(Received for publication, February 26, 1997, and in revised form, April 21, 1997)
From the Laboratory of Molecular Neurobiology, The
W. M. Burke Medical Research Institute, Cornell University Medical
College, White Plains, New York, New York 10605 and the
¶ Department of Neurology and Department of Anatomy and
Neurobiology, University of Tennessee, College of Medicine,
Memphis, Tennessee 38163
Expression of tyrosine hydroxylase (TH) is
limited to catecholamine-producing neurons and neuroendocrine cells in
a cell type-specific manner and is inducible by the cAMP-regulated
signaling pathway. Previous results indicated that the cAMP response
element (CRE) residing at 45 to
38 base pairs upstream of the
transcription initiation site is essential for both basal and
cAMP-inducible promoter activity of the 2.4-kilobase or shorter
upstream sequence of the TH gene (Kim, K. S., Lee, M. K., Carroll, J.,
and Joh, T. H. (1993) J. Biol. Chem. 268, 15689-15695; Lazaroff, M., Patankar, S., Yoon, S. O., and Chikaraishi,
D. M. (1995) J. Biol. Chem. 270, 21579-21589). Here,
we further report that the CRE is critical for the promoter activity of
the 5.6- or 9.0-kilobase upstream sequences of the rat TH gene, which
had been shown to direct the cell-specific TH expression in
vivo. To define the structure/function relationship of the CRE in
transcriptional activation of the TH gene, we performed saturated
mutational analyses of 12 nucleotides encompassing the CRE. Mutation of
any nucleotide within the octamer motif results in a significant
decrease of both basal and cAMP-inducible transcriptional activity of
the TH reporter gene construct. Among the four nucleotides adjacent to
the CRE (two 5
and two 3
), only the G residue at the immediate 3
position is important for full transcriptional activity. DNase I
footprint analysis indicates a positive correlation between in
vivo promoter activity and in vitro interaction
between the CRE motif and its cognate protein factor(s). Reconstruction
experiments using a TH promoter in which the native CRE was rendered
inactive show that the CRE can transactivate transcription in either
orientation through a window of approximately 200 base pairs upstream
of the transcription initiation site, suggesting that CRE supports
transcriptional activation of the TH gene in a
distance-dependent manner. Finally, when the distance between the CRE and TATA box was changed by inserting an additional 5 or 10 bases, it was observed that both insertional mutations increased
activity by approximately 3-fold. The cAMP inducibility was as intact
as the wild type construct. Together, these results are consistent with
a model in which transcriptional activation of the TH gene by the CRE
requires that it be located within a certain proximity of the CAP site
but does not depend on a stringent stereospecific alignment in
relationship to the TATA element.
Neuronal plasticity, essential for the adaptive function of the
nervous system, depends on the capacity of cells to alter expression of
target genes in response to environmental stimuli. Many stimuli alter
cellular activity via receptors coupled to adenylate cylase. The
resultant increases of the intracellular cAMP levels stimulate
cAMP-dependent protein kinase, resulting in phosphorylation
of the target molecules such as cAMP response element binding protein
(CREB),1 which is an essential step in
transcriptional activation by this protein factor (3, 4). In numerous
cAMP-inducible eukaryotic genes, an octamer DNA motif with the
nucleotide sequence 5-TGACGTCA-3
, termed cAMP response element (CRE),
mediates transcriptional induction by the cAMP-regulated signaling
pathway (5, 6). Recent evidence strongly suggests that the
cAMP-regulated signaling pathway plays essential roles in adaptive
functions of the nervous system in that CREB-dependent gene
transcription is critically involved in learning and memory (reviewed
in Ref. 7) as well as in behavioral manifestations of opiate dependence
(8).
Tyrosine hydroxylase (EC 1.14.16.2; tyrosine 3-monooxygenase; TH) is an
important brain enzyme because it catalyzes the conversion of
L-tyrosine to 3,4-dihydroxy-L-phenylalanine,
which is the first and rate-limiting step of catecholamine biosynthesis (9). TH is selectively expressed in catecholamine-synthesizing and
secreting cells, including dopaminergic, noradrenergic, and adrenergic
neurons in the central nervous system and sympathetic ganglia and
adrenal chromaffin cells in the periphery. The 5-flanking sequence of
the rat TH gene contains a consensus CRE motif located at
38 to
45
bp upstream of the transcription initiation site (10). Earlier studies
of the TH promoter activity by Chikaraishi and colleagues (11-13)
using the PC8b cell line as the positive system showed that the AP1
motif and an overlapping E box/dyad, located at
205 and
182 bp, is
the most important basal and cell-specific promoter element. In these
studies, the CRE was suggested to be unimportant for TH transcription.
