Cyclic AMP stimulates somatostatin gene transcription by phosphorylation of CREB at serine 133

Cell, Vol. 59, 675-660,

November

17, 1969, Copyright

0 1969 by Cell Press

Cyclic AMP Stimulates Somatostatin Gene Transcription by Phosphorylation of CREB at Serine 133 Gustav0 A. Gonzalez and Marc Ft. Montminy The Clayton Foundation Laboratories for Peptide Biology The Salk Institute 10010 North Torrey Pines Road La Jolla, California 92037

recognition site (R-R-P-S) from amino acids 130-133. Indeed, CREB is efficiently phosphorylated at Ser-133 by the C-subunit in vitro, suggesting that this site may be critical for CAMP responsiveness (Yamamoto et al., 1988). In this report we show that CREB is phosphorylated in vivo at Ser-133 in response to forskolin and that phosphorylation of Ser-133 is critical to the activation of gene transcription by CAMP

Summary Results In this paper, we demonstrate that phosphorylation of CREB at Ser-133 is induced 64old in vivo, following treatment of PC12 cells with forskolin. By contrast, no such induction was observed in the kinase A-deficient PC12 line A126lB2 (A126). Using F9 terstocarcinoma cells, which are unresponsive to CAMP, we initiated a series of transient expression experiments to establish a causal link between phosphorylation of CREB and frans-activation of CAMP-responsive genes. Inactivating the kinase A phosphorylation site by in vitro mutagenesis of the cloned CREB cDNA at Ser-133 completely abolished CREB transcriptional activity. As CREB mutants containing acidic residues in place of the Ser-133 phosphoacceptor were also transcriptionally inactive, these results suggest that phosphorylation of CREB may stimulate transcription by a mechanism other than by simply providing negative charge.

Cyclic AMP (CAMP) mediates the hormonal stimulation of a variety of eukaryotic genes through a conserved CAMP response element (CRE) (Montminy et al., 1986; Comb et al., 1986). Transcriptional induction by CAMP is rapid, peaking at 30 min and declining gradually over 24 hr (Sasaki et al., 1984; Lewis et al., 1987). This burst in transcription is resistant to inhibitors of protein synthesis, suggesting that CAMP may stimulate gene expression by inducing the covalent modification rather than de novo synthesis of specific nuclear factors. Since all of the known cellular effects of CAMP occur via the catalytic subunit (C-subunit) of CAMP-dependent protein kinase (kinase A), it appears likely that this enzyme mediates the phosphorylation of factors that are critical for the transcriptional response. Kinase A-deficient cell lines, for example, are unable to stimulate somatostatin gene transcription in response to forskolin (Montminy et al., 1988). Furthermore, microinjection of the C-subunit into cells can directly activate CRE-dependent transcription without simultaneous addition of CAMP (Riabowol et al., 1988). We have previously characterized and isolated cDNAs for a 43 kd nuclear CRE binding protein, CREB, which binds to and stimulates transcription of the somatostatin gene in vitro (Montminy and Bilezikjian, 1987; Yamamoto et al., 1988; Gonzalez et al., 1989). Sequence analysis of the CREB cDNA predicts a single kinase A consensus

