Cell, Vol. 60, 611-617,
February
23, 1990, Copyright
Characterization in Transcription
0 1990 by Cell Press
of a Bipartite Activator Factor CREB
K. K. yamamoto, G. A. Gonzalez, R Menzel, and M. R. Montminy The Clayton Foundation Laboratories for Peptide Biology The Salk institute 10010 North Torrey Pines Road La Jolla, California 92037
J. Rivier,
Summary In this paper, we characterize a trans-activating region in CREB, termed a, that interacts cooperatively with the kinase A phosphorylation motif to stimulate transcription. The a region appears to be encoded by an alternate exon that is deleted in a CREB-related cDNA named ACREB. Both proteins are expressed in eukaryotic cells, although the activity of CREB is lo-fold higher than that of ACREB. Circular dichroism data on a synthetic “a peptide” combined with results from in vitro mutagenesis experiments support the hypothesis that the a region contains an amphipathic a helix whose structure is critical to CREB activity. We propose that phosphorylation by kinase A may stimulate CREB activity in part by modulating the structure of a and thus may stimulate its ability to interact with other proteins in the polymerase II complex. Introduction CREB is a 43 kd nuclear protein that induces transcription of a number of genes in response to CAMP (Montminy et al., 1986; Yamamoto et al., 1988; Gonzalez et al., 1989). Following hormonal stimulation and subsequent activation of adenyl cyciase, the catalytic subunit (C-subunit) of kinase A appears to be transiocated to the nucleus where it phosphorylates CREB at a single serine residue, Ser133 (Nigg et al., 19%5; Gonzalez and Montminy, 1989). We determined previously that phosphorylation of CREB at Ser-133 is critical for CAMP-inducible transcription. As CREB mutants that contain acidic residues in place of the Ser-133 phosphoacceptor are inactive, phosphorylation does not appear to work simply by providing negative charge. Rather, by analogy with other substrates of kinase A, CREB may be regulated through an allosteric mechanism that allows a site distal to the phosphorylation motif to interact with the transcription apparatus. In the process of characterizing CREB-related cDNAs in PC12 cells, we obtained a cross-hybridizing clone, named ACREB, that contains a 14 amino acid deletion from residues 88 to 101 (Figure la) and appears to be homologous to a human CREB cDNA reported by Hoeffler et al. (1988). Sequence analysis revealed that CREB and ACREB cDNAs are otherwise identical and might therefore represent alternately spliced products from a single gene.
Domain
Results To determine whether CREB and ACREB could be distinguished functionally, we performed a series of transient transfection assays using eukaryotic expression vectors encoding each form. We employed the F9 teratocarcinoma line because these ceils are largely unresponsive to CAMP, thereby allowing the activity of the CREB gene to be easily monitored (Gonzalez and Montminy, 1989). When cotransfected with a C-subunit expression plasmid, the RSV-CREB construct caused a 200-fold induction in the activity of a CRE-CAT reporter gene (Figure lb). Surprisingly, ACREB was only one-tenth as active as CREB in stimulating CRE-CAT activity, indicating that the 14 amino acid region, termed a, was important for maximal CAMP inducibility. The dramatic contrast in trans-activation potential between CREB and ACREB prompted us to examine the expression of these two factors in various tissues. To discriminate between mRNAs encoding these two proteins we employed an RNAase protection assay. Using a CREB antisense probe of 807 nucieotides, we observed a single 747 nucleotide protected fragment arising from CREB mRNA and two smaller fragments of 118 and 587 nucleotides arising from ACREB mRNA (Figure 2A). Analysis of total RNA from whole brain or hypothalamic tissue revealed three fragments identical in size to those observed in reconstitution experiments using in vitro-synthesized RNAs. The relative intensity of these protected fragments suggests that ACREB mRNA is ~4- to 5-fold more abundant than CREB mRNA. The same ratio of the two transcripts was observed for all of the tissues we tested (heart, testis, kidney, adrenal, liver, and gut), indicating that the ratio of CREB and ACREB mRNA was not regulated in a tissue-specific fashion. Moreover, the absence of any RNAase protected fragments other than those observed for CREB and ACREB would argue against additional splice products. To distinguish between CREB and ACREB proteins, we prepared antiserum to a synthetic 16 amino acid peptide containing the a region (Figure 28). As predicted, the a peptide antiserum 240 was capable of immunoprecipitating CREB but not ACREB proteins that had been synthesized in vitro. In contrast, CREB antiserum 244, developed against a synthetic peptide from residues 128-162 of CREB, was capable of recognizing both proteins. To determine whether both CREB and ACREB proteins are indeed expressed in tissues, we prepared nuclear extracts from rat brain (Figure 2C). Following purification by CRE oligonucleotide affinity chromatography as described previously (Yamamoto et al., 1988) the extracts were phosphorylated with the C-subunit + [+s2]ATP, and immunoprecipitates were then prepared with the a peptide-specific antiserum 240. We observed a single 32P-iabeled 43 kd band following SDS-PAGE (Figure 2c) that was the size of the full-length CREB protein. In contrast, no such prod-
Cell 612
a.
Figure 1. Comparison Structure and Activity 327 aa.
ACRE6 OH3
CREB ,’
,’
------_____ ---___
,I’
---.__
---__.
b.
I--
PKA CONSTRUCT
L
C-
c
+
-+ CREB
341 aa
of CREB
and ACRE6
(a) Schematic representation of CREB and ACREB proteins. The kinase A phosphorylation site is indicated by the letter P over a black box. DBD/U: DNA binding domain and leutine zipper. The numbers of amino acids encoded by CREB and ACREB cDNAs are indicated on the right. The position of (I region in relation to the ACREB sequence is indicated by the broken line between the Gln and Ile residues. The sequence of Q region and neighboring residues is indicated below. Residues common to both proteins are shown in boldface. (b) CAT assay of RW-CREB and RSV-ACREB expression vectors in F9 cells. Cells were cotransfected with CRE-CAT reporter and either C-subunit expression plasmid (+) or control 9globin plasmid (-). C, control transfection without CREB or ACREB plasmids. The bar graph below shows CAT activity for each construct as measured by the percent of conversion of chloramphenicol to acetylated forms. Stippled bars indicate basal activity and black bars indicate C-subunit-stimulated activity.
