cAMP-mediated growth inhibition in fibroblasts is not mediated via mitogen-activated protein (MAP) kinase (ERK) inhibition. cAMP-dependent protein kinase induces a temporal shift in growth factor-stimulated MAP kinases.

Growth factors stimulate fibroblast cell division by activating the recently identified mitogen-activated protein kinase (MAP kinase) signaling cascade. In contrast to our previous work (Kahan, K., Seuwen, K., Meloche, S. and Pouysségur, J. (1992) J. Biol. Chem. 267, 13369-13375), several reports have suggested that an elevation in intracellular cAMP blocks cell proliferation by attenuating MAP kinase activation. Hence we re-examined the effect of a long term increase in intracellular cAMP and therefore cAMP-dependent protein kinase (PKA) activation on the MAP kinase cascade in CCL39 fibroblasts. The concomitant addition of cAMP-elevating agents prostaglandin E, (PGE1) and IBMX did not inhibit the mitogen-mediated activation of p44 MAP kinase. However, a 5-min PGE1/IBMX pretreatment abolished the MAP kinase response, in a manner correlating with the extent of PKA activity. This inhibition was temporal in nature, and while modifying the time course of growth factor-mediated p44 MAP kinase, activation did not diminish the magnitude of the response. Thus the major peak of MAP kinase activity normally present 5 min after α-thrombin addition was now evident at 10 min in the presence of PGE1/IBMX. CCL39 cell proliferation is inhibited by elevated cAMP levels. Such an inhibition could reflect either a reduction in the number of cells entering the cell cycle or a delay in the time required to go through the cycle. Bromodeoxyuridine labeling experiments revealed that the cAMP-mediated inhibition of DNA synthesis in CCL39 cells was not due to a delay in S phase entry, but was due to a reduction in the number of cells entering S phase. Thus we conclude that although PKA activation may slightly modify the time course of MAP kinase activation in response to mitogens in CCL39 cells, the PKA-mediated inhibition of cell division occurs through modulation of an intracellular target, distinct from the p42/p44 MAP kinase cascade.

In normal untransformed fibroblast cell lines, it has long been appreciated (2) that a significant elevation of intracellular cAMP levels may potently inhibit cell growth and division. The presumed mediator of this effect is the cAMP-dependent pro-tein kinase (PKA), 1 a widely studied component of a signaling cascade that links extracellular signals to a variety of cellular functions (3). Until recently, the proposed target for the action of PKA in the arrest of cell growth was unknown. As elevated cAMP levels inhibit cell proliferation mediated by either Gprotein-coupled receptors or receptors with intrinsic tyrosine kinase activity, it was assumed that the target for PKA was downstream of the initial signaling events and likely to be a central player in the mitogenic response (2).
Virtually all known mitogens stimulate cell division through the activation of the recently described mitogen-activated protein kinase (MAP kinase) cascade (4 -6). In a widely studied model of growth factor signaling, the CCL39 fibroblast cell line, a 42-and 44-kDa MAP kinase have been identified (7). Both proteins form part of a signaling complex involving the sequential activation of Ras, a MAP kinase kinase kinase (MAPKKK or MEKK), a MAP kinase kinase (MAPKK or MEK), with MAP kinase being the final member of this cascade (6,8). It is now evident that the upstream MAP kinase activators form part of an ever increasing group of kinases that provide input signals into each of the MAP kinase family members including p42/44 MAP kinase and the highly homologous Jun/stress (8,9) and p38 kinases (10). However, an obligate step in the activation of p42 and p44 MAP kinases is the passage of the stimulatory signal from the Ras oncoprotein to the MAPKKK (for review, see Ref. 11). At present, three such MAPKKK have been identified in mammalian cells, including the homologous Raf-1 (also termed cRaf-1) and B-Raf serine/threonine kinases and MEK kinase, which shares little structural similarity to the former two proteins (14 -17). An additional Raf family member A-Raf, which has been shown to interact with Ras (18) is also likely to be a MAPKKK.
p44 MAP kinase is a cytoplasmic protein in resting CCL39 cells that becomes activated following its phosphorylation on threonine and tyrosine residues (7). After activation, p44 MAP kinase is subject to redistribution within the cell and can activate target proteins such as phospholipase A 2 , p90 rsk and p62 TCF in the plasma membrane, cytoplasm, and nucleus, respectively (6,7). The role of MAP kinase has been highlighted by experiments involving the expression of either dominant-negative MAP kinase mutants (p44 MAPK-TA), p44 MAP kinase antisense, or MAP kinase phosphatase (MKP-1), each of which results in a blockade of the mitogenic response to growth factors (12,13). Thus MAP kinase can be regarded an intracellular messenger and a central component of the mitogenic response. A member of the MAP kinase signaling cascade would thus present itself as a suitable target for the growth inhibitory function of PKA.
