2000, 10 000 g and 100 000g supernatants and approximately 20 mg ml^-1 in the 250 000 g supernatants.

To assay histone Hl kinase, 2 μl of extract to be assayed was transferred into 50 μl of extraction buffer (EB; 80mM-β-glycerophosphate, 20mM-EGTA, 15mM-MgCb, ImM-DTT, 1 mM-PMSF, 10μgml^-1 leupeptin, 10μgml^-1 pepstatin, 10μgml^-1 aprotinin, pH7.3) to stabilize kinase activity. The samples were frozen in liquid nitrogen and stored at —20°C. The histone kinase activity was assayed by adding 10 μl of this sample to 6 μl of a mix containing 3.3 mg ml^-1 histone Hl (H III-S from calf thymus, Sigma), 1 mM-ATP, 0.25 μCiμl^-1 [γ-³²P]ATP (Amersham PB 218). The reaction was performed for 15 min at 20°C and stopped by pipetting 12 μl of the reaction mixture on P81 phosphocellulose paper (Whatman) precut in 1.5 cm² squares. The filters were washed three times for 15 min in 150mM-H₃PO₄ (Merck) (at least 20ml/filter), rinsed in ethanol, allowed to dry and Cerenkov-counted. The background obtained in the absence of extract was subtracted. In some cases, histone phosphorylation was analysed by SDS-PAGE. The reaction was then stopped by addition of an equal volume of SDS sample buffer and the samples were run on a 15 % minigel according to Laemmli (1970). The gels were fixed in three changes of 12.5 % isopropanol, 10% acetic acid, dried and exposed using Kodak X-AR or X-AS paper.

Arbacia punctulata cyclin mRNA was transcribed with T7 RNA polymerase from a full length clone (cyc4) in pGEMl as described by Pines and Hunt (1987). The mRNA was translated in the presence of [³⁵S] methionine in a reticulocyte lysate. The total translation mix containing [³⁵S]cyclin -was added to rhe Xenopus extract in a proportion of 1 volume per 5 vol. of extract. The extract was incubated at room temperature and 4 μl samples were transferred into 16 μl of SDS gel sample buffer. The samples were run on a 10 % gel. The gel was fixed in 45 % methanol, 7 % acetic acid and processed using the ‘Intensify’ (DuPont) procedure.

The particulate material of an extract of Xenopus eggs prepared 60 min after egg activation was obtained as follows. The 10000’supernatant was further centrifuged at 100000g’in a completely full 5 ml SW 50.1 tube. The supernatant, including the fluffy part of the pellet, was recovered and centrifuged at 150 000 gfor 30 min at 4°C in 280pl aliquots in a fixed angle rotor (TLA 100) in the TL 100 Beckman centrifuge. The pellets were resuspended in 3 vol. of acetate buffer containing 10 mM-EGTAand the ATP regenerating system. Aliquots of 100 μl were frozen and stored in liquid nitrogen.

A discontinuous sucrose gradient was prepared by layering successively 100 μl of 70 %, 50 %, 40 %, 30%, 20% w/v sucrose in acetate buffer (W/W) in 0.6 ml SW 50.1 tubes. One hundred μl of the TL 100 pellet was layered on top of the gradient and centrifuged at 100000gfor 1 h. Fractions of 50 μl were collected from the top of the gradient. The ‘kinase activator’ activity in the fractions was assayed by adding one volume of the fractions (13 μl) to one volume of a TL 100, 60 min supernatant in which the kinase could not get activated. Histone Hl kinase activation-inactivation was followed as usual in the reconstituted system.

Xenopus eggs are stably arrested in second metaphase of meiosis and contain high histone Hl kinase activity, as shown in the first lane of Fig. 1A. Fertilization or artificial activation releases the metaphase block, leading to a drop in histone Hl kinase activity. This is followed by the reappearance of Hl kinase activity, which peaks at about 80min after fertilization. This corresponds to the time of first metaphase in the intact eggs. Next, the kinase activity drops rapidly, followed by a second round of increase and disappearance peaking at 120 min. Thereafter the kinase activity in intact embryos oscillates with a period of about 30 min.

Fig. 1B shows the results of histone Hl kinase assays performed on low-speed extracts prepared from eggs sampled 60min after activation (‘60 minute’ extract). Initially these extracts contained a low level of histone kinase activity that was similar to the in vivo value (compare the first lane of Fig. IB with the 60 min time point of Fig. 1A). Upon incubation of the ‘60 minute’ extract at room temperature, the histone kinase activity increased, reached a peak at about 20 min and then suddenly dropped to a very low value at 30 min. This occurred very close to the time of kinase inactivation in intact eggs (see Fig. IA, 100min time point). Later, the histone Hl kinase activity increased again in the extract, but much more slowly than in the intact eggs. The first rise and fall of kinase activity observed in the extracts did not require protein synthesis, since it was not inhibited by cycloheximide (Fig. 1C). The second in vitro cycle absolutely required protein synthesis since activation of the kinase did not occur in the presence of cycloheximide (Fig. ID). Thus the slow appearance of kinase activity during the second in vitro cycle was probably due to a low rate of protein synthesis. It was indeed only 20 % of the rate measured in intact embryos (data not shown).