-+
ACREB
uct was observed using preimmune serum (Gonzalez et al., 1989). Less than 10% of labeled CREB was recovered upon reprecipitation with antiserum 240 (data not shown), indicating that precipitation of CREB by this procedure was nearly quantitative. lmmunoprecipitation of the same 240 supernatant with antiserum 244 (which recognizes both CREB and ACREB) revealed a ACREB product that was 4- to B-fold more abundant than CREB. Moreover, the same ratios of CREB and ACREB content were detected by Western blot analysis of the affinity-purified extract (data not shown), suggesting that the relative abundance of CREB and ACREB proteins closely parallels that of their corresponding mRNA levels. To understand the basis for the functional differences observed between CREB and ACREB proteins, we initially examined four parameters that appear to be critical for optimal transcriptional activity: nuclear localization, DNA binding, phosphorylation, and dimerization. Immunofluorescence analysis of F9 cells transfected with either CREB or ACREB expression plasmids (Figure 3a) showed that both proteins are appropriately targeted to the nuclear compartment. Furthermore, gel mobility shift assays of CREB and ACREB IacZ-fusion proteins showed that these factors have very similar DNA binding properties (Figure 3b). Moreover, kinase A treatment of brain (Figure 2c) as well as bacterial extracts (data not shown) demonstrated that
CREB and ACREB were comparable substrates for C-subunit-mediated phosphorylation of Ser-133. Using a glutaraldehyde cross-linking assay, we previously demonstrated that CREB can exist as a homodimer in solution (Dwarki et al., 1989). Similarly, glutaraldehyde treatment of in vitro-synthesized ACREB protein revealed a 90 kd cross-linked form (Figure SC), suggesting that the decreased transcriptional activity of ACREB cannot be attributed to a failure to dimerize. ACREB protein appears, therefore, to be transported to the nucleus and phosphorylated and to form dimers as efficiently as CREB, yet it remains less active. In the absence of such differences, we hypothesized that the a region might impart activity through its interaction with other proteins in the RNA polymerase II complex. To determine whether the a region assumes a secondary structure that would favor such interactions, we performed circular dichroism studies on the synthetic a pep tide (Figure 4a). Circular dichroism is a spectroscopic technique that relies on the principle that optically active substances will absorb left and right circularly polarized light to different degrees. This difference in absorbance, referred to as dichroism, is characteristic for various protein conformations and can be used to predict the secondary structure of small peptides with a high degree of certainty. a-helical peptides, in particular, have characteristic
;;rtively
Spliced
Activator
A.
B.
2~
2r
, MW 105-
CREB
,,
ACREB
,
240 244 240 244
CREB 75-
ACREB
50-
C. Affinity purified CREB (Brain) I
+ I.
lmmunoprecipitate 240 Ab / supernafant
II.
lmmunoprectpitate 244 Ab
Figure
1
2. Comparison
of CREB
and ACREB
Expression
in Brain
(A) RNAase protection assays of total RNA from rat brain and hypothalamus compared with in vitro-synthesized CREB and ACREB RNAs. On the left is a schematic representation of antisense RNA probe (SP6) and CREB and ACREB RNAs. The size of protected fragments (in nucleotides) is indicated above. RNAase protection boundaries are indicated by broken lines. On the right, a 6% denaturing gel shows products of RNAase protection assay. Arrows point to fragments of the size predicted in the adjacent schematic. Hypo, Brain: total RNA (20 vg) from hypothalamus and brain. ACREB, CREB: RNAs synthesized in vitro from ACREB and CREB cDNAs, respectively. tRNA: RNAase assay using tRNA as a control for digestion. Probe: antisense CREB RNA probe prior to RNAase digestion. MW: molecular weight marker (Mspl digest of pBR 322) with fragment size indicated on the right). (B) Characterization of antiserum specific for a region of CREB. On the left is a diagram showing the specificity of 240 and 244 antisera for a region (residues 69-101) or the region common to CREB and ACREB
circular dichroism spectra consisting of an absorbance maximum at 192 nm with absorbance minima at 208 nm and 222 nm. The spectrum we obtained, containing an energy maximum at 192 nm and energy minima at 208 nm and 222 nm, demonstrated that this peptide forms a nonrandom structure with partial a-helical content. Furthermore, a helical wheel diagram of residues 90-96 in CREB suggests that the a peptide may contain an a helix that is amphipathic in nature, with primarily hydrophobic residues on one face and mostly basic residues on the other. Spectral analysis of a second CREB peptide extending from residues 121-159 failed to reveal any secondary structure (data not shown), however, suggesting that the results we observed with the a peptide were specific for this region of the protein. To test whether the secondary structure of the a peptide region was critical to CREB activity, we designed several a region mutants and tested these by transient assay in F9 cells (Figure 4b). CREB mutant M93 L/P contains a leutine to proline substitution at position 93 that would be predicted to disrupt helix formation within the a region. When inserted into an RSV expression vector and cotransfected into F9 cells with a C-subunit expression plasmid, the M93 UP mutant, like ACREB, was almost completely unable to stimulate expression of the CRE-CAT reporter plasmid. Distortion of a peptide secondary structure would appear, therefore, to destroy completely the contribution of this region to CREB activity. By contrast, a more conservative cystine to serine substitution at position 90, mutant M90 C/S, which would be predicted to preserve secondary structure, was 5-fold more active than M93 L/P and had activity intermediate between ACRE6 and CREB. As M90 C/S and M93 L/P mutants were expressed at comparable levels to the wild-type CREB plasmid, they were also targeted efficiently to nuclei of transfected cells (data not shown) and had normal DNA binding activity. The differences in Vans-activation potential between these proteins would therefore appear to result from alterations in a peptide structure and not from differences in expression, DNA binding, or nuclear localization. To determine whether the a region would stimulate ACREB activity when separated from its neighboring sequences, we inserted the 14 amino acid domain into position 23 of the ACREB protein instead of its normal position at amino acid 88 (Figure 5b). When tested by transient assay in F9 cells, this mutant, M23a+, caused a 21-fold increase in CRE-CAT activity and was 3 times more induc-
(residues 121-159), indicated by stippled box. On the right is an SDS-PAGE of 35S-labeled CREB and ACREB proteins synthesized in vitro and immunoprecipitated with either a region-specific 240 antiserum or 244 antiserum. MW (in kd) is indicated on the left. (C) Identification of CREB and ACREB proteins in brain using antisera 240 and 244. On the left is a schematic showing the strategy used to isolate CREB and ACREB proteins individually. Following purification by CRE oligonucleotide affinity chromatography, brain extract was labeled with C-subunit plus [Y-~‘P]ATP and immunoprecipitated with 240 antiserum. The supernatant fraction was immunoprecipitated with 244 antiserum, and pellet fractions from both reactions (I and II) were analyzed by SDS-PAGE (right). MW (in kd) is indicated on the left.
Cell 614
a
(:REB
A(ZREE!
b.
