Plasmids encoding PKI(1-31), a specific inhibitor of cAMP-stimulated gene expression, inhibit the basal transcriptional activity of some but not all cAMP-regulated DNA response elements in JEG-3 cells.

Plasmids that encode a bioactive amino-terminal fragment of the heat-stable inhibitor of the cAMP-dependent protein kinase, PKI(1-31), were employed to characterize the role of this protein kinase in the control of transcriptional activity mediated by three DNA regulatory elements in the JEG-3 human placental cell line. The 5'-flanking sequence of the human collagenase gene contains the heptameric sequence, 5'-TGAGTCA-3', previously identified as a "phorbol ester" response element. Reporter genes containing either the intact 1.2-kilobase 5'-flanking sequence from the human collagenase gene or just the 7-base pair (bp) response element, when coupled to an enhancerless promoter, each exhibit both cAMP and phorbol ester-stimulated expression in JEG-3 cells. Cotransfection of either construct with plasmids encoding PKI(1-31) inhibits cAMP-stimulated but not basal- or phorbol ester-stimulated expression. Pretreatment of cells with phorbol ester for 1 or 2 days abrogates completely the response to rechallenge with phorbol ester but does not alter the basal expression of either construct; cAMP-stimulated expression, while modestly inhibited, remains vigorous. The 5'-flanking sequence of the human chorionic gonadotropin-alpha subunit (HCG alpha) gene has two copies of the sequence, 5'-TGACGTCA-3', contained in directly adjacent identical 18-bp segments, previously identified as a cAMP-response element. Reporter genes containing either the intact 1.5 kilobase of 5'-flanking sequence from the HCG alpha gene, or just the 36-bp tandem repeat cAMP response element, when coupled to an enhancerless promoter, both exhibit a vigorous cAMP stimulation of expression but no response to phorbol ester in JEG-3 cells. Cotransfection with plasmids encoding PKI(1-31) inhibits both basal and cAMP-stimulated expression in a parallel fashion. The 5'-flanking sequence of the human enkephalin gene mediates cAMP-stimulated expression of reporter genes in both JEG-3 and CV-1 cells. Plasmids encoding PKI(1-31) inhibit the expression that is stimulated by the addition of cAMP analogs in both cell lines; basal expression, however, is inhibited by PKI(1-31) only in the JEG-3 cell line and not in the CV-1 cells. These observations indicate that, in JEG-3 cells, PKI(1-31) is a specific inhibitor of kinase A-mediated gene transcription, but it does not modify kinase C-directed transcription.(ABSTRACT TRUNCATED AT 400 WORDS)

expression, however, is inhibited by PKI  only in the JEG-3 cell line and not in the CV-1 cells.
These observations indicate that, in JEG-3 cells, PKI(1-31) is a specific inhibitor of kinase A-mediated gene transcription, but it does not modify kinase Cdirected transcription. Moreover, in JEG-3 cells, expression mediated by the HCGa-derived cAMP response element depends on the activity of kinase A at basal as well as elevated levels of CAMP, whereas the cAMP regulation of the collagenase-derived phorbol ester response element, although equally dependent on kinase A (and independent of kinase C), is operative only after elevation of cAMP above basal levels. This represents a particularly clearcut instance of the participation of kinase A in cell regulation at basal levels of cAMP and illustrates the differential sensitivity of these two CAMP-responsive enhancers to the levels of the kinase A available in unstimulated JEG-3 cells. Such hierarchy in responsiveness to kinase A may enable the degree of activation of kinase A to define the qualitative as well as the quantitative nature of the cellular response.
The CAMP-dependent protein kinase is the only known intracellular effector of cAMP action in eukaryotic cells (1,2). The activity of the kinase is controlled primarily by the intracellular concentration of CAMP, whose synthesis is determined by the balance of stimulatory and inhibitory Gproteins acting to regulate adenylate cyclase activity, and whose degradation is catalyzed by a variety of cAMP phosphodiesterase enzymes (3). Isolated cells or tissues, examined in vitro in the absence of added stimulatory or inhibitory hormones, contain considerable amounts of intracellular cAMP that, if available to the kinase, would certainly promote some degree of activation. Nevertheless, many CAMP-regulated responses exhibit relatively little (or no) activity in the absence of stimulatory hormones (4). Moreover, upon addition of stimulatory hormones, CAMP-sensitive responses may exhibit substantial activation when the concomitant fractional elevation in overall cAMP content is very small relative to the basal content, often indistinguishable (4, 5). This phenomenon reflects, in part, the very strong positive cooperativity of cAMP binding to the regulatory subunits of the kinase, enabling cAMP increments over a narrow range to activate greatly kinase A' activity (6), and perhaps a compartmentation of cAMP as well. Not only does kinase A respond vigorously to small increments in total cAMP content above basal, but substantial activation of distal biologic responses often accompanies very small increases in measured kinase activity over the low levels found in the basal state (7). The basis for this sensitivity of the cell to small increases in kinase A activity over those present in the basal state is poorly understood but may in part be related to CAMPinduced inhibition of protein phosphatase activity mediated by kinase A phosphorylation of protein phosphatase inhibitor-1 (8). Thus, although the underlying mechanisms are incompletely elucidated, it is well accepted that important control of cell function is effected by slight elevations of cAMP levels and kinase A activity over those prevailing in the basal state.