As shown in the top panel of Fig. 2, extracts prepared 60 min after egg activation could be centrifuged at 250 000 g and still retain their ability to activate and turn off histone Hl kinase. The peak of activity in these supernatants varied from extract to extract, often being lower than in the corresponding 10 000 g supernatant. The appearance of kinase activity was accompanied by a large increase in ³²P labelling of endogenous proteins in the extract when assayed with added [γ-³²P]ATP (Fig. 2, middle panel). These proteins corresponded closely to the proteins we previously found to be labelled in vivo during metaphase (Karsenti et al. 1987; Felix et al. 1989).

To test if the disappearance of the kinase activity in the extract was accompanied by destruction of cyclin, we added labelled sea urchin cyclin (produced by translation of synthetic cyclin mRNA in a reticulocyte lysate with [³⁵S] methionine). The added cyclin was indeed degraded between 20 and 30 min, at the very moment of kinase inactivation (Fig. 2, bottom panel). These experiments show that basic biochemical events of the cell cycle previously characterized in vivo can be reproduced in high-speed frozen supernatants of activated Xenopus egg extracts. This should permit study of the activation and inactivation mechanism of the mitotic histone Hl kinase in vitro under almost physiological conditions. In addition, the mechanism of cyclin degradation seems equally amenable to biochemical investigation. We have begun to use this system for both purposes, and the initial results are submitted for publication elsewhere (Felix et al. 1989). We describe below some recent results obtained on the study of the activation mechanism of the mitotic kinase.

As reported in the preceding section, it is possible to obtain one cycle of kinase activation and inactivation in 250 000 g supernatants of extracts prepared 60 min after egg activation. These centrifugation conditions would usually be considered sufficient to completely eliminate all particulate material. However, the extracts used in this study were highly concentrated, and we were concerned that some low-density particulate material might not be sedimented. We therefore did several experiments to investigate this possibility. The left panel of Fig. 3 shows what happened when 200 μl of a 100 000 g supernatant was recentrifuged for 30 min at 150 000 g in the table top BeckmanTL 100 centrifuge. The kinase did not activate in the supernatant of the extract centrifuged in this way, probably because the small distance between the top and bottom of the tubes in this centrifuge allows slowly sedimenting material to be pelleted. The right panel of Fig. 3 shows that this was not a centrifugation artefact, since addition of the resuspended pellet (which itself had no histone kinase activity) to the supernatant (also devoid of kinase activity) restored a normal round of histone Hl kinase activation and inactivation. It thus appeared that particulate material that is difficult to pellet by normal centrifugation of concentrated extracts was involved in the activation process of the mitotic histone Hl kinase.

In the preceding experiment, the pellet and the inactive supernatant were both obtained from eggs homogenized 60 min after activation, a time when further protein synthesis is not needed to support activation of the mitotic histone Hl kinase. It was of interest to test whether the particulate material had to be prepared at a specific time of the cell cycle in order to activate the kinase. We therefore prepared a 100 000 g supernatant from eggs sampled only 20 min after activation. At this time in the cell cycle, the component that has to be synthesized in order to trigger kinase activation (which we strongly suspect is cyclin, according to the recent results of Minshull et al. 1989, and Murray and Kirschner, 1989) should be either missing or in low concentration. Indeed, incubation of such an extract at room temperature for 1 h did not result in any activation of the histone kinase (data not shown). This extract was fractionated as described above into a TL100 pellet and supernatant. Homologous reconstitution between the 20 min pellet and supernatant did not result in kinase activation (Fig. 4D). In Fig. 4A, we show an homologous reconstitution between a 60 min TL 100 supernatant and its pellet. As expected, the slight activation of the histone kinase that occurred in the supernatant (open symbols) was strongly increased by addition of the pellet. Addition of the ‘20 minute’ pellet to the ‘60 minute’ supernatant did not activate the kinase. Instead, it inhibited the residual activation that occurred in the supernatant alone (Fig. 4B). Adding the ‘60 minute’ pellet to the ‘20 minute’ supernatant did not result in kinase activation either (Fig. 4C).

Therefore, the slowly sedimenting particulate material involved in the mechanism of kinase activation is active only when prepared from eggs that have begun to enter mitosis. Moreover, active particulate material is not sufficient to activate the kinase in a supernatant prepared from eggs sampled during interphase.