However, more recent evidence from several laboratories using other
TH-positive cell lines supports the idea that the cAMP-regulated
signaling pathway, via the CRE, importantly regulates transcription of
the TH gene. First, transient expression assays in TH-expressing cell
lines such as the human neuroblastoma SK-N-BE(2)C and rat PC12 (1) or
mouse CATH.a and PATH.a (2) show that the CRE is critical for both
basal and cAMP-inducible transcription of the rat TH gene. Secondly, TH
gene expression is significantly attenuated at the transcriptional level in several cAMP-dependent protein kinase-deficient
PC12 cell lines (14). Finally, coexpression of the catalytic subunit of
cAMP-dependent protein kinase dramatically increases the
transcriptional activity of the rat TH gene in a
dose-dependent manner, whereas coexpression of the specific
cAMP-dependent protein kinase inhibitor blocks cAMP-stimulated
induction and reduces basal transcriptional activity (15).
To understand TH gene regulation by the CRE in greater detail, we investigated the structure/function relationship of the CRE in transcriptional activation. To define individual nucleotides that are important for transcriptional activational function, we performed saturated mutagenesis of the 12-bp region of the rat TH gene that encompasses the CRE and examined the effect of individual mutation on basal and cAMP-inducible transcription by transient transfection assay using TH-expressing SK-N-BE(2)C cell line. In vivo promoter activity was correlated with in vitro interaction between the CRE and transcription factors by comparing the DNase I footprint patterns of the wild type and CRE-mutated upstream sequences using nuclear extracts of the SK-N-BE(2)C as well as purified CREB protein. We also determined whether the position of the CRE relative to the TATA box can affect its transcriptional activation function. Surprisingly, this study indicates that transcriptional activation by the CRE is distance-dependent but does not require a stringent stereospecific alignment in relation to the TATA element.
Human neuroblastoma SK-N-BE(2)C cells (16) were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.), containing 10% heat-inactivated fetal calf serum (Hyclone Lab.), 100 units/ml of penicillin (Life Technologies, Inc.), and 100 µg/ml of streptomycin (Life Technologies, Inc.). CATH.a cells (17) were grown in RPMI 1640 medium (Life Technologies, Inc.) supplemented with 10% horse serum (Hyclone Lab), 5% fetal bovine serum, and 100 units/ml of penicillin, and 100 µg/ml of streptomycin. Cells were maintained in a humidified 5% CO2 incubator at 37 °C.
TH-CAT Fusion ConstructsTH2400CAT fusion construct contained 2.4-kb upstream sequences of the rat TH gene fused to the bacterial CAT gene as described (1). Longer reporter gene constructs, TH9000CAT and TH5600CAT, containing 9.0- and 5.6-kb upstream sequences, respectively, were also generated. TH9000CAT was constructed by replacing the 1.6-kb HindIII-XhoI DNA fragment of TH2400CAT by an 8.3-kb HindIII-XhoI DNA fragment of pTH9.0 (18). A 3.4-kb SphI DNA fragment was then deleted from TH9000CAT, resulting in TH5600CAT. To make CRE mutant forms of these longer constructs, TH2400(41C40G)CAT, which contains a double mutation within the CRE (15), was used in place of TH2400CAT in the above procedure.
Site-directed mutagenesis was performed based on the procedure of
Nakamaye and Ekstein (19) as described previously (1). To generate
saturated mutations in the TH CRE, the present procedure employed a
degenerate 24-mer oligonucleotide,
5-GGCCAGG*C*T*G*A*C*G*T*C*A*A*A*GCCCCT-3
, in which the
asterisks denote degenerate bases with 5% substitution of each of the
other three nucleotides (Bio-Synthesis, Inc., Denton, TX). Among 136 individual clones isolated and sequenced, 56 were found to be mutants.