Much greater uncertainty exists, however, as to the contribution of kinase A to the regulation of cell function at basal levels of CAMP. The idea that a "resting" or "basal" state prevails when cells or tissues are incubated in vitro in the absence of added ligands is oversimplified, in that isolated cells release a variety of autocrine factors that can modulate the activity of adenylate cyclase (e.g. adenosine, eicosanoids, etc.) (9), and serum also contains a variety of potential cyclase regulators. Nevertheless, if cyclase activity is fully inhibited, as e.g. by addition of a ligand that acts through Gi, a considerable pool of intracellular cAMP persists. It has been suggested that this basal cAMP is bound or sequestered in a form that cannot activate the CAMP-dependent protein kinase (5). Paradoxically, however, no quantitatively significant cellular CAMP-binding sites have been identified, other than the two sites on kinase A regulatory subunits (1,2,6). It appears more plausible that some fraction of kinase A activity is operative in the basal state, but for many substrates this phosphorylation is largely reversed by the action of protein phosphatases. Consistent with this view, for example, is the observation that addition to isolated rat adipocytes of the adenosine agonist, PIA, which acts to inhibit adenylate cyclase through Gi, depresses measured kinase A activity to very low levels (<5% of total) and inhibits basal lipolysis completely (9). Kinase A activity, however, probably persists even in the presence of PIA, inasmuch as lipolysis is stimulated by the addition of the protein phosphatase inhibitor okadeic acid to PIA-inhibited adipocytes (lo), implying that some phosphorylation of hormone-sensitive lipase continues even in the presence of sufficient PIA to inhibit cyclase fully.
In studying cAMP regulation of gene expression by examining the expression of transfected DNA, we have encountered a striking example of the operation of kinase A at basal levels of CAMP. Studies of a variety of cAMP and TPA-responsive genes have identified short DNA sequences that are necessary and sufficient to confer cAMP or TPA-regulated expression when ligated to reporter genes. The sequence motif of the cAMP response element in the somatostatin (11) and human chorionic gonadotropin-a (12)(13)(14) subunit genes is the octamer, 5'-TGACGTCA-3', whereas in the human enkephalin gene, the sequence, 5'-TGCGTCA-3', is observed within the cAMP response element (15). Of great interest, the consensus ' The abbreviations used are: kinase A, CAMP-dependent protein kinase; kinase C, protein kinase C; CRE, cAMP response element; TPA, tetradecanoyl phorbol acetate; bp, base pair; 8-Br-cAMP, 8bromoadenosine 3', 5'-cyclic monophosphate; CPT-CAMP, 8-(4-chlo-ropheny1thio)-adenosine 3', 5'-cyclic monophosphate; HCGa, human chorionic gonadotropin-a subunit; MES, 2-(N-morpholino)ethanesulfonic acid; RSV LTR, Rous sarcoma virus long terminal repeat; PIA, N6-[R-(-)-l-methyl-2-phenylethyl]adenosine; CAT, chloramphenicol acetyltransferase; kb, kilobase; CRP, CAMP-sensitive protein. sequence of the phorbol ester response element found in the human collagenase gene (16), the rat stromelysin gene (171, and the human metallothionein gene (18) is the related heptamer, 5'-TGA(C or G)TCA-3'. Previous studies have shown that in JEG-3 human placental cells, the expression of recombinant CAT reporter plasmids containing either the CAMP response element octamer or the phorbol ester response element heptamer both exhibit basal enhancer activity and stimulated expression on addition of CAMP, whereas only the phorbol ester response element heptamer but not the CAMP response element octamer is also stimulated by TPA (19,20). Different proteins in JEG-3 cell extracts were shown to bind to the cAMP response element octamer and the phorbol ester response heptamer (19,20). Herein we report that a plasmid that codes for a highly specific peptide inhibitor of the CAMPdependent protein kinase, PKI( 1-31), when cotransfected into JEG-3 cells with plasmids that contain the HCGa and enkephalin cAMP response element, gives a potent inhibition of their expression both in the presence and absence of added exogenous cAMP analogs. In contrast, the basal and phorbol ester-regulated expression mediated by the phorbol ester response element heptamer, derived from the collagenase gene, are unaffected by cotransfection with PKI, although the response of these constructs to added cAMP is strongly inhibited. These features appear to reflect a differential sensitivity of these two CAMP-responsive enhancer elements to the kinase A activity available in unstimulated JEG-3 cells.