To this effect, several publications have recently shown that the ability of Ras to activate Raf-1 was impaired in cells treated with cAMP-elevating agents, leading to a loss in ability to activate MAP kinase (19 -23; see Ref. 24 for review). It would therefore appear that PKA prevents growth factor mediated cell division by the attenuation of an obligate step in the activation of MAP kinase.
We have previously demonstrated that increased cAMP levels in CCL39 cells, while blocking the mitogenic response (30), do not inhibit the ability to stimulate p44 MAP kinase (1). In an attempt to resolve this apparent contradiction, we have reexamined the effect of PKA activation on p44 MAP kinase stimulation. We show that long term elevation of cAMP levels induces a temporal shift in the activation of both MAP kinase and its activator MAPKK. However, the magnitude and duration of p44 MAP kinase activation is not altered. Thus we show that in fibroblasts, PKA-mediated growth inhibition is not mediated by a modulation of MAP kinase activation, but by the modification of at least one additional target.

Materials
Highly purified human ␣-thrombin and recombinant basic FGF were generous gifts of Dr. J. W. Fenton II (New York, State University of Health, Albany, NY) and Dr. D. Gospodarowicz (University of California, Medical Center, San Francisco, CA), respectively. [ 3 H]Adenine and [␥-32 P]ATP was obtained from Amersham Corp. Antiserum (␣IIcp42), which specifically immunoprecipitates p42 MAP kinase, was as described previously (25). Antisera (anti-ERK-1), which specifically immunoprecipitates p44 MAP kinase, was a kind gift from Dr. E. Van Obberghen (26). Antiserum Kelly #3, which immunoprecipitates both p42 and p44 MAP kinase, will be described in detail elsewhere. 2 Antiserum (Kawa), which specifically immunoprecipitates p45 MAPKK, was as described previously (27). Anti-BrdUrd IgG were from Amersham. All other materials were obtained from Sigma unless otherwise stated.

Methods
Cells and Culture Conditions-CCL39 cells are an established line of Chinese hamster lung fibroblasts (American Type Culture Collection). Cells expressing human muscarinic (m1) receptors (clone M1-81) were obtained as described previously (28). Cells lacking PKA activity (clone CCL39PKA Ϫ ) were obtained by transfecting CCL39 cells by the calcium phosphate method with the MT-REV(AB)-neo expression vector (kind gift of G. S. McKnight), which contains the coding region of the RI␣ subunit gene of PKA with three point mutations that prevent cAMP binding and activation of the enzyme (29). Transcription of the construct is directed by an inducible metallothionein promoter. However, the basal activity was sufficient to produce the kinase deficient phenotype. Following transfection, stable clones were selected in G418 (400 g/ml) and their PKA activity assessed regularly (see below) to ensure stability of the phenotype. 3 Cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Inc.), supplemented with 7.5% fetal calf serum, antibiotics (50 units/ml penicillin and 50 g/ml streptomycin), and 25 mM sodium bicarbonate at 37°C in an humid atmosphere (5% CO 2 , 95% air). Cells were serially passaged upon reaching confluence, and all experiments were performed on subculture passages 5-15.
To obtain quiescent cells arrested in the G 0 /G 1 phase of the cell cycle, confluent or preconfluent cultures were incubated for 24 h in serum-free medium.
Measurement of DNA Synthesis Reinitiation-Cells were seeded on glass coverslips at a density of 250,000 cells/well of a six-well plate and rendered quiescent just prior to confluence by serum removal overnight. Cells were then stimulated with the appropriate agonist and BrdUrd was added (10 M final). After 22 h or suitable time point, the incorpo-ration was stopped by aspiration of the media and several washes in PBS at 4°C. Cells were fixed with methanol/acetone (70:30) at Ϫ20°C for 15 min, followed by a further round of fixation in the presence of 4 M HCl for 10 min and then by extensive washes in PBS. Cells were incubated with 100 l of anti-BrdUrd IgG at a 1:100 dilution with nuclease in PBS, 1% bovine serum albumin for 1 h in a humid chamber. After extensive washes in PBS, cells were incubated for 30 min with fluorescein isothiocyanate anti-mouse IgG at a 1:100 dilution in PBS/ bovine serum albumin. A further round of rinsing in PBS was performed, and the coverslips were mounted. Cells were clearly visualized with a Nikon Diaphot microscope, ϫ 40 lens.