The results of a preliminary attempt to fractionate the particulate material involved in activation of the kinase on a discontinuous sucrose gradient are shown in Fig. 5. A TL 100 pellet was centrifuged for 1 h at 100000g on a 20–70% discontinuous sucrose gradient, and the fractions added to different aliquots of a TL 100 supernatant. The kinetics of kinase activation was followed for each reconstitution. In this experiment, the kinase showed a low amplitude cycle in the TL 100 supernatant alone (the level of activation of the kinase in the TL 100 supernatants varied somewhat from extract to extract). We attribute the residual cycle of kinase activity observed in some supernatants to incomplete removal of the particulate material. In any case, Fig. 5 clearly shows that a normal level of kinase activation could be restored in the TL 100 supernatants by adding back fractions 3 to 5 of the sucrose gradient (the 20–30% sucrose interface). No other fractions had detectable activity. This confirms that the kinase activator is associated with a slowly sedimenting particle (as a reference, centrosomes would sediment in 50–60% sucrose under the same conditions), and is in agreement with the sedimentation properties of this activity in the concentrated extracts. We have not yet determined precisely the sedimentation coefficient or the density of the activator.

As already mentioned in the introduction, activation of the mitotic kinase requires the presence of cyclin (Minshull et al. 1989; Murray et al. 1989). It also involves a cascade of phosphorylation and dephosphorylation events, as suggested by genetic analysis in yeast (Russell and Nurse, 1987). In fact, the genetic analysis predicts that activation of p34^cdc2 kinase occurs when another kinase (the product of the weel⁺gene) is inhibited by phosphorylation. This suggests that final activation of p34^cdc2may be achieved by dephosphorylation, a hypothesis that is strongly supported by biochemical results from several laboratories working on both Xenopus and starfish (Labbé et al. 1988a,6; Gautier et al. 1989; Dunphy and Newport, 1989; Draetta et al. 1989).

As a first approach to test the involvement of protein phosphorylation in kinase activation we added EDTA to the extracts to block all ATP-Mg²⁺-dependent reactions, or used 6-dimethylaminopurine (6-DMAP). This adenine analog blocks mitosis in fertilized sea urchin eggs at a concentration of 10^-4M, and seems to be mainly a protein kinase inhibitor (Néant et al. 1988). Both these reagents strongly inhibited ³²P incorporation into the proteins of the treated extracts. As Fig. 6 shows, EDTAand 6-DMAP completely inhibited activation of the histone kinase in 10 000 g and 100 000 g supernatants. The activity of the histone kinase could be measured in the extracts containing phosphorylation inhibitors because the EDTA or 6-DMAP were diluted below their active concentration when the extract was diluted in the histone Hl kinase assay buffer. Neither EDTAand 6-DMAP had any effect on the measured kinase activity when added to extracts in which the histone kinase was already activated.

To our surprise, the 250 000 g supernatants gave a completely different result with these inhibitors. Neither EDTA nor 6-DMAP prevented the increase in histone kinase activity, which reached a level equivalent to in vivo values. However, once activated, the activity never dropped. This is described in more detail elsewhere (Felix et al. 1989).

These results draw attention to two important points. First, a particulate material present in 100 000 g supernatants is not needed for activation of the histone Hl kinase when protein phosphorylation is inhibited. Second, the particulate material can be removed by centrifugation at 250000g for 2h. In the absence of the particulate material, the kinase activates normally when all phosphorylations are inhibited in the extract. This is consistent with the idea that the final activation step of the histone Hl kinase represents dephosphorylation and further suggests the existence of a particulate inhibitor of the kinase, which is inactivated by phosphorylation (either directly or indirectly).

In order to test whether the final kinase activation step was due to protein dephosphorylation, we added different concentrations of β-glycerophosphate to 100 000 g and 250 000g supernatants prepared 60 min after egg activation. As shown in Fig. 7, left panel, activation of the histone Hl kinase in 100 000 g supernatants was completely inhibited by 20mM- β glycerophosphate. This is the concentration routinely used to stabilize MPF and histone Hl kinase activity in extracts of non activated eggs. This same concentration of phosphatase inhibitor had no effect on the activation of the kinase in 250000 g’ supernatants, but prevented the normal disappearance of kinase activity (Fig. 7, right panel). Higher β-glycerophosphate concentrations inhibited activation of the histone kinase in the 250000 g supernatants. The effect of β-glycerophosphate was not due to chelation of Mg²⁺ ions, as was shown by adding a molar equivalent of Mg²⁺ acetate to the extract together with β-glycerophosphate (data not shown). Other phosphatase inhibitors (NaF, a-naphthyl phosphate) gave similar results (data not shown).