EXPERIMENTAL PROCEDURES
Characterization of the PKZ(1-31) Peptide-PKI(1-31) (21) and Kemptide (LRRASLG) were synthesized on an Applied Biosystems automated solid-phase peptide synthesizer and purified by desalting followed by reversed-phase high pressure liquid chromatography using a Vydac Cla preparative column with a Waters Liquid Chromatograph. The solvent system was trifluoroacetic acid (0.1%) and acetonitrile. Peptide composition and mass were determined by amino acid analysis, and the peptides were sequenced on an Applied Biosystems gas-phase sequencer. The Ki of the PKI(1-31) as an inhibitor of the kinase A was determined using bovine protein kinase catalytic subunit (83 pg/ml) incubated with 160 p~ [y3'P]ATP, 66 mM MES, pH 6.8, 13 mM MgCls, 10 mM 8-mercaptoethanol in the presence of varying amounts of PKI  and at concentrations of Kemptide (LRRASLG) ranging from 2 to 15 PM, in a final volume of 60 p1. The rate of Kemptide phosphorylation was analyzed according to Henderson (22) and yielded a Ki of 4.3 nM. The specificity of synthetic PKI(1-31) for kinase A was evaluated by its ability to modify the activity of kinase C, the cGMP-dependent protein kinase, and the catalytic subunit of protein phosphatase 1 (Fig. 1).

19), nLXX-PKI(1-31) and
nLXX-[Gly'8~'9]PKI(1-31) (21) have been described previously. The collagenase-CAT expression vector (called -1200/+63-CAT in Ref. 16) was a gift of P. Herrlich (Karlsruhe, West Germany). Plasmids derived from nLXX exhibit poor replication in prokaryotic hosts and a tendency to grow as dimers; these multimeric forms appear to transfect eukaryotic cells and inhibit CAMP-stimulated expression less efficiently than the monomers (data not shown). To overcome these problems, the mutant and wild-type PKI(1-31) transcription units were recloned into the vector pBluescribe (Stratagene). The TLXX vector sequences were removed by digesting with NruI, which cuts in the 5'-end of the RSV-LTR promoter-enhancer, and with StuI, which cuts 3' of the site of poly(A) addition provided by sequences from pSV2 (for map see Ref. 21). pBluescribe was digested with KpnI and treated with T4 DNA polymerase to create a flush end, and then cut with SmaZ. NruI-to-StuI fragments encoding mutant and wild-type PKI(1-31) were cloned into the pBluescript vector to produce pBS-[Gly'8~'9]PKI(1-31) and pBS-PKI(1-31), respectively. These plasmids replicate as monomers and yield -10-fold more plasmid/liter of culture, compared with the TLXX-based plasmids. Plasmids were purified from alkaline or Triton X-100 lysates by banding on two sequential CsCl density gradients. Plasmid purity and DNA concentrations were analyzed using agarose gel electrophoresis with mass standards.  . Percent of initial activity was determined for CAMP-dependent protein kinase catalytic subunit, protein kinase C, CAMP-dependent protein kinase, and protein phosphatase-1 at various concentrations of PKI(1-31). CAMP-dependent protein kinase (0) was assayed as indicated under "Experimental Procedures," except that mixed histone (5 mg/ml) was utilized as substrate. Protein kinase C was assayed in the presence (A) and absence (A) of phosphatidyl serine, utilizing rat brain protein kinase C purified as indicated (57). Cyclic GMPdependent protein kinase activity was measured (58) utilizing mixed histone (1 mg/ml) as substrate in the presence (X) and absence (+) of 1 p~ cGMP. Bovine heart protein phosphatase-1 (0) was assayed by dephosphorylation of [32P]phosphorylase a (10 p~) as described (59).

Cell Culture, DNA Transfections, and CAT Assays-JEG-3 cells (HTB 36) and CV-1 cells (CCL 70)
were obtained from the ATCC and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Subconfluent cultures in 60-or 100-mm dishes were transfected (25) with 10-20 pg of plasmid DNA via the Capol method; 24 h later the media was removed and the cells were washed and refed. Regulators (8-Br-CAMP to 1 mM, CPT-CAMP to 0.1 mM, TPA to 100 nM, or 3-isobutyl-1-methylxanthine to 0.5 mM; see figure legends) were added for 15-18 h before cells were harvested for CAT assays, which were performed as described (13, 21, 26).