Determination of Intracellular cAMP-Confluent cultures in 12-well plates were rendered quiescent, and intracellular ATP pools were labeled by incubation in serum free DMEM containing [ 3 H]adenine (2 Ci/ml) for 24 h. The cells were washed three times with HEPESbuffered DMEM (pH 7.4) before being stimulated in HEPES-buffered DMEM (pH 7.4) for the indicated times in the presence of suitable agonist. The incubation was stopped by rapid aspiration of the media, followed by extracting the cells with ice-cold 5% trichloroacetic acid.
Protein Kinase A Activity-Confluent cultures in 12-well plates were rendered quiescent and were washed once with HEPES-buffered DMEM (pH 7.4) before being stimulated in HEPES-buffered DMEM (pH 7.4) for the indicated times in the presence of suitable agonist. Assays were stopped by rapid aspiration of media followed by rinsing with ice-cold PBS and addition of lysis buffer (50 mM Tris/HCl (pH 7.5), 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 40 mM ␤-glycerophosphate, 200 M sodium orthovanadate, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 M pepstatin A, 1% Triton X-100). Cells were then removed from the surface of the plate and centrifuged briefly at 12,000 ϫ g in a benchtop centrifuge to pellet nonlysed cells. The supernatants were then assayed for PKA activity exactly as described (31) in the presence of 100 M Leu-Arg-Arg-Ala-Ser-Leu-Gly (Kemptide) as substrate. Total PKA activity was determined in the presence of 10 M cAMP. PKA activity was defined as that sensitive to the inhibitor peptide PKA inhibitor peptide (1 M) (32).
Immune Complex Kinase Assay of MAP Kinase-Quiescent cells in 12-well plates were incubated in HEPES-buffered DMEM medium prior to stimulation with growth factors for the indicated times at 37°C. The cells were then washed twice with cold PBS and lysed in 0.5 ml of Triton X-100 lysis buffer (50 mM Tris/HCl (pH 7.5), 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 40 mM ␤-glycerophosphate, 200 M sodium orthovanadate, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 M pepstatin A, 1% Triton X-100) for 25 min at 4°C. After clarification by centrifugation at 12,000 ϫ g for 15 min at 4°C, the lysates were precleared for 1 h at 4°C with 1 l of normal rabbit serum and protein A-Sepharose (Pharmacia Biotech Inc.). The lysates were then incubated for 2 h at 4°C with 4 l of either antiserum ␣IIcp42 (p42 MAP kinase), anti-ERK-1 (p44 MAP kinase), or Kelly #3 (p42/p44 MAP kinase), preadsorbed to protein A-Sepharose beads. Immune complexes were collected by centrifugation and washed four times with Triton X-100 lysis buffer and once with kinase buffer (20 mM HEPES (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 10 mM p-nitrophenyl phosphate). Myelin basic protein kinase activity was assayed by resuspending the final pellet in a total volume of 40 l of kinase buffer containing 0.25 mg/ml myelin basic protein and 50 M [␥-32 P]ATP (specific activity ϭ 5500 cpm/pmol). Reactions were initiated with ATP and incubated at 30°C for 10 min. Assays were stopped by the addition of 40 l of 2 ϫ Laemmli's sample buffer. After heating to 95°C for 5 min, the samples were analyzed by SDS-gel electrophoresis on 10% gels. The gels were stained with Coomassie Blue, dried, and subjected to autoradiography. Phosphate incorporation was measured by excising substrate bands from the gel and counting the radioactivity in a liquid scintillation counter.