These experiments suggest that both activation and inactivation of the histone H1 kinase involve dephosphorylation events, but in 250 000 g supernatants, the relative absence of particulate material means that the kinase can be activated even when phosphorylation is blocked and phosphatase inhibitors are present. Higher concentrations of phosphatase inhibitors presumably act by directly inhibiting the dephosphorylation of p34^các2 that is thought to be the final step of kinase activation. These confusing results probably owe their explanation to the existence of a complex chain of protein modifications.

In this paper, we describe the preparation of extracts from activated Xenopus eggs that are capable of more than one cycle of histone Hl kinase activation and inactivation. (Cicarelli et al. 1988; Dabauvalle et al. 1988; Labbé et al. 1988a). We have not formally identified this activity with that of the frog homologue of the cdc2⁺gene product of S. pombe. The cdc2⁺ gene product has strong histone Hl kinase activity from yeast to humans, and this activity is activated only during mitosis (for review, see Lokha, 1989). The histone Hl kinase activity is abolished in crude extracts of a temperature-sensitive mutant of cdc2⁺ in yeast (Moreno et al. 1989). The mitotic histone kinase activity of Xenopus eggs has been partially purified (Labbé et al. 1988a) and fractionates as a single peak through four fractionation steps, and the activity coincides with p34^cdc2. Finally, most of the frog egg histone Hl kinase activity eluted as a single peak from DEAE cellulose and binds to Sepharose beads carrying the yeast pl3^s”^c7 protein (Félix et al. 1989). This protein has been shown to interact directly with p34^ciZc2 (Dunphy et al. 1988; Arion et al. 1988). Therefore, it is extremely probable that the histone kinase activity we observed in this work is due to p34^edc2.

There have been several previous reports of Xenopus egg extracts in which the cell cycle could be reconstituted (Cyert and Kirschner, 1988; Hutchison et al. 1987, 1988). In these studies, either the activity of MPF, morphological markers or the occurrence of DNA synthesis were used to follow the cell cycle. Our goal was to define a cell-free system that reconstituted the cell cycle in vitro with a minimum of alteration to the in vivo situation, in order to have a secure starting point to further simplify the system and study the physiological cascade of reactions involved in the control of mitosis. We therefore needed to define precise molecular markers that could be used to compare the in vivo and in vitro conditions. We believe that protein phosphorylation provides such a marker, since the pattern of ³²P-labelled proteins observed in our extracts is virtually identical to that seen in intact fertilized eggs (Felix et al. 1989). Moreover, the mitotic kinase is activated and inactivated with essentially the same kinetics in our extracts as in the first cell cycle of fertilized eggs, and the regulation of cyclin degradation is correctly preserved. Moreover, it is possible to freeze high-speed extracts which preserve the activation-inactivation mechanism of the histone kinase after thawing.

We find that a particulate subcellular fraction, which is even present at a significant level in the 250 000 f supernatants, is involved in the mechanism of kinase activation. This material can be eliminated by longer centrifugation, or by dilution and recentrifugation of the extract. We find that the p34^cdc2 kinase itself is not sedimented under any condition (Felix et al. 1989). Therefore, some component necessary for its activation must be particulate. Moreover, the cycle of kinase activation and inactivation can be reconstituted by recombining the pellet and the inactive supernatant. We call the particulate material the ‘Light Particulate Activator’ (LPA). ‘Light Particulate’, because it sediments only at very high speed in the crude extracts and does not enter very far into a sucrose gradient, and ‘Activator’ because LPA seems to be required to activate the kinase in a high-speed supernatant prepared from pre-mitotic eggs. LPA prepared from eggs homogenized just before metaphase cannot activate the histone Hl kinase in interphase supernatants. This indicates that other components - perhaps cyclins - must be present in the cytoplasm to support its activity. It is perhaps more surprising that LPA prepared from interphase eggs cannot activate the kinase in pre-mitotic supernatants. This is more difficult to understand, and seems to indicate that LPA is in a different state in interphase and in pre-mitotic eggs.

We also found that histone Hl kinase activated in 250 000g supernatants, which contain at least some LPA, when protein phosphorylation was completely inhibited by EDTA or 6-DMAP. This strongly suggests that phosphorylation of the histone Hl kinase is not necessary for its activation. It should be recalled that phosphatase inhibitors block activation of the kinase in these extracts, suggesting that activation of the kinase is brought about by dephosphorylation of some component of the system. Accordingly, LPA could provide either this hypothetical phosphatase or an activator of the phosphatase. Much more biochemical analysis will be required, however, before all this confusing phenomenology can be clearly and rationally explained. We think that the system described, here will be useful for such studies.