Each determination of CAT activity represented the average of two (Figs. 6 and 7 ) or three (Figs. [2][3][4][5]8,9) independent transfections. When CAT activity exceeded 40% conversion, extracts were diluted and reassayed in the linear range; activity was then reported as percent conversion times dilution factor, resulting in apparent conversion greater than 100% when a strong promoter-enhancer was used, such as in the plasmid pa1500 CAT (see Fig. 6). For protein kinase C desensitization experiments, the same protocol was followed, except that the cells were fed medium containing 100 nM TPA for 24 h prior to transfection, and TPA was included in the medium during PKI(1-31), which encode a mutant, inactive PKI (21). We have observed that vectors that carry the RSV LTR, such as the nLXX derivatives or pRSVpga1, can give substantial inhibition of the expression of cotransfected DNA. This "nonspecific" inhibition may be caused by competitive binding of limiting, general purpose transcription factors by the viral enhancer elements. Artifacts due to this property are avoided by including a fixed amount of RSV LTR DNA in each transfection and varying the proportion of vector containing active PKI(1-31) sequences.

RESULTS
The PKI( 1-31) expression plasmids were employed to provide a transfectable inhibitor of the endogenous cellular CAMP-dependent protein kinase. These plasmids contain the RSV LTR enhancer-promoter ligated to sequences coding for Extracts were prepared and assayed for CAT activity, which is expressed as percent conversion of chloramphenicol to acetylated chloramphenicol/h of assay (see "Experimental Procedures"). Note the linear scale on the ordinate. Error bars indicate one standard deviation ( n = 3). The numbers above the bars indicate the CAT activity (as % conversion) for the basal condition and the -fold induction for the regulator-treated conditions. The significance of the inductions, relative to basal conditions for mutant PKI-transfected cells, are indicated as follows: n.s., not significant; +, p < 0.02; **, p < 0.001. an initiator methionine and the amino-terminal 31 residues of the heat-stable protein kinase inhibitor from rabbit skeletal muscle. The corresponding synthetic peptide inhibits the catalytic subunit of the CAMP-dependent protein kinase with a Ki of 4 nM (data not shown), but does not alter the activity of kinase C, the cGMP-dependent protein kinase, or the catalytic subunit of protein phosphatase-1 at micromolar concentrations of peptide (Fig. 1). Previous studies have shown that *LXX-PKI( 1-31) inhibits the CAMP-stimulated expression of a cotransfected enkephalin-CAT reporter (pENKAT-12) in CV-1 and COS cells, but does not alter heavy metal stimulation of expression of a mouse metallothionein enhancer-promoter/growth hormone fusion gene (21).

PKI(l-31) Inhibits the CAMP-stimulated, but Not Basal or TPA-stimulated, Expression of Constructs Containing a Collagenase-deriued Phrbol Ester Response Element in JEG-3
Cells-The effect of PKI(1-31) on gene transcription was further evaluated by examining the expression in JEG-3 cells of collagenaseCAT, a plasmid that contains 1200 base pairs of 5"flanking sequence from the human collagenase gene ligated to coding sequences of the chloramphenicol acetyl transferase gene. The expression of this plasmid was originally characterized in HeLa cells and shown to be markedly stimulated by TPA (>lOO-fold) and slightly or not at all affected by cAMP (<2-fold) and was thus judged to contain a phorbol ester response element (Ref. 27). ' We find that expression of collagenaseCAT in JEG-3 cells is stimulated by both cAMP and TPA (Fig. 2), and both agents together provide a superadditive increment (Fig. 3). This feature permitted an examination of the effects of both PKI(1-31) inhibition of kinase A as well as TPA down-regulation of kinase C on basal and agonist-stimulated expression. Cotransfection into JEG-3 cells of collagenaseCAT with pBS-PKI(1-31) completely inhibited the CAMP-stimulated expression of collagenaseCAT as compared to cotransfection with the plas-mid pBS-[Gly1S,'g]PK1(1-31) (which codes for a peptide devoid of kinase inhibitory activity), but did not alter the basal expression or the response to added TPA. The selectivity of the inhibition induced by PKI  indicates clearly that this effect is not due to a nonspecific depression of DNA uptake or overall transcription in JEG-3 cells, but reflects a specific action of PKI(1-31) on the CAMP-dependent protein kinase. Of interest, exposure of JEG-3 cells to TPA (0.1 p M ) for 1 or 2 days prior to an acute challenge with TPA and/or cAMP abolished the response of collagenaseCAT to further addition of TPA but did not alter basal expression; the response to CAMP, although attenuated, remained substantial (Fig. 3). Thus, the basal expression of collagenaseCAT does not depend upon the catalytic activity of kinase A or kinase C. Moreover, the cAMP and TPA targets that modify collagenaseCAT expression exhibit no cross-regulation prior to the kinase step, e.g. TPA does not act on collagenaseCAT by increasing cAMP levels. Conversely, in the presence of both regulators, a synergistic increase in expression is observed, suggesting that the independent action of each kinase converges in a positively cooperative fashion after the phosphorylation reaction.