Immune Complex Assay of p45 MAPKK-Quiescent cells in 12-well plates were incubated in HEPES-buffered DMEM medium prior to stimulation with growth factors for the indicated times at 37°C. The cells were then washed twice with cold PBS and lysed in 0.5 ml of Triton X-100 lysis buffer (50 mM Tris/HCl (pH 7.5), 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 40 mM ␤-glycerophosphate, 200 M sodium orthovanadate, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 M pepstatin A, 1% Triton X-100) for 25 min at 4°C. After clarification by centrifugation at 12,000 ϫ g for 15 min at 4°C, the lysates were incubated for 2 h at 4°C with 3 l of antiserum (Kawa, Ref. 33) preadsorbed to protein A-Sepharose beads (30 l). The capacity of immunoprecipitated MAPKK to stimulate inactive MAPK was assessed in a kinase assay where epitope-tagged p44 MAPK was immunoprecipi-tated from growth-arrested cells that express high quantities of this protein (33). MAPKK Immune complexes were collected by centrifugation and washed four times with Triton X-100 lysis buffer and once with kinase buffer (20 mM HEPES (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 10 mM p-nitrophenyl phosphate). Myelin basic protein kinase activity was assayed exactly as described above.
Protein Determinations-Protein determinations were performed using the BCA protein assay kit (Pierce) with bovine serum albumin as standard.
Reproducibility of Data and Data Presentation-Immunoprecipitations were routinely performed as one point assays, as intra-assay variation was always found to be less than 12%. The data presented in figures are pooled from at least two individual experiments representative of at least three such experiments performed, which yielded qualitatively identical results.

Activation of MAP Kinase Is Not Inhibited by Concomitant
Addition of cAMP-elevating Agents-In mitogen-stimulated CCL39 hamster lung fibroblasts, a sustained intracellular elevation of cAMP serves as a potent inhibitor of cell cycle re-entry (30). We have previously published that the target(s) for this action does not lie in the p42/p44 MAP kinase-activating cascade (1), in marked contrast to what has since been reported by several groups (19 -23). Hence to re-address this question in detail, we examined the ability of a range of growth factors that function through activation of either receptors with intrinsic tyrosine kinase activity or receptors which modify G-proteinlinked signaling pathways, to activate the MAP kinase pathway in the presence of agents which elevate intracellular cAMP (Fig. 1). In CCL39 fibroblasts and derived clones such as M1-81 cells, which are transfected with and express M1 muscarinic receptors (28), MAP kinase activation in quiescent cells is rapid, peaking at approximately 5 min after agonist addition. As we have noted previously, the concomitant addition of an agent such as PGE 1 , which together with the phosphodiesterase inhibitor, IBMX, provokes a robust increase in the intracellular cAMP levels (1, 30), does not modify the stimulation of p44 MAP kinase elicited by three different mitogens: ␣-thrombin, FGF, or serum in M1-81 cells. A similar result was obtained with the non-mitogen carbachol (Fig. 1). Identical results were obtained using an antiserum specifically recognizing p42 MAP kinase and an additional antiserum that immunoprecipitates p42 and p44 MAP kinase (not shown). This experiment expands upon and re-inforces our previous work (1).
Inhibition of MAP Kinase Activation following Pretreatment of Cells with cAMP Elevating Agents-If the conditions are now changed, and the cells are pretreated with PGE 1 /IBMX for 5 min prior to agonist addition, then a different picture em-merges (Fig. 2). With a PGE 1 /IBMX pretreatment, the ability of ␣-thrombin, FGF, serum, and the non-mitogen carbachol to stimulate p44 MAP kinase (Fig. 2a) and its upstream activator, p45 MAP kinase kinase (Fig. 2B), is severely impaired. When PGE 1 is used without IBMX, then the inhibition, although significant, is less pronounced (results not shown). Hence, in M1-81 fibroblasts, an elevation in the intracellular cAMP levels may attenuate MAP kinase activation if conditions are appropriate.