C. I
CREB CRE
Figure
3. Comparison
of CREB
i/2 CRE
and ACREB
”
ACREB 112 CRE CRE
Protein
240 ACREB’
’
Targeting,
DNA
Binding,
and Dimerization
Activities
(a) lmmunofluorescence analysis of CREB and ACRE6 proteins in F9 ceils following transfection with expression plasmids encoding either form. Nomarski (N) or immunofluorescence (IF) micrographs following treatment of transfected cells with CREB antiserum 244 are shown. Arrows point to the nuclei of antigen that contain cells. (b) Gel shift assays of bacterial IacZ-CREB or ACREB fusion proteins using double-stranded CRE or half-site mutant (l/2 CRE) probes. I and II indicate putative dimer and monomer forms of CREB, respectively. (c) Glutaraldehyde cross-linking assay of ACREB protein synthesized in vitro. Asterisk indicates that protein was %-labeled. Lane 1, immunoprecipitation of %-labeled ACREB protein with antiserum 244 after glutaraldehyde treatment. Lane 2, immunoprecipitation of ACREB protein with a region-specific 240 antiserum. Lane 3, immunoprecipitation of %-labeled ACREB protein plus unlabeled CREB protein with 240 antiserum after glutaraldehyde treatment. MW, molecular weight in kd.
ible by C-subunit than ACREB. Although M23a+ was less potent than wild-type CREB protein, the ability of the a region to enhance ACREB activity when removed from its normal context indicates that this domain can indeed function autonomously. The a region thus defines both a
structural and a functional motif within the CREB molecule. Having determined that the a region participates in rfans-activation as a functional unit, we sought to understand whether, and in what manner, this domain interacts
Alternatively 615
Spliced
Activator
a.
Figure 4. Analysis Function
r,
,96
190
b.
200
220
210
230
240
250
NANOMETERS
’
101 Gln Ser Ser CyS LyS Asp Leu Lys Arg Iau Phe Ser Gly Thr Gln CAG TCT TCC TGT AAG GAC TTA AAA AGA CTT TTC TCC GGA ACT CAG
07
CREB aPEPTIDE:
J.
J
TCI .Ser MUTANTS:
CCA prq M93 UP
M90 C/S
of cr Region
Structure
and
(a) Helical wheel diagram (residues 90-96) and circular dichroism spectrum (residues 69-101) of synthetic peptide containing the 14 amino acid a region. 0, degrees of rotation (x 10s) shown as a function of wavelength (in nm). +, - indicates charge characteristics of polar amino acids. (b) CAT assay of wild-type and mutant CREB expression plasmids following transfection into F9 cells. The sequence of amino acid substitution mutants M90 C/S and M93 UP in relation to wild-type CREB protein in a region from amino acids 69101 is indicated. CAT assay of various plasmids is shown below. C, control transfection without CREB plasmids showing endogenous CREB activity in F9 cells. -, + indicates whether C-subunit (+) or control 6-globin (-) expression plasmid was cotransfected with CREB.
I--II PKA
-+-+-+-+-+ -C-
CONSTRUCT
%REBJ
C
L ACREB’
CREB
ACREB Construct
‘M90 C/S’ ’ M93 UP’
M90 C/S
M93 UP
. with the pK-A phosphorylation site to confer CAMP responsiveness. To monitor a activity in isolation, we used the CREB mutant Ml, in which the pK-A phosphorylation site is destroyed by a Ser-133 to Ala-133 substitution (Gonzalez and Montminy, 1989). Following transfection into F9 cells, CREB mutant Ml stimulated CRE-CAT expression 3-to 5-fold higher than ACREB, having basal activity similar to wild-type CREB (Figure 5a). In unstimulated cells, CREB would appear to have basal activity that is therefore dependent upon the a region. The ratio of CREB to ACREB basal activity over six separate experiments averaged 3:i. When cotransfected with the C-subunit plasmid, the ratio of CREB to ACREB activity increased to lO:l, suggesting that the a region not only imparts constitutive activity to CREB but also participates in C-subunitstimulated activity with the pK-A phosphorylation site. In stimulated cells, therefore, C-subunit-mediated activation
of CREB may rely on a cooperative interaction between the Ser-133 phosphoacceptor and the a region. Based on our hypothesis that CREB is subject to allosteric regulation, pK-A-mediated phosphorylation of CREB may cause structural changes in the protein that affect the ability of the a region to interact with other proteins in the RNA polymerase II transcription complex. Indeed, the presence of multiple basic residues within the a peptide suggests a strong potential for ion pair formation with the Ser-133 phosphate. Such ion pairs have been shown to account for the conformational changes induced by pKA-mediated phosphorylation of glycogen phosphorylase A (Sprang et al., 1988). Do these regions need to interact in cis, or can CREB mutants lacking either region complement one another in trans? To test whether the a region and pK-A motif could interact in this manner, we cotransfected F9 cells with CREB mutant Ml (which lacks the
Cdl
616
BASAL
C 0.4%
1
1
CREB 5.65%
Figure 5. Analysis of a Region and pK-A Motif Interactions (a) CAT assay showing activity of CREB, ACREB, and Ml expression plasmids (in F9 cells) in the absence of cotransfected C-subunit plasmid. C, control transfection of F9 cells with CRECAT reporter in the absence of added CREB or C-subunit plasmids. The percent of conversion is indicated below. (b) Tins-activation potential of CREB-related constructs when cotransfected with C-subunit plasmid. Absolute activity for each construct is normalized to wild-type CREB activity (100%). Values shown represent an average of five separate experiments. In all cases, CAT activity was measured in extracts after normalizing to ft-galactosidase activity from a cotransfected RBV-6gal control plasmid. The schematic on the left shows a diagram of each protein. o denotes a region, P signifies the pK-A phosphorytation motif, and the stippled box indicates DNA binding and dimerixation region. The M23o+ plasmid was constructed by inserting the a region into ACREB at position 23 instead of its normal position at amino acid 66. The arrow indicates that o region was moved in an N-terminal direction. ACREB + Ml indicates that these plasmids were cotransfected into F9 cells.
ACTIVITY
1
ACREB 0.85%
1
Ml 4.0%
1
b. PK-A DEPENDENT ACTIVITY P v
a
+e
CREB ACREB
100 12.4 12.8 31.6
16
pK-A motif) and ACREB (which lacks the a motif). When compared with CREB, the Ml + ACREB transfection remained lo-fold less active (Figure 5b), thus suggesting the necessity for a cis interaction between the pK-A motif and the a region. Although this cis interaction may reflect a direct interaction between basic residues of the a region and the Ser-133 phosphate, the a region may also influence the secondary structure of the pK-A motif without such ion pairing.