Deletion analysis of collagenaseCAT (27) showed that at least 80% of the 100-fold TPA-stimulated induction observed in HeLa cells resided in a segment near -72 with respect to the site of transcription initiation that contains the single phorbol ester response element consensus, 5'-TGAGTCA-3'. Deutsch and co-workers (19) have previously shown that this element alone can mediate responsiveness to both cAMP and TPA in JEG-3 cells; the plasmid co17a100CAT contains the 7-bp phorbol ester response element found in collagenaseCAT (5'-TGAGTCA-3') fused to the pal00 CAT. The pa100 CAT vector, which contains the fragment from -100 to +44 of the 5'-flanking region of the human chorionic gonadotrophin-a gene ligated to CAT, exhibits very low, barely detectable expression in JEG-3 cells (13, 19). Introduction of the heptamer (yielding co17a100CAT) boosts basal expression by 4fold (data not shown) and the addition of CAMP, TPA, and both agents together further increased expression of co17a100CAT by 2.9-, 2.4-, and 18.8-fold, respectively (Fig.  4). Cotransfection with pBS-PKI(1-31) inhibited the cAMP induction of co17a100CAT expression with little or no effect on the basal or TPA-stimulated expression; the synergistic response to the combination of regulators was reduced by pBS-PKI(1-31) cotransfection almost to the level of the response to TPA alone (77% inhibition; see Fig. 4). TPA pretreatment of JEG-3 cells transfected with co17a100CAT blocked the response to rechallenge with TPA and reduced the synergistic response observed on addition of both second messengers together to that seen with cAMP alone; TPA pretreatment had little or no effect on the basal or CAMPstimulated expression (Fig. 5). These results indicate that in JEG-3 cells, the single 7-bp DNA sequence, 5"TGAGTCA- Error bars indicate one standard deviation ( n = 3). The numbers above the bars indicate the CAT activity (as % conversion) for the basal condition and the -fold induction for the regulator-treated conditions. The significance of the inductions, relative to basal conditions of cells not pretreated with TPA, are indicated as follows: n.s., not significant; * , p < 0.01; * * , p < 0.001.

PKI(1-31) Effects o n Gene Expression
3', based on the phorbol ester response element from the 5' flank of the collagenase gene, can mediate basal expression, as well as a response to cAMP or to TPA. Neither kinase A nor kinase C participates in the basal expression mediated by this element. Moreover, the response to either cAMP or TPA is independent of the other stimulator, in that a specific blocking treatment can abolish one response with no effect on the other. When the action of kinase A and kinase C are concurrently expressed, however, a functionally synergistic interaction between their respective targets ensues.
PKI(1-31) Inhibits Both Basal and CAMP-stimulated Expression of HCGa-derived CAMP Response Element in JEG-3 Cells-The reporter gene pa1500 CAT was employed to examine the contribution of protein kinase A to the basal and CAMP-stimulated activity of a construct containing a "pure" cAMP response element. This plasmid consists of 1.5 kb of 5'-flanking sequence from the HCGa gene ligated to the chloramphenicol acetyltransferase gene. The HCGa 5"flanking region contains two copies of the sequence, 5'-TGACGTCA-3', situated within two identical 18-bp sequences arranged as adjacent direct repeats. This plasmid exhibits very high expression in JEG-3 placental cells in transient transfection experiments, both in the presence and absence of 8-Br-CAMP but is not stimulated by TPA (13, 24). JEG-3 cells were cotransfected with pa1500 CAT (5 pg) mixed with 10 pg of TLXX-derived DNA, containing various combinations of wild-type TLXX-PKI(1-31) or mutant TLXX-[Gly'8.'9]PKI(1-31). As the proportion of TLXX-PKI(1-31) was progressively increased, CAT activity was increasingly inhibited, both in the basal state (Fig. 6A) and in the presence of cAMP (Fig. 6B), and the inhibition proceeded in an essentially parallel fashion. This result indicates that kinase A is the target of the inhibition and that basal expression of pa1500 CAT in JEG-3 cells requires the activity of kinase A. An experiment was performed to ascertain if yet higher concentrations of PKI(1-31) could effect more complete inhibition. At these higher inhibitor levels, the dose-response relationships in the presence and absence of cAMP were still parallel (Fig. 6C). At 40 pg of cotransfected PKI, both basal and CAMP-stimulated transcription were inhibited -85% compared with activity when the vector alone was cotransfected. Vector alone ( TLXX) had significant inhibitory activity at these higher doses (i.e. 40 pg) that precluded extension of the dose-response analysis to ascertain whether higher doses of inhibitor plasmid might have caused a more complete inhibition.