Inhibition of p44 MAPK Activation Correlates with an Increase in Intracellular cAMP Levels and PKA Activity-In an attempt to define the conditions leading to inhibition of MAP kinase activation, we examined the time courses of each of cAMP production and PKA activation in response to PGE1/ IBMX challenge, together with the ability of PGE 1 /IBMX to block MAP kinase activation in response to ␣-thrombin (Fig. 3). Upon addition of PGE 1 /IBMX to quiescent M1-81 cells, the intracellular level of cAMP rapidly increases, with the maximal increase obtained after approximately 20 min of agonist addition. Intracellular cAMP remains at 80% of this maximal value for several hours following PGE 1 /IBMX addition (not shown). 2 min following addition of PGE 1 /IBMX the intracellular cAMP levels reach 70% of the maximal value obtained after 20 min of stimulation (Fig. 3A). This time course is paralleled by the increase in PKA elicited by PGE 1 /IBMX. Maximal activity is again obtained after 20 min of agonist addition. As for cAMP, the level of PKA activity remains elevated for several hours following addition of the agonist indicated. However, activation of PKA by PGE 1 /IBMX proceeds with a slight lag when compared with the time course of cAMP production. At 2 min, the PKA activity is at only 30% of its maximal stimulated activity (Fig. 3B). In an additional range of experiments, PGE 1 on its own produced a similar spectrum of activity, although in each case it was less potent than in the presence of IBMX and the duration of the response was significantly shorter (results not shown). Thus it is apparent that the activation status of PKA in M1-81 cells closely follows the level of intracellular cAMP, but with a slight delay. We then examined the time course of inhibition of MAP kinase activation (Fig. 3C) by PGE1/IBMX. As previously noted (Fig. 1), addition of PGE 1 /IBMX at the same time (time 0) or 1 min after addition of either FGF or ␣-thrombin fails to attenuate the ability of both agonists to stimulate MAP kinase. The ability of PGE 1 /IBMX to inhibit agonist-stimulated MAP kinase is time-dependent and starts to be evident only when PGE 1 /IBMX is added at least 2 min before growth factor addition. Inhibition is maximal after 5 min of pretreatment (Fig. 3C). Therefore the cAMP-mediated inhi-bition of MAP kinase activation closely correlates with the activation of PKA. However, it appears that the ability of PKA to inhibit growth factor-mediated MAP kinase activation depends on the prior activation of PKA. If the enzyme has not attained a significant level of activity before the addition of MAP kinase activating agent, then it is no longer possible to inhibit MAP kinase activation by PKA.
PKA Activation Does Not Completely Block p44 MAPK Activation, but Alters the Time Course of Activation-Having established that activation of PKA could indeed attenuate MAP kinase activation in M1-81 fibroblasts under appropriate conditions, we next examined if an increase in PKA activity completely blocked the ability to stimulate MAP kinase or if the effect was transient in nature. MAP kinase activity was determined in parental CCL39 cells in response to ␣-thrombin at a range of different time points. As noted previously, a 5-min pretreatment with PGE 1 /IBMX attenuated the ability of ␣-thrombin to stimulate MAP kinase, when viewed 5 min after ␣-thrombin addition (Fig. 4A). However, to our surprise, the ability of PGE 1 /IBMX to exert this inhibition was entirely transient in nature. It may be noted that the ␣-thrombinmediated activation of MAP kinase at 10 and 20 min is considerably greater than that seen in the absence of PGE 1 /IBMX pretreatment. (Fig. 4A, inset). Hence, PKA activation does not inhibit the activation of MAP kinase, but modifies its time course of activation. If the overall activity of MAP kinase is . Prior to addition of these growth factors, the cells were pretreated with PGE 1 (10 Ϫ6 M)/IBMX (1 mM) for the times indicated (from 10 min before growth factor addition to 1 min after growth factor addition). Cells were lysed 5 min after growth factor addition and p44 MAPK activity measured as described in the legend to Fig. 1. Data are presented as the percentage inhibition of FGF (Ⅺ)-or ␣-thrombin (q)stimulated p44 MAP kinase after a 5-min stimulatory period, in the absence of PGE 1 /IBMX. Data are mean values Ϯ S.E. from either two (A, B) or three (C) experiments. Error bars have been omitted from C for clarity, with error being less than 8% of the mean. compared, then it is actually greater following PKA activation. At later time points the PKA induced modification in the time course of MAP kinase activation is more difficult to quantify. A similar profile is obtained when serum is used as MAP kinase activating agent for each of CCL39 and Rat-1 fibroblasts and for rat vascular smooth muscle cells (Fig. 4B). In each case following PGE 1 /IBMX pretreatment, the peak of MAP kinase activity at 5 min in response to serum addition is completely absent. However, MAP kinase activity 1 h following serum addition is comparable for CCL39 cells, Rat-1 cells, and vascular smooth muscle cells, regardless of the presence of cAMPelevating agents (Fig. 4B). These results indicate that the target of PKA-mediated inhibition of cell division is unlikely to be the MAP kinase cascade.