Conclusions A functional role for ACREB protein in the CAMP signaling process remains elusive. The absence of a domain so critical for transcriptional activation leads us to suspect that ACREB may serve to downregulate the response to CAMP, perhaps by serving as a competitive inhibitor of CREB. Indeed the ability of ACREB to form heterodimers with CREB would further support this regulatory scheme (Figure 3~). In this case, the synthesis (or degradation) of ACREB and CREB proteins should be differentially regulated by CAMP. Alternatively, ACREB may stimulate transcription in a promoter-specific manner. Furthermore, the presence of additional phosphorylation sites in CREB (i.e., casein kinase II and kinase C) suggests that ACREB
activity may be regulated through cellular mechanisms other than CAMP In summary, we have described two proteins, CREB and ACREB, that differ only by the insertion of a 14 amino acid a region. ACREB is lo-fold less active yet 5-fold more abundant than CREB protein. It appears likely that CREB and ACREB mRNAs may arise by alternative splicing of a common precursor. Preliminary Southern blotting analysis of rat genomic DNA using fragments confined to the CREB activator domain suggests the presence of two or three related sequences. Nevertheless, final proof of this mechanism awaits characterization of genomic clones for CREB. Our studies also define a novel type of transactivation domain whose activity is dependent upon two distinct regions of the CREB protein: the a region and the pK-A phosphorylation site. Structural analysis of the CREB protein will allow us to distinguish individual amino acids that interact with the Ser-133 phosphate to confer this activity. Experimental
Procedures
Cloning and Plasmid Constructions ACREB cDNA was obtained after screening a Xgt 11 PC-12 library with a CREB cDNA probe as previously described (Gonzalez et al., 1969). The RBVACREB expression plasmid was constructed by inserting the
Alternatively 617
Spliced
Activator
ACRE6 cDNA, which contains the entire coding region (961 nucleotides) plus 100 bp of 5’ and 52 bp of 3’ untranslated sequences, into the Rous sarcoma virus expression vector RSV-SG (Gonzalez and Montminy, 1989). CREB mutants M90 C/S and M93 UP were constructed by PCR amplification of a 300 bp CREB cDNA fragment using 39-base antisense oligonucleotide primers that contain appropriate mutations and a sense oligonucleotide that corresponds to the 5’ noncoding region. Amplified fragments were digested with restriction enzyme BspMll and inserted into a CREB cDNA vector. The PCRamplified regions for each mutant were then sequenced on both strands. Mutant M23a+ was constructed by annealing sense and antisenae oligomers that encode the u peptide sequence from amino acids 89-101 and inserting the double-stranded oligo into a Hincll site of the ACREB cDNA. . RNAase Protection Asssys Total RNA was prepared from rat brain and hypothalamus to Chomczynski and Sacchi (1987). CREB antisense RNA prepared as previously described (Gonzalez and Montminy, RNAase protection assays were performed according to (1963).
according probe was 1969) and Zinn et al.
Affinity Purification and Phosphorylatlon of CREB CREB protein was prepared from rat brain extracts and purified by CRE oligonucleotide affinity chromatography (Yamamoto et al., 1966). CREB was labeled with purified C-subunit of kinase A (kindly provided by S. Taylor) as described previously (Gonzalez and Montminy, 1989). Peptide Synthesis, AntIsera, and Ciwlar Dichrolsm Analysis CREB (121159) and a peptides were synthesized using primary sequence data in Figure la. The a peptide consists of residues 89-101 plus Gly-Tyr at the C-terminus. Both peptides were conjugated to human a-globulins, and antisera 240 and 244 were then prepared as previously described (Gonzalez et al., 1989). For circular dichroism studies, a peptide (90 ug/ml) was analyzed in .Ol M phosphate buffer (pH 7.2) plus 40% trifluoroethanol. Secondary structure was estimated using the AVIV program, which is based upon an algorithm and reference spectra of J. T. Yang. The program makes a statistical fit of three reference spectra representing a helix, 6 sheet, and 6 turn. Cell Lines, l?ansfections, and lmmunofluorwcence Studies F9 teratocarcinoma cells were maintained in Dulbeccds modified Eagles medium (DMEM) supplemented with 10% fetal calf serum. Transfections and immunofluorescence studies were performed as described previously (Gonzalez and Montminy, 1989). CAT activity was measured from cell extracts after normalizing to 6-galactosidas.e activity derived from cotransfected RSV+gal plasmid. In Vitro Translatlons, Cross-Llnklng, and Bacterial Extracts CREB and ACREB cDNAs were inserted into pGEM vectors and transcribed in vitro with T7 polymerase. Transcripts were then translated in reticulocyte lysate extracts using rj5Sjmethionine. For cross-linking studies, labeled proteins were incubated at 2ooC with 005% glutaraldehyde for 30 min. Cross-linked proteins were then immunoprecipitated and complexes were resolved by SDS-PAGE. Bacterial extracts containing various CREB-related proteins were prepared and analyzed as described previously (Gonxalez et al., 1989). Acknowledgments We thank T Hunter, M. McKeown, B. Sefton, and I. Verma for commentson this manuscript. We also thank J. Andrews for circular dichroism spectra, J. Porter and R. Kaiser for peptide synthesis and purification, and J. Vaughan for characterization of antisera. This work was supported by National Institutes of Health grants GM 37828 and DK 26741. 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 U.S.C. Section 1734 solely to indicate this fact. Received
October
27, 1989; revised
December
12, 1989.
References 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, hf., Birnberg, N. C., Seascholtz, A., Herbert, E., and Goodman, H. M. (1986). A cyclic-AMP and phorbol ester-inducible DNA element. Nature 323, 353-356. Dwarki, V. J., Montminy, M. R., and Verma, I. M. (1989). Both the basic region and the “leucine zipper” domain of CREB protein are essential for transcriptional activation, EMBO J., in press. Gonzalez, G. A., and Montminy, M. R. (1989). Cyclic AMP stimulates somatostatin gene transcription by phosphorylation of CREB at serine 133. Cell 59, 675-680. 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. Hoeffler, J. P, Meyer, T E., Yun, Y., Jameson, J. L., and Habener, J. F. (1988). Cyclic-AMP-responsive DNA-binding protein: structure based on a cloned placental cDNA. Science 242, 1430-1432. Montminy, M. R., and Bilezikjian, protein to the cyclic-AMP response Nature 328, 175-178.