To determine if the kinase A target that mediates basal expression acts on the same DNA sequence as that which mediates expression stimulated by addition of cAMP analogs, the effects of TLXX-PKI(1-31) on the expression of pa100+36 CAT were examined. This plasmid contains the two identical 18-bp DNA sequences, which are found as a tandem repeat in the 5' flank of the HCGa gene, ligated to pa100 CAT. Addition of the 36-bp tandem repeat to pa100 CAT yields a very marked increase in the basal expression and confers cAMP stimulatability on the CAT activity; overall expression in the presence and absence of cAMP is, however, only -10% that observed for pa1500 CAT, presumably due to the removal of other regulatory sequences (13,23). Cotransfection of pa100+36 CAT with TLXX-PKI(1-31) results in an inhibition of CAT expression as compared with cotransfection with a like amount of TLXX, both in the basal state and in the presence of 8-Br-CAMP (Fig. 7); the magnitude of the PKI inhibition of pa100+36 CAT is similar to that seen with the parent pa1500 CAT.
Thus, the same DNA sequence element (5"TGACGTCA- cates the amount of cotransfected wild-type PKI; where that value was less than 10 pg, the difference was made up with mutant PKI plasmid. Cells were refed 24 h after transfection with medium containing no additions (Fig. 6A, 0) or with 1 mM 8-Br-CAMP (Fig. 6B,  0 ) . Cells were harvested 18 h later and assayed for CAT activity. Each value in the depicted experiment represents the average of two independent transfections, reported as % conversion times dilution factor (see "Experimental Procedures"). C, cells were transfected as described in A and B, except that the total amount of PKI plasmid  for 18 h before harvest and assay. Error bars indicate one standard deviation ( n = 3). The numbers above the bars indicate the CAT activity (as % conversion) for the basal condition and the -fold induction for the regulator-treated conditions. CAT activity that resulted from each regulator treatment of mutant PKI-transfected cells was compared with the corresponding value for the wild-type PKI-transfected cells; the significance is indicated as follows: *, p < 0.01; **, p < 0.001.
CAT reporter gene substituted for the enkephalin coding region (15). Consequently, we examined the expression of pENKAT-12 in JEG-3 cells and found that its expression, although quantitatively much less than pa100+36 CAT, is also stimulated by cAMP but not TPA. Cotransfection of pENKAT-12 with pBS-PKI(1-31) produced inhibition of both basal and CAMP-stimulated expression as compared with cotransfection with pBS-[Gly's~'g]PKI(1-31) (Fig. 8). Thus, regulation of basal and CAMP-stimulated expression of pENKAT-12 in JEG-3 cells is quite similar to that observed for HCGa-derived plasmids. The expression of pENKAT-12 in CV-1 cells was re-examined, and the resistance of basal expression to suppression by cotransfection with aLXX-PKI(1-31), relative to aLXX-[Gly'8~'9]PKI(1-31) observed previously (21), was confirmed and extended (Fig. 9). Unfortunately, the expression of the HCGa-derived plasmids in CV-1 cells was too low to permit the reciprocal examination of their regulation in that cellular milieu. Thus, the basal expression of both the HCGa and enkephalin cAMP response elements in JEG-3 cells is inhibited by PKI .

DISCUSSION
The present study shows that in JEG-3 cells the catalytic activity of protein kinase A is required for the basal enhancer function of the cAMP response element, 5'-TGACGTCA-3', as well as for the augmentation of its enhancer function that occurs when cAMP levels rise above basal. By contrast, the element 5'-TGAGTCA-3', which, in JEG-3 cells, can act as a basal enhancer as well as mediate a response to either CAMP or TPA, does not require the activity of kinase A or kinase C for its basal enhancer function but only for the response to elevations of the respective second messengers above their basal levels. These conclusions are based on two inhibitory manipulations, down-regulation of protein kinase C by prolonged exposure of cells to TPA, and inhibition of protein kinase A by expression of plasmids that encode an aminoterminal 31 amino acid residue fragment of the heat-stable inhibitor of kinase A. The ability of the PKI(1-31)-encoding plasmids to selectively inhibit CAMP-stimulated, but not the basal or TPA-stimulated expression of collagenaseCAT and co17a100CAT (Figs. [2][3][4][5], is anticipated from the high affinity and narrow specificity for the kinase A demonstrated by the PKI(1-31) peptide (Fig. 1). Coupled with earlier results (21), these observations demonstrate the highly specific nature of these PKI(1-31)-encoding plasmids as transfectable inhibitors of the endogenous kinase A. It is necessary, however, to employ aLXX-PKI , in conjunction with the parent TLXX and aLXX-[Gly'8~'9]PKI(1-31) vectors (or the parallel pBS derivatives) so as to control for the proclivity of transfected RSV LTR DNA to inhibit nonspecifically the expression of cotransfected DNA, probably by interacting with endogenous transcriptional cofactors. In view of these considerations, and the data presented in Figs. 6 and 7, we conclude that in JEG-3 cells, the inhibition of basal expression of the HCGa-derived enhancer constructs produced by ?rLXX-PKI(1-31) is due to expression of an active protein kinase A inhibitor. A similar conclusion was suggested by Maurer and colleagues (28), who recently reported that cotransfection of a prolactin-CAT reporter construct with a PKI DNA was associated with inhibition of both basal and hormone-stimulated expression in GH, cells, analogous to the effects of PKI on the HCGa constructs we report here.