Of these mutant clones, 19 single substitution mutants and 11 double
mutants were further analyzed in this study. Other mutants were
constructed to examine the position dependence of the CRE in
transcriptional activation function. Oligonucleotides with sequences
5
-GAGGGGCTTTGACGTCAGCCTGG-3
and 5
-CCAGGCTGACGTCAAAGCCCCTC-3
, representing both strands of the CRE region at
54 to
32 bp of the
rat TH gene, were annealed and inserted at BglII (
168 bp), BamHI (
503 bp), XhoI (
773 bp),
PstI (
2400 bp), or SmaI (+1700 bp) sites of the
42A38G mutant construct that contain a double substitution mutation
(see Fig. 2). The copy number and orientation of the inserted DNA were
confirmed by sequence analysis. In addition, oligonucleotides of the
sequences 5
-CGCCCTCTTTAAATGACGTCAAAGCC-3
, 5
-TCTTTAAAGGCCACGTGAGGCTGACGTCA-3
, and
5
-TCTTTAAAGGCCACGTGACTGCAGGCTGACGTCA-3
were used to generate
2400TH(
8)CAT, 2400TH(+5)CAT, and 2400TH(+10)CAT, respectively.
The modified sequences of the upstream region of these mutant reporter
genes were verified by DNA sequencing analysis.
Transfection and Enzyme Assays
Cells were transfected using
the calcium phosphate coprecipitation method (20) as described (21).
For SK-N-BE(2)C cells, 50% confluent 60-mm dishes were transfected
with 2 µg of the reporter construct, 1 µg of pRSV-gal, and pUC19
plasmid to a total of 5 µg of DNA. For CATH.a cells, doubled amounts
of plasmids were used. For the experiments described in Fig. 1, an
equimolar amount of each reporter construct was used for transfection.
Transfected cells were collected 36 h after transfection, and
activities of CAT and
-galactosidase were determined as described
elsewhere (20, 22). To ensure that all CAT assays were performed in the
linear range, the final sample volume was adjusted following preliminary assays so that chloramphenicol conversion amounted to
5-50%. The CAT activity was normalized by the
-galactosidase activity to correct for differences in transfection efficiency among
different DNA precipitates.
Nuclear Extract Preparation and DNase I Footprinting
Crude
nuclear extracts were prepared from SK-N-BE(2)C cells based on
described procedures (23). To obtain a CRE area probe labeled on the
noncoding strand, p2400THCAT and the 42A38G mutant construct were cut
with Bsu36I and end-labeled with Klenow enzyme and
[-32P]TTP according to procedures suggested by
vendors. The probe was then digested with NcoI and purified
by polyacrylamide gel electrophoresis as described previously (24).
Labeled probes (30,000 cpm) were incubated with nuclear extracts or
purified CREB protein, treated with DNase I, and resolved on a 6%
polyacrylamide/8 M urea-sequencing gel as described (21).
Location of the protected area was determined by Maxam-Gilbert
sequencing of labeled probes.
As
shown in Fig. 1A, double mutation within the
CRE diminished the promoter activity of the 2.4-kb sequence of the rat
TH gene as well as activities of longer 5.6- and 9.0-kb upstream sequences in the human neuroblastoma SK-N-BE(2)C cell line. These longer upstream sequences were shown to confer cell type-specific expression of reporter genes in transgenic mice (18, 25). The CRE is
likewise essential for the promoter activities of these upstream
sequences in another TH-expressing cell line, CATH.a (17), which had
been derived from the central nervous system (Fig. 1B). In
addition, these CRE mutant constructs do not respond to forskolin
treatment (data not shown), indicating that the CRE is required for
both basal and cAMP-inducible transcriptional activities of the 5 TH
upstream sequences.
Using site-directed mutagenesis procedure with a degenerate
oligonucleotide we isolated 19 different point mutations that include
at least one mutation in every position of twelve bases encompassing
the CRE motif and its four proximal bases (two 5 and two 3
to the CRE
octamer) (Fig. 2). The effects of these mutations on
basal transcription and cAMP inducibility were examined in the context
of the 2400 bp upstream sequence of the rat TH gene by the transient
transfection assay using the TH-expressing SK-N-BE(2)C cells. First,
mutation of any nucleotide within the CRE resulted in a significant
loss of basal transcription activity, demonstrating that the octamer
CRE motif of the TH gene is the functionally optimal sequence motif.