and Discussion

Forskolin stimulates transient expression of the somatostatin-CAT fusion gene A( - 71) lo- to 20-fold in PC12 cells, but has no effect in kinase A-deficient Al28-162 cells (Montminy et al., 1988). To test the hypothesis that CAMP enhances transcription of the somatostatin gene by inducing the phosphorylation of CREB through kinase A, we labeled PC12 and Al26 cells with inorganic 32P and then treated them with forskolin or ethanol vehicle for 20 min. Cells were harvested in standard SDS lysis buffer and immunoprecipitated with W39 CREB antiserum (Gonzalez et al., 1989). Following SDS-PAGE, the 32P-labeled 43 kd bands corresponding to CREB were eluted and digested with trypsin. Two-dimensional phosphopeptide maps of control and forskolin-treated samples showed a single 32P-labeled spot (Figure 1A) migrating at the same relative position as CREB tryptic peptide W51 (Gonzalez et al., 1989). Using gas phase microsequencing, we had previously determined that the W51 peptide contains the kinase A phosphorylation motif (R-R-P-S). Forskolin treatment caused a 6-fold increase in the phosphorylation of this peptide in PC12 cells, but had no effect in Al28-lB2 cells. The increased phosphorylation in PC12 cells does not appear to arise from de novo synthesis of CREB, as incorporation of [35S]methionine into the protein was unaffected by forskolin treatment from 30 min (data not shown) up to 4 hr (Figure 18). To determine whether CREB is both necessary and sufficient to stimulate transcription of the somatostatin gene in response to CAMP we initiated a series of transient expression experiments. Because PC12 cells have high levels of endogenous CREB activity (Montminy and Bilezikjian, 1987; Yamamoto et al., 1988) we did not consider these cells suitable for mutagenesis studies. F9 teratocarcinema cells, in contrast, are unresponsive to CAMP, becoming inducible only after differentiation with retinoic acid (Rickles et al., 1989). Kinase A (Plet et al., 1982) and CREB levels (M. M., unpublished data) are reduced 3- to !&fold in undifferentiated versus differentiated F9 cells, suggesting that a deficiency in both components may account for the lack of CAMP responsiveness. Consistently, A( - 71) CAT activity was not stimulated by forskolin after transfection of F9 cells with this reporter plasmid (data not shown). When cotransfected with a Rous sarcoma virus (RSV) plasmid expressing CREB or a metallothionein vector expressing the C-subunit of kinase A (Mellon et al., 1989) A( - 71) activity was only modestly induced by each

Cdl

676

PC12

Al26

8. (Mr)

0

2

4

10575-

l o-

50-

CONTROL

*

3927-

a-

;

FORSKOLIN

:

oFORSKOLIN

l-

Figure

1. Phosphorylation

and Synthesis

of CREB

in PC12 Cells

(A) Phosphorylation of CREB in response to forskolin treatment of PC12 and protein after immunoprecipitation with W39 CREB antiserum. Arrow points tryptic peptide W51. (B) Biosynthesis of CREB in PC12 cells in response to forskolin after 0, 2, = relative molecular weight (in thousands). Cells were treated with forskolin mCi/ml) for 15 min. Cell extracts were immunoprecipitated with affinity-purified

(Figure 2). Cotransfection of both the C-subunit and CREB genes, however, caused a dramatic 200-fold induction in A( - 71) CAT activity, suggesting that both CREB and C-sub unit activities are critical to the induction of gene transcription. No such induction was observed when CREB was cotransfected with an inactive mutant C-subunit plasmid pCaK72M (gift of M. Uhler) encoding a lysine to methionine substitution at position 72 (data not shown). Furthermore, C-subunit and CREB activities appear to be specific for the CRE sequence, as neither CREB nor the C-subunit, alone or in combination, could stimulate CAT expression when coexpressed with an RSV-CAT reporter gene (Figure 2). To establish a causal link between C-subunit expression and CREB frans-activation, we prepared mutant forms of CREB that were defective in the kinase A phosphorylation motif by in vitro mutagenesis of the cloned CREB cDNA (Figure 3a). After insertion into RSV vectors, each mutant was evaluated for frans-activating potential in F9 cells by cotransfection with the C-subunit expression plasmid. Mutant Ml contains a conservative serine to alanine substitution at position 133, which destroys the kinase A phosphorylation site while maintaining charge balance (Figure 3a). Combined transcription-translation of the Ml cDNA in vitro showed an immunoreactive protein that migrated identically to the wild-type CREB protein on SDS-PAGE (Figure 3b). When tested in F9 cells, however,

Al26 cells. Two-dimensional phosphotryptic maps of 32P-labeled CREB to the 32P-labeled spot, which migrates at the same position as CREB and 4 hr of forskolin treatment. Arrow points to %-labeled CREB. Mr (10 mM) for times indicated and then labeled with IsrS]methionine (0.1 W39 CREB antiserum. Samples were then resolved by SDS-PAGE.

Ml was completely unable to activate transcription of the somatostatin gene, suggesting that Ser-133 was indeed critical to the trans-activation of CREB by the C-subunit (Figure 4a). To test whether phosphorylation of Ser-133 activates transcription simply by providing negative charge at this residue, we constructed a mutant, M2, containing a Ser133 to Asp-133 substitution (Figure 3a). As with Ml, the M2 cDNA encoded a 43 kd immunoreactive CREB product (Figure 3b). When coexpressed with the C-subunit expression plasmid in F9 cells, however, the M2 protein was ineffective in stimulating A( - 71) CAT activity (Figure 4a). A CREB mutant containing a Ser-133 to Glu-133 substitution was similarly inactive (G. G., unpublished data). These results suggest that phosphorylation of CREB at Ser-133 may activate transcription by a mechanism other than by providing negative charge. In the event that the Ml and M2 mutants were inactive owing to reduced expression from their RSV promoters, we performed RNAase protection assays on total RNA from F9 cells transfected with wild-type or mutant (Ml and M2) RSV-CREB plasmids. Using an 800 base CREB antisense RNA probe, we observed a 745 nucleotide RNAase-protected fragment on denaturing polyacrylamide gels, the reduced size of which was was due to RNAase digestion of plasmid polylinker sequences. The 745 base fragment was present at comparable levels in samples