The ability of PKI to inhibit the basal enhancer function of the HCGa-derived (and proenkephalin) constructs, including the 36-bp synthetic version of the tandem-repeated octameric CRE, might simply reflect some slight activation of the kinase A engendered by the growth conditions or handling of the JEG-3 cells. The significant observation in our view is that this kinase A activity is insufficient to recruit the CAMPresponsive enhancer functions of the 5' flank of the human collagenase gene, or of the heptameric sequence, 5'-TGAGTCA, derived therefrom; this CAMP-dependent response becomes manifested only after kinase A activity is elevated beyond basal levels. These two types of CAMPresponsive enhancer elements are thus differentially sensitive to the kinase A activity expressed in unstimulated JEG-3 cells. Consistent with this view is the observation that the cAMP response of the TPA response element constructs was completely inhibited by PKI under the conditions of our experiments, while the response of the CRE constructs was inhibited only 80%. Based on the greater sensitivity of the CRE to the kinase A activity ambient in the basal state, it is not surprising that a more profound inhibition of kinase A is required to block completely the cAMP response mediated by this element, as compared to the TPA response element.
The molecular basis for the differential sensitivity of the CAMP-responsive enhancers to basal kinase A observed in the present studies is not yet known, but several factors deserve discussion. Kinase A exhibits considerable variation in K,,, for its physiologic protein substrates (29); thus, an obvious question is whether the same or different kinase A target substrates mediate kinase A-dependent regulation on both these enhancer elements. The fidelity with which the synthetic octamer and heptamer response elements reproduce the qualitative features of cAMP regulation exhibited by the much longer parental 5"flanking segments suggests that the synthetic response elements interact (directly or through other proteins) with at least some of the important kinase A targets. Previous studies have demonstrated that distinct proteins in JEG-3 cells bind to the closely related octameric and heptameric enhancer sequences (19). A protein or family of proteins of 38-45 kDa, named CREB (11) or ATF (30), bind preferentially to the octameric cAMP response element, 5'-TGACGTCA-3' (31) but not to the heptamer, 5'-TGA(C or G)TCA-3' (19). The cloning of cDNAs for two such proteins from human placenta and rat brain (32, 33) has been described; the recombinant placental protein exhibits the specific binding properties attributed to CREB. Both proteins contain at least one amino acid sequence, Arg-Arg-Pro-Ser-Tyr, that conforms to the favored substrate motif for the CAMP-dependent protein kinase (34). A purified CREB is, in fact, phosphorylated by kinase A i n uitro, and its transcriptional activity assayed in vitro is increased thereby (35). Thus, it appears that kinase A leads to gene activation mediated by the octamer element, 5'-TGACGTCA-3', through the phosphorylation of the protein CREB. The present data suggest that the kinase A activity available in JEG-3 cells in the absence of stimulatory ligands is sufficient to give partial phosphorylation and activation of CREB. The fraction of CREB protein phosphorylated in the basal state, however, cannot be inferred from these data; studies in intact cells will be necessary to define the fractional phosphorylation of CREB at basal and stimulated levels of CAMP. Moreover, the relationship between the concentration of phosphorylated CREB and the rate of gene transcription cannot yet be specified; available evidence suggests that this dependence will be nonlinear, in that the transcriptional activating function of phosphorylated CREB is expressed primarily after assembly into homo-(and possibly hetero-) dimers (35). The relationship between CREB phosphorylation and transcriptional activation will require the development of in uitro transcription systems utilizing purified or recombinant CREB, wherein the activity of kinase A and the concentration of other necessary elements can be precisely specified.
The kinase A targets that mediate the CAMP-stimulated expression driven by the heptameric element (5"TGACTCA-3') are not known. Based on competition experiments, CREB does not appear to bind directly to the heptamer in uitro (19).