L. M. (1967). Binding of a nuclear element of the somatostatin gene.
Montminy, M. R., Sevarino, K. A., Wagner, J. A., Mandel, G., and Goodman, R. H. (1986). Identification of a cyclic-AMP responsive element within the rat somatostatin gene. Proc. Natl. Acad. Sci. USA 83, 6682-6686. 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, 2801-2806. Riabowol, K. T, Fink, J. S., Gilman, M. Z., Walsh, R. H., and Feramisco, J. R. (1988). The catalytic dependent protein kinase induces expression of CAMP-responsive enhancer elements. Nature 336,
D. A., Goodman, subunit of CAMPgenes containing 83-86.
Sprang, S. R., Acharya, K. R., Goldsmith, E. J., Stuart, D. I., Varvill, K., Fletterick, R. J., Madsen, N. B., and Johnson, L. N. (1988). Structural changes in glycogen phosphorylase induced by phosphorylation. Nature 336, 215-221. Yamamoto, K. K., 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 p-interferon 34, 865-879.
of two disgene. Cell
February
23, 1990, Copyright
Characterization in Transcription
0 1990 by Cell Press
of a Bipartite Activator Factor CREB
K. K. yamamoto, G. A. Gonzalez, R Menzel, and M. R. Montminy The Clayton Foundation Laboratories for Peptide Biology The Salk institute 10010 North Torrey Pines Road La Jolla, California 92037
J. Rivier,
Summary In this paper, we characterize a trans-activating region in CREB, termed a, that interacts cooperatively with the kinase A phosphorylation motif to stimulate transcription. The a region appears to be encoded by an alternate exon that is deleted in a CREB-related cDNA named ACREB. Both proteins are expressed in eukaryotic cells, although the activity of CREB is lo-fold higher than that of ACREB. Circular dichroism data on a synthetic “a peptide” combined with results from in vitro mutagenesis experiments support the hypothesis that the a region contains an amphipathic a helix whose structure is critical to CREB activity. We propose that phosphorylation by kinase A may stimulate CREB activity in part by modulating the structure of a and thus may stimulate its ability to interact with other proteins in the polymerase II complex. Introduction CREB is a 43 kd nuclear protein that induces transcription of a number of genes in response to CAMP (Montminy et al., 1986; Yamamoto et al., 1988; Gonzalez et al., 1989). Following hormonal stimulation and subsequent activation of adenyl cyciase, the catalytic subunit (C-subunit) of kinase A appears to be transiocated to the nucleus where it phosphorylates CREB at a single serine residue, Ser133 (Nigg et al., 19%5; Gonzalez and Montminy, 1989). We determined previously that phosphorylation of CREB at Ser-133 is critical for CAMP-inducible transcription. As CREB mutants that contain acidic residues in place of the Ser-133 phosphoacceptor are inactive, phosphorylation does not appear to work simply by providing negative charge. Rather, by analogy with other substrates of kinase A, CREB may be regulated through an allosteric mechanism that allows a site distal to the phosphorylation motif to interact with the transcription apparatus. In the process of characterizing CREB-related cDNAs in PC12 cells, we obtained a cross-hybridizing clone, named ACREB, that contains a 14 amino acid deletion from residues 88 to 101 (Figure la) and appears to be homologous to a human CREB cDNA reported by Hoeffler et al. (1988). Sequence analysis revealed that CREB and ACREB cDNAs are otherwise identical and might therefore represent alternately spliced products from a single gene.
Domain
Results To determine whether CREB and ACREB could be distinguished functionally, we performed a series of transient transfection assays using eukaryotic expression vectors encoding each form. We employed the F9 teratocarcinoma line because these ceils are largely unresponsive to CAMP, thereby allowing the activity of the CREB gene to be easily monitored (Gonzalez and Montminy, 1989). When cotransfected with a C-subunit expression plasmid, the RSV-CREB construct caused a 200-fold induction in the activity of a CRE-CAT reporter gene (Figure lb). Surprisingly, ACREB was only one-tenth as active as CREB in stimulating CRE-CAT activity, indicating that the 14 amino acid region, termed a, was important for maximal CAMP inducibility. The dramatic contrast in trans-activation potential between CREB and ACREB prompted us to examine the expression of these two factors in various tissues. To discriminate between mRNAs encoding these two proteins we employed an RNAase protection assay. Using a CREB antisense probe of 807 nucieotides, we observed a single 747 nucleotide protected fragment arising from CREB mRNA and two smaller fragments of 118 and 587 nucleotides arising from ACREB mRNA (Figure 2A). Analysis of total RNA from whole brain or hypothalamic tissue revealed three fragments identical in size to those observed in reconstitution experiments using in vitro-synthesized RNAs. The relative intensity of these protected fragments suggests that ACREB mRNA is ~4- to 5-fold more abundant than CREB mRNA. The same ratio of the two transcripts was observed for all of the tissues we tested (heart, testis, kidney, adrenal, liver, and gut), indicating that the ratio of CREB and ACREB mRNA was not regulated in a tissue-specific fashion. Moreover, the absence of any RNAase protected fragments other than those observed for CREB and ACREB would argue against additional splice products. To distinguish between CREB and ACREB proteins, we prepared antiserum to a synthetic 16 amino acid peptide containing the a region (Figure 28). As predicted, the a peptide antiserum 240 was capable of immunoprecipitating CREB but not ACREB proteins that had been synthesized in vitro. In contrast, CREB antiserum 244, developed against a synthetic peptide from residues 128-162 of CREB, was capable of recognizing both proteins. To determine whether both CREB and ACREB proteins are indeed expressed in tissues, we prepared nuclear extracts from rat brain (Figure 2C). Following purification by CRE oligonucleotide affinity chromatography as described previously (Yamamoto et al., 1988) the extracts were phosphorylated with the C-subunit + [+s2]ATP, and immunoprecipitates were then prepared with the a peptide-specific antiserum 240. We observed a single 32P-iabeled 43 kd band following SDS-PAGE (Figure 2c) that was the size of the full-length CREB protein. In contrast, no such prod-
Cell 612
a.
Figure 1. Comparison Structure and Activity 327 aa.
ACRE6 OH3
CREB ,’
,’
------_____ ---___
,I’
---.__
---__.
b.
I--
PKA CONSTRUCT
L
C-
c
+
-+ CREB
341 aa
of CREB
and ACRE6
(a) Schematic representation of CREB and ACREB proteins. The kinase A phosphorylation site is indicated by the letter P over a black box. DBD/U: DNA binding domain and leutine zipper. The numbers of amino acids encoded by CREB and ACREB cDNAs are indicated on the right. The position of (I region in relation to the ACREB sequence is indicated by the broken line between the Gln and Ile residues. The sequence of Q region and neighboring residues is indicated below. Residues common to both proteins are shown in boldface. (b) CAT assay of RW-CREB and RSV-ACREB expression vectors in F9 cells. Cells were cotransfected with CRE-CAT reporter and either C-subunit expression plasmid (+) or control 9globin plasmid (-). C, control transfection without CREB or ACREB plasmids. The bar graph below shows CAT activity for each construct as measured by the percent of conversion of chloramphenicol to acetylated forms. Stippled bars indicate basal activity and black bars indicate C-subunit-stimulated activity.