cAMPStimulated 677

Phosphorylation

CRE-CAT

r

%

conversion

of CREB

C

pK-A

0.2

1.9

CREB

RSV-CAT pK-A +‘I CREB C

1 1.3

40.6 1 5.5

pK-A

) 2.4

pK-A +’ CREB CREB

1 1.6 1 1.6

200 -

C

pKA

C

pKA

CREB

pKA+ CREB

Figure 2. Effect of CREB and Kinase A on A( - 71) CAT and RSWAT Espression in Undifferentiated F9 Teratocarcinoma Cells Cells were transfected with either the A( - 71) CAT or RSV-CAT reporter gene plus C-subunit (pKA), CAEB (CREB). or both expression plasmids (pKd + CREB). Control (C) = F9 cells transfected with reporter gene alone. CAT activity is expressed as percent conversion shown below each lane. The graph below shows CAT activity relative to control values (R.A.).

a.

from wild-type, Ml, and M2 transfections, but was not observed in RNA from cells transfected with a nonexpressing pGEM-CREB vector (Figure 4b). To demonstrate that the inactivity of Ml and M2 plasmids was not due to preferential degradation of the mutant CREB proteins, we performed immunoprecipitation experiments (Figure 4~). Cells transfected with wild-type, Ml, or M2 plasmids expressed comparable levels of immunoreactive CREB products. Furthermore, indirect immunofluorescence microscopy revealed that both wild-type and mutant (Ml and M2) CREB products were appropriately targeted to nuclei of transfected F9 cells (Figure 4d). The inability of the CREB mutants Ml and M2 to stimulate transcription, therefore, would appear to arise from the deletion of the Ser-133 phosphoacceptor site and not from underexpression of the transfected plasmids. The presence of two basic residues (Arg-135 and Lys136) C-terminal to Ser-133 forms a potential kinase C phosphorylation motif (S-Y-R-K). Indeed, Ser-133 can also be phosphorylated by kinase C in vitro (Yamamoto et al., 1988) prompting us to investigate whether the C-subunit can activate CREB when the kinase C motif is ablated. Substituting Arg-135 and Lys-136 with Met-135 and Glu136 in M3 (Figure 3a) destroyed the consensus kinase C motif. In vitro translation of M3 RNA (Figure 3b) showed the expected 43 kd band and a smaller 30 kd immunoreactive product, which is consistent with alternate translational initiation at Met-135 Furthermore, immunoprecipitation and immunofluorescence assays (Figures 4c and 4d)

b. (W

C. Ml ?

TAC AGG AAA. Tyr Arq Lys.

M1

.AGG AGG CCT GCC .Arg Arq Pro -Ala

TAC AGG AAA. Tyr Arg Lys.

PI.2

.AGG AGG CCT GAC .Arq Arg Pro Asp -

TAC AGA AAA. Tyr Arg Lys.

.AGG AGG CCT TCC Arg Arq Pro Ser*

TAC ATG GAA. Tyr tht Glu.

Type

M2 I

M3

w-r

c Lei”

P -.’ ,

pK-C

pK-A .AGG AGG CCT TCC .Arg Arg Pro Ser*

Wild

“a

. 105 75-

39.

27-

17-

Figure

3. Characterization

of CREB

Point Mutants

(a) Nucleotides and corresponding amino acid sequences of point mutants Ml, M2, and M3 compared with wild-type (WT) CREB cDNA near the Ser-133 phosphoacceptor site. Consensus kinase A (pK-A) and kinase C (pK-C) phosphorylation sites are overlined and labeled. Positions of amino acids are shown in parentheses. Serine phosphoacceptors are indicated by asterisks. Mutated residues are underlined and in bold. (b) lmmunoprecipitation of %i-labeled wild-type and mutant CREB proteins expressed in vitro. Mr = molecular weight standards (in thousands). Ml, M2, M3, and WT lanes correspond to mutant and wild-type CREB proteins. (c) Effect of C-subunit phosphorylation on CRE binding activity of bacterial CREB fusion protein in vitro. Gel shift assay of extract using 32P-labeled double-stranded CRE oligonucleotide probes. I and II = putative dimer and monomer forms of CREB, respectively. C = control; P = CREB phosphorylated by the C-subunit in vitro.