The major polypeptide known to bind directly to the heptamer is AP-l/c-Jun (36)(37)(38), which acting with c-Fos in a heterodimer mediates the response of the heptamer to TPA (39-46). Consensus kinase A sites can be identified on c-Fos (47); thus, the interaction of c-Fos with AP-l/c-Jun might allow cAMP regulation via the heptamer. It is conceivable the CREB might associate with the heptamer in an indirect manner analogous to c-Fos, by forming a (thus far hypothetical) heterodimer with AP-l/c-Jun through the leucine zipper motif (48). In such a heterodimer, the intrinsic potency of phosphorylated CREB as a transcriptional activator may be quite different than in the CREB homodimer, providing a basis for differential sensitivity to kinase A despite a shared kinase A target (CREB) for both enhancer elements. It is interesting that cAMP and TPA regulation of expression via the heptamer are completely independent unless both second messengers are present together, whereupon a superadditive response is seen (Figs. 4 and 5). This pattern suggests that the cAMP and TPA targets that act on the heptamer do not interact with each other in a functionally significant way in JEG-3 cells until after the phosphorylation of both has occurred. Thereafter, however, the synergistic effects of cAMP and TPA on expression indicate that a positively cooperative interaction does occur. This could involve phosphorylation of a single transcriptional activator at different sites by kinase A or kinase C, wherein phosphorylation by either kinase activates independently, and simultaneous phosphorylation by both kinases yields a synergistic functional activation. Alternatively, the kinase A and kinase C targets may be entirely distinct proteins that acquire the ability to interact with each other in a functionally cooperative way only after both have undergone phosphorylation. The homo-and heterooligomeric associations between transcriptional regulatory proteins, before and after their phosphorylation, appear to be an important theme in this area of cell regulation (49)(50)(51).
In considering the mechanisms underlying the differential sensitivity of gene transcription to basal levels of kinase A, the discussion has focused primarily on the likelihood that different transcriptional activator protein complexes bind to these closely related enhancer sequences. It should be recalled, however, that catabolite repression in Escherichia coli, wherein the expression of several genes is regulated by ambient cAMP levels, also exhibits a hierarchy in sensitivity to ambient cAMP levels (52, 53), despite the participation of only a single CAMP-sensitive protein (CRP). In that wellcharacterized system (see Ref. 54 for review), the differential sensitivity must be conferred entirely by variations in the structure of the cis-acting elements. There, it is likely that a hierarchy of binding affinities between DNA and CRP/cAMP is established by variations in the sequence of the pentameric consensus element in the CRP-binding site, and by the presence of a second, occasionally symmetric, element 6-bp downstream of the pentamer. Other DNA structural motifs that may contribute to variations in the potency of bound CRP/ cAMP as a transcriptional activator include the location of the CRP-binding site relative to the transcriptional start site, the presence of contiguous binding sites for other proteins, and the presence of a second CRP-binding site. Some of these features are likely to be operative in the tuning of eukaryotic gene expression. It is already clear that sequences immediately adjacent to the octameric CRE (55) as well as the nature of the specific eukaryotic promoter (56) strongly influence the potency of the CAMP-responsive enhancer element in the HCGa gene.
All of the data discussed thus far apply to observations made in JEG-3 cells. This cell does provide a good milieu in which to study expression mediated by the HCGa-derived octameric cAMP response element, in that the endogenous HCGa gene is normally expressed in these cells. Nevertheless, the regulatory mechanisms operative in this cell may be quantitatively or qualitatively different in other cells. This is well illustrated by the expression of pENKAT-12 in JEG-3 uersus CV-1 cells (neither necessarily reflecting the normal milieu in which the human enkephalin gene is expressed). In both cells, expression is substantially stimulated by added cAMP and largely unresponsive to TPA, CAMP-stimulated expression is inhibited by PKI(1-31) in both cell lines. In JEG-3 cells, however, basal pENKAT-12 expression is strongly inhibited by PKI(1-31) (Fig. 8), whereas in CV-1 cells, basal expression of ENKAT-12 is entirely resistant to PKI(1-31) (Fig. 9). Whether this reflects different transcription factors mediating basal expression in the two cells or a different availability of kinase A at basal levels of cAMP (or both) is not known.
In conclusion, the data presented provide a very clearcut example of the participation of so-called basal levels of cAMP and kinase A activity in the regulation of cell function. Moreover, they illustrate the differential sensitivity of two CAMPresponsive enhancer sequences to a given intracellular level of kinase A activity, namely the basal level. The significant implication of this observation is that the qualitative nature of the cellular response can be controlled not only by activating different receptors or different signaling pathways, but by varying the intensity of stimulation at a single receptor or at a single signaling pathway.