Intriguingly, mutations to different nucleotides at a particular
position appear to have differential effects; for instance, replacement
of C residue by T at
39 bp (39T in Fig. 2C) diminished
more than 90% of the basal transcription but an A substitution
(39A in Fig. 2C) retained approximately half of
the promoter activity. Substitution mutations of surrounding bases at
36,
46, and
47 bp show little or no effect on the promoter
activity, indicating that these nucleotide sequences do not affect the
CRE function. However, the G residue at
37 bp, immediately 3
to the
CRE, appears to be important because its change to an A or a T residue
results in a decrease of basal transcription activity by 60-70%. The
11 double mutations all exhibited a significant decrease of promoter
activity (Fig. 2). Residual activities of double mutants (at least 10%
of that of the wild type promoter) may arise from promoter activities of other elements, e.g. AP1 (1, 11). Finally, mutations that exhibit a significant decrease in basal transcription also exhibited a
significant defect in the cAMP inducibility. Conversely, mutant constructs with basal transcription activity >30% of the wild type
without exception showed cAMP inducibility equivalent to that of the
wild type construct.
Mutations of the CRE likely affect binding affinities of the cognate
transcription factors and thereby account for the decreased transcriptional activities. DNase I footprint analysis demonstrates that the nucleotides starting from 32 to
50 bp upstream of the transcription initiation site, which encompasses the CRE, are protected
by nuclear proteins when the wild type sequence is used as the probe
(Fig. 3A). It is to be noted that the TATA
region was not detectably protected in this footprinting assay,
indicating that the DNA-protein interaction at the TATA of the TH gene
is weak. This weak binding affinity of the TATA, at least in part, may
underlie the essential function of the CRE in TH transcription. The
corresponding CRE area of the mutant 42A38G was not protected at all by
the same nuclear proteins (Fig. 3B). This mutant construct contains base substitutions at
42 and
38 nucleotides, C to A and A
to G, respectively, and exhibited minimal promoter activity (Fig. 2).
Furthermore, purified CREB footprinted the wild type CRE area at the
identical region but not that of 42A38G mutant sequence (Fig.
3C), supporting the suggestion that CREB is a transcription factor acting at the TH CRE.
Positional Effects on Transcriptional Activation via the TH CRE
To address whether the CRE activates TH transcription in a
position- and distance-dependent manner, reconstruction
experiments were performed using the 42A38G mutant construct (Fig.
4A; see also Figs. 2 and 3). In this
experimental paradigm, we tested the effects of the location, copy
number, and orientation of the CRE motif on its ability to support the
basal transcription as well as the cAMP inducibility. As shown in Fig.
4B, when located at 168 bp upstream of the transcription
initiation site, the CRE supported both basal transcription and
cAMP-mediated induction of the reporter gene as efficiently as the wild
type construct. At the same locus, two copies of the CRE in opposite
orientation drive basal CAT expression at approximately twice the level
of the wild type construct. However, the level of cAMP inducibility was
the same as that of the native promoter. Additional copies appear to
exert no further increases in basal and cAMP-inducible transcription of
the reporter gene. Strikingly, the CRE did not support basal
transcription at further upstream regions, e.g. at
503,
773, and
2400 bp upstream of the transcription initiation site. In
addition, the CRE at these loci no longer mediated transcriptional induction following treatment with forskolin. To determine whether multiple copies of the CRE could overcome the incompetence of transcription activation at these loci, we inserted two and three copies of the CRE at
503 bp, two copies at
773 bp, and two and four
copies at
2400 bp. These constructs did not significantly support
basal transcription of the reporter gene. In contrast, cAMP
inducibility was restored to the level of the native promoter when
three and four copies of the CRE were inserted at
503 and
2400 bp,
respectively. When located 3
to the reporter gene at +1600 bp, a
single copy of the CRE supported neither basal transcription nor cAMP
inducibility of the reporter construct.