-+
ACREB
uct was observed using preimmune serum (Gonzalez et al., 1989). Less than 10% of labeled CREB was recovered upon reprecipitation with antiserum 240 (data not shown), indicating that precipitation of CREB by this procedure was nearly quantitative. lmmunoprecipitation of the same 240 supernatant with antiserum 244 (which recognizes both CREB and ACREB) revealed a ACREB product that was 4- to B-fold more abundant than CREB. Moreover, the same ratios of CREB and ACREB content were detected by Western blot analysis of the affinity-purified extract (data not shown), suggesting that the relative abundance of CREB and ACREB proteins closely parallels that of their corresponding mRNA levels. To understand the basis for the functional differences observed between CREB and ACREB proteins, we initially examined four parameters that appear to be critical for optimal transcriptional activity: nuclear localization, DNA binding, phosphorylation, and dimerization. Immunofluorescence analysis of F9 cells transfected with either CREB or ACREB expression plasmids (Figure 3a) showed that both proteins are appropriately targeted to the nuclear compartment. Furthermore, gel mobility shift assays of CREB and ACREB IacZ-fusion proteins showed that these factors have very similar DNA binding properties (Figure 3b). Moreover, kinase A treatment of brain (Figure 2c) as well as bacterial extracts (data not shown) demonstrated that
CREB and ACREB were comparable substrates for C-subunit-mediated phosphorylation of Ser-133. Using a glutaraldehyde cross-linking assay, we previously demonstrated that CREB can exist as a homodimer in solution (Dwarki et al., 1989). Similarly, glutaraldehyde treatment of in vitro-synthesized ACREB protein revealed a 90 kd cross-linked form (Figure SC), suggesting that the decreased transcriptional activity of ACREB cannot be attributed to a failure to dimerize. ACREB protein appears, therefore, to be transported to the nucleus and phosphorylated and to form dimers as efficiently as CREB, yet it remains less active. In the absence of such differences, we hypothesized that the a region might impart activity through its interaction with other proteins in the RNA polymerase II complex. To determine whether the a region assumes a secondary structure that would favor such interactions, we performed circular dichroism studies on the synthetic a pep tide (Figure 4a). Circular dichroism is a spectroscopic technique that relies on the principle that optically active substances will absorb left and right circularly polarized light to different degrees. This difference in absorbance, referred to as dichroism, is characteristic for various protein conformations and can be used to predict the secondary structure of small peptides with a high degree of certainty. a-helical peptides, in particular, have characteristic
;;rtively
Spliced
Activator
A.
B.
2~
2r
, MW 105-
CREB
,,
ACREB
,
240 244 240 244
CREB 75-
ACREB
50-
C. Affinity purified CREB (Brain) I
+ I.
lmmunoprecipitate 240 Ab / supernafant
II.
lmmunoprectpitate 244 Ab
Figure
1
2. Comparison
of CREB
and ACREB
Expression
in Brain
(A) RNAase protection assays of total RNA from rat brain and hypothalamus compared with in vitro-synthesized CREB and ACREB RNAs. On the left is a schematic representation of antisense RNA probe (SP6) and CREB and ACREB RNAs. The size of protected fragments (in nucleotides) is indicated above. RNAase protection boundaries are indicated by broken lines. On the right, a 6% denaturing gel shows products of RNAase protection assay. Arrows point to fragments of the size predicted in the adjacent schematic. Hypo, Brain: total RNA (20 vg) from hypothalamus and brain. ACREB, CREB: RNAs synthesized in vitro from ACREB and CREB cDNAs, respectively. tRNA: RNAase assay using tRNA as a control for digestion. Probe: antisense CREB RNA probe prior to RNAase digestion. MW: molecular weight marker (Mspl digest of pBR 322) with fragment size indicated on the right). (B) Characterization of antiserum specific for a region of CREB. On the left is a diagram showing the specificity of 240 and 244 antisera for a region (residues 69-101) or the region common to CREB and ACREB
circular dichroism spectra consisting of an absorbance maximum at 192 nm with absorbance minima at 208 nm and 222 nm. The spectrum we obtained, containing an energy maximum at 192 nm and energy minima at 208 nm and 222 nm, demonstrated that this peptide forms a nonrandom structure with partial a-helical content. Furthermore, a helical wheel diagram of residues 90-96 in CREB suggests that the a peptide may contain an a helix that is amphipathic in nature, with primarily hydrophobic residues on one face and mostly basic residues on the other. Spectral analysis of a second CREB peptide extending from residues 121-159 failed to reveal any secondary structure (data not shown), however, suggesting that the results we observed with the a peptide were specific for this region of the protein. To test whether the secondary structure of the a peptide region was critical to CREB activity, we designed several a region mutants and tested these by transient assay in F9 cells (Figure 4b). CREB mutant M93 L/P contains a leutine to proline substitution at position 93 that would be predicted to disrupt helix formation within the a region. When inserted into an RSV expression vector and cotransfected into F9 cells with a C-subunit expression plasmid, the M93 UP mutant, like ACREB, was almost completely unable to stimulate expression of the CRE-CAT reporter plasmid. Distortion of a peptide secondary structure would appear, therefore, to destroy completely the contribution of this region to CREB activity. By contrast, a more conservative cystine to serine substitution at position 90, mutant M90 C/S, which would be predicted to preserve secondary structure, was 5-fold more active than M93 L/P and had activity intermediate between ACRE6 and CREB. As M90 C/S and M93 L/P mutants were expressed at comparable levels to the wild-type CREB plasmid, they were also targeted efficiently to nuclei of transfected cells (data not shown) and had normal DNA binding activity. The differences in Vans-activation potential between these proteins would therefore appear to result from alterations in a peptide structure and not from differences in expression, DNA binding, or nuclear localization. To determine whether the a region would stimulate ACREB activity when separated from its neighboring sequences, we inserted the 14 amino acid domain into position 23 of the ACREB protein instead of its normal position at amino acid 88 (Figure 5b). When tested by transient assay in F9 cells, this mutant, M23a+, caused a 21-fold increase in CRE-CAT activity and was 3 times more induc-
(residues 121-159), indicated by stippled box. On the right is an SDS-PAGE of 35S-labeled CREB and ACREB proteins synthesized in vitro and immunoprecipitated with either a region-specific 240 antiserum or 244 antiserum. MW (in kd) is indicated on the left. (C) Identification of CREB and ACREB proteins in brain using antisera 240 and 244. On the left is a schematic showing the strategy used to isolate CREB and ACREB proteins individually. Following purification by CRE oligonucleotide affinity chromatography, brain extract was labeled with C-subunit plus [Y-~‘P]ATP and immunoprecipitated with 240 antiserum. The supernatant fraction was immunoprecipitated with 244 antiserum, and pellet fractions from both reactions (I and II) were analyzed by SDS-PAGE (right). MW (in kd) is indicated on the left.
Cell 614
a
(:REB
A(ZREE!
b.