Sustained PKA Activation Reduces the Number of Cells Entering S Phase-PKA-mediated inhibition of cell division is commonly revealed by blockade of the passage of growth factorstimulated quiescent cells into S phase of the cell cycle (2,30). In order to rigorously determine whether the PKA-mediated shift in the time course of MAP kinase activation could provoke a delay in the re-entry of M1-81 fibroblasts into the cell cycle and hence an inhibition of the rate of cell population growth, we examined the kinetics of cell cycle re-entry (Table I). Stimulation of quiescent cells with either serum (10%), ␣-thrombin (1 unit/ml), or FGF (25 ng/ml) resulted in the re-entry of 87, 47, or 42%, respectively, of the cell population into S phase of the cell cycle, as evidenced by incorporation of the thymidine analogue, BrdUrd (Table I). Carbachol, which is non-mitogenic for M1-81 cells, failed to significantly stimulate entry of M1-81 cells into S phase. Both PGE 1 /IBMX and PGE 1 on its own were potent inhibitors of BrdUrd incorporation, attenuating ␣-thrombin and FGF-mediated increases in BrdUrd incorporation by approximately 76 and 68%, respectively. As has been shown previously, both PGE 1 /IBMX and PGE 1 were slightly less potent inhibitors of serum-stimulated BrdUrd incorporation (57% inhibition). PGE 1 /IBMX and PGE 1 on their own were with-out effect on either basal or carbachol mediated BrdUrd incorporation.
PKA Activation Does Not Delay, but Blocks, S Phase Reentry-Following growth factor stimulation, quiescent CCL39 cells and their nontransformed M1-81 and ATR variants start to enter S phase and hence replicate their DNA at approximately 12 h. The maximal rate of DNA synthesis in such a population is at approximately 20 h. We examined a time course of serum-stimulated BrdUrd incorporation with or without PGE 1 /IBMX (Table II). When the percentage of serumstimulated cells entering S phase is compared at different times, from 12 to 20 h, it may be noted that the inhibition produced by pretreatment with PGE 1 /IBMX is very similar at each time point tested (approximately 50%). These data show that PKA activation does not delay the growth factor-mediated passage of cells from quiescence into S phase but rather reduced the number of cells entering S phase.
Inhibition of Growth Factor-stimulated p44 MAPK Is Mediated by PKA-Although additional targets for cAMP are beginning to be appreciated (34 -37), it is generally accepted that cAMP exerts virtually all its intracellular effects through the activation of PKA. Hence, to verify that the cAMP-mediated modification to the kinetics of MAP kinase activation was entirely mediated by PKA activation, we examined MAP kinase activation in CCL39 cells lacking PKA activity (38) (Fig. 5). In CCL39 PKA Ϫ cells, pretreatment with PGE 1 /IBMX was completely unable to modify the kinetics of ␣-thrombin-stimulated MAP kinase. Identical results were obtained when the activating agent was serum or FGF (not shown). As the ability of PGE 1 /IBMX to elevate intracellular cAMP is unimpaired in CCL39PKA Ϫ cells, 3 we conclude that the cAMP-mediated modification to the kinetics of MAP kinase activation as well as inhibition of DNA synthesis 3 is mediated through activation of PKA.

DISCUSSION
One of the original indications that cAMP, the first second messenger discovered (39), could act as a growth inhibitory agent was provided by Burk, who demonstrated that the growth of both normal and transformed baby hamster kidney cells was retarded by the addition of reagents that prevented intracellular degradation of cAMP (40). The universality of cAMP's growth inhibitory effect, at least for cells of fibroblastic origin (41), was later appreciated (2). However, a suitable target for cAMP, exerting its effect presumably through the activation of PKA, has until recently remained elusive. Recently, a number of groups have reported that one likely target for active PKA was the ubiquitous MAP kinase signaling cascade (19 -23). Specifically, PKA activation has been shown to attenuate an obligate step in the MAP kinase cascade, the activation of TABLE I Sustained PKA activation reduces the number of cells entering S phase M1-81 cells were seeded on glass coverslips at a density of 250,000 cells/well of a six well plate and rendered quiescent just prior to confluency by serum removal overnight. Cells were stimulated the next day with each of; ␣-thrombin (1 unit/ml), FGF (25 ng/ml), serum (10%), or carbachol (10 Ϫ3 M). Cells were also treated (or not) with either PGE 1 (10 Ϫ6 M) or PGE 1 (10 Ϫ6 M), 1 mM IBMX for 5 min prior to growth factor addition. Following a 22-h stimulatory period in the continual presence of BrdUrd (10 M final), the cells were fixed and percentage of cells which had incorporated BrdUrd counted as described under "Experimental Procedures." At least 150 cells were counted for each treatment. Data are from a single experiment representative of three performed.   