Cdl 676

a.

Mt

WT

Ml

M2

M3

C. W,)

%

conversion

0.3 ) 5.7

1 0.2

1 0.2

RSV Ml ilmums-

M2

M3 Mt anlnllR-

1 6.5

3927-

Mt

Wf

Ml

M2

b.

s Mr

Wl

Ml

M2

T7

*

600: 622 527 -

Figure

4. Analysis

of CREB

Mutants

at the Kinase

A Phosphorylation

Site

(a) Analysis of wild-type and mutant CREB proteins by transient expression assay. Representative CAT assays of F9 cells transfected with wild-type (WT) or mutant (Mt, Ml, M2, and M3) RSV-CREB plasmids. The mutant Mt, which contains a nonsense mutation at residue 131 and does not express immunoprecipitable CREB protein, was used to control for nonspecific effects of the RSV-CREB plasmid on transcription. Each CREB plasmid was cotransfected with A( - 71) CAT reporter, C-subunit vector MtC, and RSV-6-galactosidase as described in Figure 2. Percent conversion is indicated and relative activity (R.A.) is shown graphically below. Assays were normalized for 6-galactosidase activity. Each mutant was tested in at least three separate assays. (b) RNAase protection of total RNA from F9 cells transfected with wild-type (WT) or mutant (Ml, M2) CREB plasmids. Mr = molecular weight standards (in fragment. T7 = RNA from cells transfected with pGEM-CREB plasmid DNA. tRNA .~ bases). , Arrow ooints to 745 nucleotide RNAase-protected = control assay using tRNA only. To prepare 3zP-labeled antisense CREB RNA probe, pGEM-CREB plasmid, linearized with Aatll, was transcribed in vitro (Chomczynski and Sacchi, 1967) using T7 RNA polymerase. The 600 nucleotide probe was resolved on a 6% urea-polyacrylamide gel, removed, and electroeluted. Total RNA was prepared from transfected cells (Chomczynski and Sacchi, 1967) and treated with DNAase I to remove transfected plasmid DNAs. Equal amounts of RNA (6 ng, verified by agarose gel electrophoresis) were analyzed by RNAase protection of the CREB RNA probe according to Zinn et al. (1963). (c) lmmunoprecipitation of as.9labeled CREB products from F9 cells transfected with CREB plasmids. F9 cells were transfected with wild-type (WT), mutant (Ml, M2, M3), and truncated (Mt) plasmids as indicated. After 24 hr in medium containing [35S]methionine, cells were harvested and immunoprecipitated with affinity-purified W39 CREB antiserum. labeled products were resolved by SDS-PAGE. Mr = relative molecular weight in thousands. (d) lmmunofluorescence analysis of F9 cells following transfection with wild-type (WT) or mutant (Ml, M2, M3) RSV-CREB expression plasmids. Transfected CREB protein was detected using affinity-purified W39 CREB antiserum. No immunoreactive products were detectable in control (untransfected) cells. In each case, CREB proteins were appropriately targeted to the nuclear compartment.