C. I
CREB CRE
Figure
3. Comparison
of CREB
i/2 CRE
and ACREB
”
ACREB 112 CRE CRE
Protein
240 ACREB’
’
Targeting,
DNA
Binding,
and Dimerization
Activities
(a) lmmunofluorescence analysis of CREB and ACRE6 proteins in F9 ceils following transfection with expression plasmids encoding either form. Nomarski (N) or immunofluorescence (IF) micrographs following treatment of transfected cells with CREB antiserum 244 are shown. Arrows point to the nuclei of antigen that contain cells. (b) Gel shift assays of bacterial IacZ-CREB or ACREB fusion proteins using double-stranded CRE or half-site mutant (l/2 CRE) probes. I and II indicate putative dimer and monomer forms of CREB, respectively. (c) Glutaraldehyde cross-linking assay of ACREB protein synthesized in vitro. Asterisk indicates that protein was %-labeled. Lane 1, immunoprecipitation of %-labeled ACREB protein with antiserum 244 after glutaraldehyde treatment. Lane 2, immunoprecipitation of ACREB protein with a region-specific 240 antiserum. Lane 3, immunoprecipitation of %-labeled ACREB protein plus unlabeled CREB protein with 240 antiserum after glutaraldehyde treatment. MW, molecular weight in kd.
ible by C-subunit than ACREB. Although M23a+ was less potent than wild-type CREB protein, the ability of the a region to enhance ACREB activity when removed from its normal context indicates that this domain can indeed function autonomously. The a region thus defines both a
structural and a functional motif within the CREB molecule. Having determined that the a region participates in rfans-activation as a functional unit, we sought to understand whether, and in what manner, this domain interacts
Alternatively 615
Spliced
Activator
a.
Figure 4. Analysis Function
r,
,96
190
b.
200
220
210
230
240
250
NANOMETERS
’
101 Gln Ser Ser CyS LyS Asp Leu Lys Arg Iau Phe Ser Gly Thr Gln CAG TCT TCC TGT AAG GAC TTA AAA AGA CTT TTC TCC GGA ACT CAG
07
CREB aPEPTIDE:
J.
J
TCI .Ser MUTANTS:
CCA prq M93 UP
M90 C/S
of cr Region
Structure
and
(a) Helical wheel diagram (residues 90-96) and circular dichroism spectrum (residues 69-101) of synthetic peptide containing the 14 amino acid a region. 0, degrees of rotation (x 10s) shown as a function of wavelength (in nm). +, - indicates charge characteristics of polar amino acids. (b) CAT assay of wild-type and mutant CREB expression plasmids following transfection into F9 cells. The sequence of amino acid substitution mutants M90 C/S and M93 UP in relation to wild-type CREB protein in a region from amino acids 69101 is indicated. CAT assay of various plasmids is shown below. C, control transfection without CREB plasmids showing endogenous CREB activity in F9 cells. -, + indicates whether C-subunit (+) or control 6-globin (-) expression plasmid was cotransfected with CREB.
I--II PKA
-+-+-+-+-+ -C-
CONSTRUCT
%REBJ
C
L ACREB’
CREB
ACREB Construct
‘M90 C/S’ ’ M93 UP’
M90 C/S
M93 UP
. with the pK-A phosphorylation site to confer CAMP responsiveness. To monitor a activity in isolation, we used the CREB mutant Ml, in which the pK-A phosphorylation site is destroyed by a Ser-133 to Ala-133 substitution (Gonzalez and Montminy, 1989). Following transfection into F9 cells, CREB mutant Ml stimulated CRE-CAT expression 3-to 5-fold higher than ACREB, having basal activity similar to wild-type CREB (Figure 5a). In unstimulated cells, CREB would appear to have basal activity that is therefore dependent upon the a region. The ratio of CREB to ACREB basal activity over six separate experiments averaged 3:i. When cotransfected with the C-subunit plasmid, the ratio of CREB to ACREB activity increased to lO:l, suggesting that the a region not only imparts constitutive activity to CREB but also participates in C-subunitstimulated activity with the pK-A phosphorylation site. In stimulated cells, therefore, C-subunit-mediated activation
of CREB may rely on a cooperative interaction between the Ser-133 phosphoacceptor and the a region. Based on our hypothesis that CREB is subject to allosteric regulation, pK-A-mediated phosphorylation of CREB may cause structural changes in the protein that affect the ability of the a region to interact with other proteins in the RNA polymerase II transcription complex. Indeed, the presence of multiple basic residues within the a peptide suggests a strong potential for ion pair formation with the Ser-133 phosphate. Such ion pairs have been shown to account for the conformational changes induced by pKA-mediated phosphorylation of glycogen phosphorylase A (Sprang et al., 1988). Do these regions need to interact in cis, or can CREB mutants lacking either region complement one another in trans? To test whether the a region and pK-A motif could interact in this manner, we cotransfected F9 cells with CREB mutant Ml (which lacks the
Cdl
616
BASAL
C 0.4%
1
1
CREB 5.65%
Figure 5. Analysis of a Region and pK-A Motif Interactions (a) CAT assay showing activity of CREB, ACREB, and Ml expression plasmids (in F9 cells) in the absence of cotransfected C-subunit plasmid. C, control transfection of F9 cells with CRECAT reporter in the absence of added CREB or C-subunit plasmids. The percent of conversion is indicated below. (b) Tins-activation potential of CREB-related constructs when cotransfected with C-subunit plasmid. Absolute activity for each construct is normalized to wild-type CREB activity (100%). Values shown represent an average of five separate experiments. In all cases, CAT activity was measured in extracts after normalizing to ft-galactosidase activity from a cotransfected RBV-6gal control plasmid. The schematic on the left shows a diagram of each protein. o denotes a region, P signifies the pK-A phosphorytation motif, and the stippled box indicates DNA binding and dimerixation region. The M23o+ plasmid was constructed by inserting the a region into ACREB at position 23 instead of its normal position at amino acid 66. The arrow indicates that o region was moved in an N-terminal direction. ACREB + Ml indicates that these plasmids were cotransfected into F9 cells.
ACTIVITY
1
ACREB 0.85%
1
Ml 4.0%
1
b. PK-A DEPENDENT ACTIVITY P v
a
+e
CREB ACREB
100 12.4 12.8 31.6
16
pK-A motif) and ACREB (which lacks the a motif). When compared with CREB, the Ml + ACREB transfection remained lo-fold less active (Figure 5b), thus suggesting the necessity for a cis interaction between the pK-A motif and the a region. Although this cis interaction may reflect a direct interaction between basic residues of the a region and the Ser-133 phosphate, the a region may also influence the secondary structure of the pK-A motif without such ion pairing.