II  PKA activation does not delay, but blocks, S phase re-entry  M1-81 cells seeded as in the legend to table 1 were treated (or not) with PGE 1 (10 Ϫ6 M, 1 mM IBMX) for 5 min prior to serum (10%) addition. Cells were given a "pulse" of BrdUrd (10 M final) during the last 2 h of incubation and the experiment was terminated after either 14, 16, 18, or 20 h of serum stimulation. Cells were then fixed and percentage of cells in the whole population that had incorporated BrdUrd determined as described in the legend to Table I the MAPKKK, Raf-1 by the Ras oncoprotein (5,11,42). In this paper, we show that although activation of PKA inhibits the proliferation of CCL39 cells (Ref. 30 and Tables I and II), the target of this inhibition is not the classical MAP kinase cascade, since the activation of both p42/p44 MAP kinase is not inhibited by the presence of elevated intracellular cAMP, but has an activation delay of approximately 5-10 min (Fig. 4) The mechanism of Raf-1 activation is complex and incompletely understood. Ras appears to be required to "re-direct" Raf-1 to the plasma membrane (43), where Raf-1 is activated by a Ras-independent phosphorylation event(s) involving both serine/threonine and tyrosine kinases (44,45). Inactivation of Raf-1 requires the action of either a serine/threonine and/or a tyrosine phosphatase (46) and may be prevented by members of the 14-3-3 family of proteins that have previously been shown to bind to both active and inactive Raf-1 (47)(48)(49)(50). Following inactivation, Raf-1 may now return to the cytosol for a new round of activation. This cyclical process has previously been outlined (24) and has obvious implications concerning the mechanism of inhibition of Raf-1 by PKA. If PKA specifically inhibits an early step in the Raf activation process, then we would expect PKA to be unable to inhibit an already active Raf molecule. This situation is entirely supported by our data (Fig.  2) where PKA has to be fully active (i.e. preaddition of cAMPelevating agents) before addition of growth factor. Otherwise, MAP kinase activation proceeds as normal (Fig. 1).
The mechanism of inhibition of Raf-1 by PKA has recently been suggested to involve two separate events (51). In addition to weakening the interaction of Raf-1 with Ras (20), thus preventing the initial translocation of Raf to the plasma membrane, PKA phosphorylates the Raf-1 kinase domain, inhibiting autophosphorylation of Raf-1 kinase (20,21,52). The latter explains the ability of PKA to inhibit v-Raf (which does not require Ras to be functionally active) and hence the ability of elevated intracellular cAMP levels to revert v-Raf-transformed NIH 3T3 cells (53). At present, of the three MAPKKK enzymes that have been identified, Raf-1, B-Raf, and MEK kinase, all appear to be inhibited by PKA activation (52,54,55). A fourth member of the Raf oncoprotein family, A-Raf, exists which is likely to be a MAPKKK, although data for this functional effect are currently lacking (15). Interestingly, MEK kinase appears to be one member of what may be a large family of protein kinases as at least four genes putatively coding for MEK kinase homologues exist. In addition, a significant number of reports exist that suggest that additional MAPKKK proteins are expressed in both mammalian cells (56 -58) and in Xenopus oocytes (59). A surprising recent finding by Moscat and colleagues is that a protein kinase C (PKC) family member, PKC , may also function as a MAPKKK (68), again increasing the complexity of the system.
The simplest explanation to account for the ability of PKA to modify the time course, but not to inhibit the activation of MAP kinase in CCL39 cells, is that in this cell line, although the presently identified MAPKKK are inhibited by PKA, there exists an as yet unidentified MAPKKK (see above) that is not inhibited by PKA. Such a PKA-insensitive MAPKKK expressed by CCL39 cells would have to have a time course of activation considerably slower than than of the identified PKA sensitive MAPKKK. However, one must add the caveat that the time course of activation of a MAPKKK insensitive to PKA would be identical, irrespective of whether Ras is able to bind to Raf family members which are sensitive to PKA. An additional possibility is that in CCL39 cells, although the presently identified MAPKKK are inhibited by PKA, there exists an as yet unidentified MAPKKK that is stimulated by PKA. This may indeed be the case in PC12 cells, where elevated cAMP has been reported to be (26) or not be able (54) to generate a significant stimulation of MAP kinase. However, this is unlikely to be the case for CCL39 cells as elevated levels of cAMP do not on their own stimulate MAP kinase ( Figs. 1 and 2).