cAMPStimulated 679

Phosphorylation

of CREB

showed that the M3 protein was expressed and targeted to nuclei of transfected cells. Likewise, M3 retained wildtype activity when cotransfected with the C-subunit expression plasmid (Figure 4a), suggesting that kinase C is not involved in the activation of CREB at Ser-133. The kinase A motif does not appear to participate in DNA binding. We have previously demonstrated, for example, that phosphorylation of purified CREB by kinase A in vitro has no effect on DNAase I footprinting or gelshifting activities (Montminy and Bilezikjian, 1987; Yamamoto et al., 1988). Furthermore, a bacterial f%galactosidase-CREB fusion protein containing the C-terminal 317 amino acids of CREB shows no change in DNA binding activity upon phosphorylation by the C-subunit in vitro (Figure 3~). Our results suggest that the C-subunit of kinase A may directly phosphorylate CREB and thereby activate transcription upon stimulation with CAMP Indeed, treatment of cells with CAMP appears to induce transport of the C-subunit to the nucleus (Nigg et al., 1985). Furthermore, microinjection of the C-subunit into cells can directly activate transcription of CAMP-responsive genes c-fos and vasoactive intestinal peptide (Riabowol et al., 1988). Our results demonstrate that CREB contains a classic kinase A motif that is phosphorylated in response to CAMP and that is absolutely required for activation by the C-subunit in F9 cells. How does phosphorylation activate CREB? Although a number of nuclear factors have been shown to stimulate transcription through domains that are largely acidic in character (Ptashne, 1988), our results suggest that phosphorylation of CREB does not stimulate transcription simply by providing negative charge. It is tempting to speculate that, by analogy with other substrates that are allosterically regulated by kinase A phosphorylation, CREB may also undergo a conformational change that allows a site distal to the kinase A motif to interact with the transcription apparatus. Indeed, the inability of the CREB mutant M2 to stimulate transcription, either constitutively or upon cotransfection with the C-subunit, would argue in favor of this model. Further mutagenesis studies will help to define other structural determinants involved in the activation process. Experimental

Procedures

Immunoprecipltations and Phosphopeptide Mapping All immunoprecipitations were performed using CREB antiserum W39 as described previously (Gonzalez et al., 1969). PC12 and A126-B2 cells were labeled with inorganic =P for 5 hr and then treated with either 10 mM forskolin or ethanol vehicle (control) for 20 min. 32P-labeled ceil extracts were immunoprecipitated with W39 antiserum and were then fractionated by SDS-PAGE. =P-labeled CREB bands, visualized by autoradiography, were cut out and placed in elution buffer. After recovery from gel slices, equal amounts (in cpm) of 3zP-labeled CREB were digested with trypsin and analyzed by twodimensional mapping as described previously (Yamamoto et al., 1988). Purified CREB. labeled in vitro with kinase A and digested with trypsin, was analyzed in parallel with test samples to confirm identity of spots observed on chromatograms. Plasmid Constructions The RSV-CREB expression kb fragment of the CREB

plasmid was constructed by isolating a 1.2 cDNA containing the entire coding region

(1023 bp) plus 100 bp of B’and 52 bp of 3’untranslated sequences. This CREB fragment was inserted into the RSV expression vector RSVSG (gift of S. Gould), which contains 450 bp of the RSV long terminal region plus an SV40 polyadenylation site. CREB mutants were constructed by polymerase chain reaction amplification of a 200 bp Stul-Kpnl cDNA fragment using 30 base oligonucleotide primers containing mutations indicated above (Figure 3a). Amplified fragments were reinserted into the CREB cDNA and then sequenced on both strands of the DNA. Mutant and wild-type constructs were then inserted into the RSV-SG and pGEM vectors. Bacterial expression plasmid KY-CREB was constructed by the insertion of a 1 kb cDNA fragment encoding the C-terminal 317 amino acids of CREB into pGEM. Cell Lines, Transfections, and lmmunofluorescence Studies PC12 and F9 teratocarcinoma cells were maintained in Dulbeccos modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum plus 5% horse serum. Transfections were performed as described previously (Yamamoto et al., 1988) using 5 ug of reporter (A[ - 711 CAT or RSV-CAT), 2.5 pg of RSV-5galactosidase plasmid, and when indicated, 5 ug of C-subunit plasmid MtC (kinase A) (Mellon et al.. 1989) and 7.5 ug of RSVCREB expression plasmid. For each transfection, pGEM plasmid was included to maintain a total of 20 ug of plasmid DNA. Cells were assayed for CAT activity after normalizing to 8galactosidase activity. All assays were performed at least three times. For immunofluorescence analysis, cells were fixed in 3% formalde hyde solution 7 hr after transfection. Affinity-purified W39 CREB antiserum was added to cells at a 1:30 dilution for 15 min. Fluoresceinated goat anti-rabbit antibody was then added at a 1:50 dilution and allowed to incubate for 10 min. After each antibody, cells were washed 10 times in phosphate-buffered saline (pH 7.4). Cells were viewed and photographed on a Leitz microscope. In Vitro Translations and Bacterial Extracts Mutant and wild-type CREB cDNAs in pGEM were transcribed in vitro with T7 RNA polymerase. and transcripts were translated in reticulocyte lysate extracts using [s?S]methionine. Translation products were then immunoprecitated with W39 CREB antiserum and resolved by SDS-PAGE. Bacterial extract preparation, C-subunit phosphorylation in vitro, and gel shift assays were performed as previously described (Gonzalez et al., 1989). Acknowledgments We thank K. Yamamoto for RNAase protection data, S. &Knight for the gift of MtC plasmid, M. Uhler for the gift of MaCK72M plasmid, and W. Vale for advice and support. We also thank members of the Max Planck Lab (Salk Institute) for oligonucleotide synthesis. This work was supported by United States Public Health Service grant GM 37828 and conducted in part by the Clayton Foundation for Research, California Division. M. R. M. is a Clayton Foundation Investigator. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC. Section 1734 solely to indicate this fact. Received