Conclusions A functional role for ACREB protein in the CAMP signaling process remains elusive. The absence of a domain so critical for transcriptional activation leads us to suspect that ACREB may serve to downregulate the response to CAMP, perhaps by serving as a competitive inhibitor of CREB. Indeed the ability of ACREB to form heterodimers with CREB would further support this regulatory scheme (Figure 3~). In this case, the synthesis (or degradation) of ACREB and CREB proteins should be differentially regulated by CAMP. Alternatively, ACREB may stimulate transcription in a promoter-specific manner. Furthermore, the presence of additional phosphorylation sites in CREB (i.e., casein kinase II and kinase C) suggests that ACREB
activity may be regulated through cellular mechanisms other than CAMP In summary, we have described two proteins, CREB and ACREB, that differ only by the insertion of a 14 amino acid a region. ACREB is lo-fold less active yet 5-fold more abundant than CREB protein. It appears likely that CREB and ACREB mRNAs may arise by alternative splicing of a common precursor. Preliminary Southern blotting analysis of rat genomic DNA using fragments confined to the CREB activator domain suggests the presence of two or three related sequences. Nevertheless, final proof of this mechanism awaits characterization of genomic clones for CREB. Our studies also define a novel type of transactivation domain whose activity is dependent upon two distinct regions of the CREB protein: the a region and the pK-A phosphorylation site. Structural analysis of the CREB protein will allow us to distinguish individual amino acids that interact with the Ser-133 phosphate to confer this activity. Experimental
Procedures
Cloning and Plasmid Constructions ACREB cDNA was obtained after screening a Xgt 11 PC-12 library with a CREB cDNA probe as previously described (Gonzalez et al., 1969). The RBVACREB expression plasmid was constructed by inserting the
Alternatively 617
Spliced
Activator
ACRE6 cDNA, which contains the entire coding region (961 nucleotides) plus 100 bp of 5’ and 52 bp of 3’ untranslated sequences, into the Rous sarcoma virus expression vector RSV-SG (Gonzalez and Montminy, 1989). CREB mutants M90 C/S and M93 UP were constructed by PCR amplification of a 300 bp CREB cDNA fragment using 39-base antisense oligonucleotide primers that contain appropriate mutations and a sense oligonucleotide that corresponds to the 5’ noncoding region. Amplified fragments were digested with restriction enzyme BspMll and inserted into a CREB cDNA vector. The PCRamplified regions for each mutant were then sequenced on both strands. Mutant M23a+ was constructed by annealing sense and antisenae oligomers that encode the u peptide sequence from amino acids 89-101 and inserting the double-stranded oligo into a Hincll site of the ACREB cDNA. . RNAase Protection Asssys Total RNA was prepared from rat brain and hypothalamus to Chomczynski and Sacchi (1987). CREB antisense RNA prepared as previously described (Gonzalez and Montminy, RNAase protection assays were performed according to (1963).
according probe was 1969) and Zinn et al.
Affinity Purification and Phosphorylatlon of CREB CREB protein was prepared from rat brain extracts and purified by CRE oligonucleotide affinity chromatography (Yamamoto et al., 1966). CREB was labeled with purified C-subunit of kinase A (kindly provided by S. Taylor) as described previously (Gonzalez and Montminy, 1989). Peptide Synthesis, AntIsera, and Ciwlar Dichrolsm Analysis CREB (121159) and a peptides were synthesized using primary sequence data in Figure la. The a peptide consists of residues 89-101 plus Gly-Tyr at the C-terminus. Both peptides were conjugated to human a-globulins, and antisera 240 and 244 were then prepared as previously described (Gonzalez et al., 1989). For circular dichroism studies, a peptide (90 ug/ml) was analyzed in .Ol M phosphate buffer (pH 7.2) plus 40% trifluoroethanol. Secondary structure was estimated using the AVIV program, which is based upon an algorithm and reference spectra of J. T. Yang. The program makes a statistical fit of three reference spectra representing a helix, 6 sheet, and 6 turn. Cell Lines, l?ansfections, and lmmunofluorwcence Studies F9 teratocarcinoma cells were maintained in Dulbeccds modified Eagles medium (DMEM) supplemented with 10% fetal calf serum. Transfections and immunofluorescence studies were performed as described previously (Gonzalez and Montminy, 1989). CAT activity was measured from cell extracts after normalizing to 6-galactosidas.e activity derived from cotransfected RSV+gal plasmid. In Vitro Translatlons, Cross-Llnklng, and Bacterial Extracts CREB and ACREB cDNAs were inserted into pGEM vectors and transcribed in vitro with T7 polymerase. Transcripts were then translated in reticulocyte lysate extracts using rj5Sjmethionine. For cross-linking studies, labeled proteins were incubated at 2ooC with 005% glutaraldehyde for 30 min. Cross-linked proteins were then immunoprecipitated and complexes were resolved by SDS-PAGE. Bacterial extracts containing various CREB-related proteins were prepared and analyzed as described previously (Gonxalez et al., 1989). Acknowledgments We thank T Hunter, M. McKeown, B. Sefton, and I. Verma for commentson this manuscript. We also thank J. Andrews for circular dichroism spectra, J. Porter and R. Kaiser for peptide synthesis and purification, and J. Vaughan for characterization of antisera. This work was supported by National Institutes of Health grants GM 37828 and DK 26741. 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 U.S.C. Section 1734 solely to indicate this fact. Received
October
27, 1989; revised
December
12, 1989.
References 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, hf., Birnberg, N. C., Seascholtz, A., Herbert, E., and Goodman, H. M. (1986). A cyclic-AMP and phorbol ester-inducible DNA element. Nature 323, 353-356. Dwarki, V. J., Montminy, M. R., and Verma, I. M. (1989). Both the basic region and the “leucine zipper” domain of CREB protein are essential for transcriptional activation, EMBO J., in press. Gonzalez, G. A., and Montminy, M. R. (1989). Cyclic AMP stimulates somatostatin gene transcription by phosphorylation of CREB at serine 133. Cell 59, 675-680. 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. Hoeffler, J. P, Meyer, T E., Yun, Y., Jameson, J. L., and Habener, J. F. (1988). Cyclic-AMP-responsive DNA-binding protein: structure based on a cloned placental cDNA. Science 242, 1430-1432. Montminy, M. R., and Bilezikjian, protein to the cyclic-AMP response Nature 328, 175-178.
L. M. (1967). Binding of a nuclear element of the somatostatin gene.
Montminy, M. R., Sevarino, K. A., Wagner, J. A., Mandel, G., and Goodman, R. H. (1986). Identification of a cyclic-AMP responsive element within the rat somatostatin gene. Proc. Natl. Acad. Sci. USA 83, 6682-6686. 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, 2801-2806. Riabowol, K. T, Fink, J. S., Gilman, M. Z., Walsh, R. H., and Feramisco, J. R. (1988). The catalytic dependent protein kinase induces expression of CAMP-responsive enhancer elements. Nature 336,
D. A., Goodman, subunit of CAMPgenes containing 83-86.
Sprang, S. R., Acharya, K. R., Goldsmith, E. J., Stuart, D. I., Varvill, K., Fletterick, R. J., Madsen, N. B., and Johnson, L. N. (1988). Structural changes in glycogen phosphorylase induced by phosphorylation. Nature 336, 215-221. Yamamoto, K. K., 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 p-interferon 34, 865-879.
of two disgene. Cell