In mammalian cells, virtually all of the cAMP effects can be attributed to the activation of PKA. However, protein kinase A-dependent cAMP responses have been described. The ability of cAMP to inhibit the GLUT4 but not the GLUT1 glucose transporter may be mediated by direct binding of cAMP to GLUT4 (34). In addition, a range of different ion channels, some of which contain a putative cyclic nucleotide binding site (60), may be modulated in a stimulatory manner by direct binding of cAMP (35)(36)(37). In CCL39 cells, it would appear that the ability of elevated cAMP levels to modify the kinetics of MAP kinase activation are entirely due to the activation of PKA, as the activation of MAP kinase in CCL39 cells lacking PKA activity is completely insensitive to elevated cAMP (Fig. 5).
To the best of our knowledge, in all cell types studied, activation of MAP kinase is rapid, with peak activity occurring approximately 5 min and no later than 10 min after addition of agonist. In contrast to the situation in primary dog thyrocytes, where TSH does not stimulate MAP kinase (61), in primary cultures of human thyroid follicles, one may note an interesting parallel between the time course of growth factor-stimulated MAP kinase in the presence of activated PKA in CCL39 cells and the time course of MAP kinase activation shown in response to TSH (62). In this system, peak MAP kinase activity occurs at approximately 20 min after addition of TSH. Interestingly, TSH also stimulates production of cAMP and hence activation of PKA in thyrocytes (63). One may speculate that the reason for the apparently "slow" time course of MAP kinase activation in human thyroid cells in response to TSH is due to the inhibition of PKA sensitive MAPKKK enzymes, with the signal being propagated by PKA insensitive MAPKKK. In addition, Al-Alawi and colleagues have recently shown that although elevation of cAMP in thyroid cells may attenuate Raf-1 activation, TSH induced mitogenesis still proceeds in a Ras-dependent manner (63,64). Hence it is highly probable TSH stimulates MAP kinase and cell division in certain species of thyroid cells via an unidentified PKA insensitive MAPKKK.
We had hoped that this report would clarify the situation regarding PKA and MAP kinase, at least for cells of fibroblastic origin (CCL39 and Rat-1 cells). However, reports in the literature show that the activation of MAP kinase can be both stimulated (26,65) and inhibited (19 -23) by an elevation of intracellular cAMP. Additionally, a recent report shows that neuropeptide modulation of voltage sensitive K ϩ currents in the body-wall neuromuscular junction of Drosophila larvae requires concomitant elevation of cAMP and activation of Raf (66). A variety of scenarios are therefore available. However, our data is supported from recently published work in PC12 cells (54), where Vaillancourt and colleagues have shown that elevated cAMP levels, although inhibiting the growth factor mediated activation of B-Raf, have no apparent effect on the activation of MAPKK and MAP kinase. As for CCL39 cells, the simplest explanation for this result is that additional MAP-KKK exist in PC12 cells which are insensitive to PKA (54). Previous studies in the literature are hampered by a total lack of detailed kinetics of MAP kinase activation, hence it is not possible to conclude whether or not previous results are in discordance with our own (20 -23). One possible exception is A14 cells where cAMP appears to inhibit MAP kinase in a protracted manner (19).
The two major points raised by this report are: what additional Ras-dependent PKA-insensitive MAPKKK are expressed in fibroblasts and what is (are) the real target(s) for the PKAmediated inhibition of cell division? Experiments to define the former are in progress, and our attempts to define the latter are centered on the the cell cycle machinery. An obligate step for the mitogen-stimulated passage of quiescent cells through G 1 to S phase appears to be the induction of cyclin D1. 4 Mitogenstimulated induction of cyclin D1 is blocked by the long term elevation of cAMP levels in CCL39 cells. Hence, bypassing this blockade may be a suitable way to overcome the growth inhibitory effect of long term PKA activation. Current experiments are seeking to address this possibility. As has been discussed (see above), the MAP kinase family of serine/threonine kinases can be subdivided into three distinct groups; the classical p42/ p44 MAP kinases, the Jun kinase/stress-activated kinase family, and the p38/osmotically activated kinases (67). The role of the latter two kinase families in mitogenic signaling remains to be determined. It will, thus, be of interest to discover what effect, if any, activation of PKA has on the latter two recently identified MAP kinase family members.