July 26, 1989; revised

September

7, 1989.

Refennces Chomczynski, P, and Sacchi, N. (1987). Single-step method isolation by acid guanidinium thiocyanate-phenol-chloroform tion. Anal. Biochem. 162, 156-199.

of RNA extrac-

Comb, M., Birnberg, N. C., Seascholtz, A., Herbert, E., and Goodman, H. M. (1986). A cyclic-AMPand phorbol ester-inducible DNA element. Nature 323, 353-356. Gonzalez, G. A., Yamamoto, K. K., Fischer, W. H., Karr, D., Menzel, P., Biggs, W., Ill, Vale, W. W., and Montminy, M. R. (1989). A cluster of phosphorylation sites on the cyclic AMP-regulated nuclear factor CREB predicted by its sequence. Nature 337, 749-752. Lewis, E. J., Harrington, C. A., and Chikaraishi, D. M. (1987). Transcrip tional regulation of the tymsine hydroxylase gene by glucocorticoid and cyclic AMP Proc. Natl. Acad. Sci. USA 84, 3550-3554.

Cell 680

Mellon, P L., Clegg, C. H., Correll, L. A., and McKnight, G. S. (1989). Regulation of transcription by cyclic AMP-dependent protein kinase. Proc. Natl. Acad. Sci. USA 86, 4887-4892. Montminy, M. R., and Bilezikjian, protein to the cyclic-AMP response Nature 328, 175-178.

L. M. (1987). Binding of a nuclear element of the somatostatin gene.

Montminy, M. Ft., Sevarino, K. A., Wagner, J. A., Mandel, G., and Goodman, R. H. (1988). Identification of a cyclic-AMP-responsive element within the rat somatostatin gene. Proc. Natl. Acad. Sci. USA 83, 8882-8888. Nigg, E. A., Hilz, H., Eppenberger, H. M., and Dutly, F. (1985). Rapid and reversible translocation of the catalytic subunit of CAMP-dependent protein kinase type II from the Golgi complex to the nucleus, EMBO J. 4, 2801-2808. Plet, A., Evain, D., and Anderson, W. B. (1982). Effect of retinoic acid treatment of F9 embryonal carcinoma cells on the activity and distribution of cyclic AMP-dependent protein kinase. J. Biol. Chem. 257, 889-893. Ptashne, M. (1988). How eukaryotic ture 335, 883-889.

transcriptional

activators

work. Na-

Riabowol, K. T., Fink, J. S., Gilman, M. Z., Walsh, D. A., Goodman, R. H., and Feramisco, J. R. (1988). The catalytic subunit of CAMPdependent protein kinase induces expression of genes containing CAMP-responsive enhancer elements. Nature 336, 83-88. Rickles, R. J., Darrow A. L., and Strickland, S. (1989). Differentiationresponsive elements in the 5’ region of the mouse tissue plasminogen activator gene confer two-stage regulation by retinoic acid and cyclic AMP in teratocarcinoma cells. Mol. Cell. Biol. 9, 1891-1704. Sasaki, K., Cripe, T. P, Koch, S. R., Andreone, T L., Peterson, D. D., Beale, E. B., and Granner, K. K. (1984). Multihormonal regulation of phosphoenolpyruvate carboxykinase gene transcription. J. Biol. Chem. 259, 15242-15251. Yamamoto, K. R., Gonzalez, G. A., Biggs, W. H., Ill, and Montminy, M. R. (1988). Phosphorylation-induced binding and transcriptional efficacy of nuclear factor CREB. Nature 334, 494-498. Zinn, K., DiMaio, D., and Maniatis, T. (1983). Identification tinct regulatory regions adjacent to the human f&interferon 34, 865-879.

of two disgene. Cell