The TH CRE Does Not Require Stereospecific Alignment in Relation to the TATA Box for Transcriptional Activation
The CRE of the TH
gene resides 8 nucleotides upstream of the TATA box, showing the
closest proximity to the TATA box among known cAMP-regulated genes.
This unique spatial arrangement suggests that the stereospecific
alignment between the CRE and the TATA may be important for
transcription activation. To test this, we examined whether alterations
of the distance between the CRE and TATA affect transcriptional
activity (Fig. 5). Five nucleotides were inserted
between the CRE and TATA, adding a half helical turn in the reporter
construct TH2400(+5)CAT. Surprisingly, this mutant construct drove CAT
expression 2.8-fold more strongly than did the wild type reporter
construct. Insertion of 10 nucleotides between the CRE and the TATA,
which would produce approximately a full helical turn between them,
similarly increased basal CAT expression by 3-fold. The CAT expression
by these two spatial mutant constructs was induced to a level
comparable with that by the wild type construct following treatment
with forskolin. Taken together, the CRE of the TH gene does not require
stereospecific arrangement in relationship to the TATA box for its
transcriptional activation activity. Furthermore, our results suggest
that the native location of the CRE may contribute to determining the
level of basal expression of the TH gene.
TH gene transcription is inducible by the cAMP-regulated signaling
pathway (10, 26-28). Although there are some conflicting data in
regard to the critical role in TH transcriptional regulation (11-13),
probably due to cell line differences, the CRE, which resides at 38
to
45 bp upstream of the transcription initiation site, appears to be
responsible for not only the cAMP-mediated induction but also basal
transcription of the TH gene (1, 2, 15). The dual role of the CRE as a
basal and cAMP-inducible transcription element appears to be a common
regulatory feature for many CRE-containing genes (5, 29-34). To
understand TH gene regulation by the cAMP-regulated signaling pathway
in greater detail, this study defined the relationship between the
structure and function of the CRE in transcriptional activation of the
TH-CAT reporter gene.
Mutation of the CRE in the context of 9.0-, 5.6-, and 2.4-kb upstream sequences of the rat TH gene largely diminished their transcriptional activities in TH-expressing SK-N-BE(2)C and CATH.a cell lines (Fig. 1). Previous transgenic mice experiments demonstrated that longer upstream sequences of the rat TH gene, e.g. 4.8 or 9.0 kb, can direct expression of the reporter genes in a tissue-specific manner (18, 25). Based on our data showing that the CRE is critical for the promoter activities of the 5.6- and 9.0-kb upstream sequences in both TH positive cell lines, we conclude that the CRE may be important for in vivo TH gene transcription.
The present analysis (Fig. 2) showed that (i) base substitution of any
nucleotide within the octamer motif results in a significant decrease
of the promoter activity (55-90%), demonstrating that every base of
the octamer motif is important for intact transcription activation
function of the CRE; (ii) among the nucleotides adjacent to the TH CRE,
only the G residue residing immediately 3 to the CRE is important for
full transcriptional activity; (iii) mutation of a specific nucleotide
may have differential effects depending on the substituted nucleotide;
and (iv) a mutation that shows a significant reduction in basal
transcription no longer mediates an intact cAMP induction.
Although the upstream area at 32 to
50 bp of the wild type sequence
was footprinted by both nuclear proteins of SK-N-BE(2)C cells and
bacterially expressed CREB, that of the mutant 42A38G construct with
minimal promoter activity was not protected at all, indicating a
positive correlation between in vivo promoter activity and
in vitro DNA-protein interaction (Fig. 3). In addition, this
observation is consistent with the idea that CREB is the transcription
factor or one of the factors that bind to and transactivate the CRE. In
support of this, coincubation of nuclear extracts with anti-CREB
antibody results in formation of a supershifted band in electrophoretic
mobility shift assay (35).2 However, in
view of the increasing number of identified CRE-binding proteins
(reviewed in Refs. 36, 37), it is likely that additional protein
factors may be involved in binding to the CRE and regulating TH
transcription. Antibody coincubation experiments indeed suggested that
additional protein factors including ATF1 and CREM may bind to the TH
CRE.2 Further, forskolin treatment of PC12 cells and
reserpine treatment of rats induced expression of the inducible cAMP
early repressor, an isoform of the CREM family (38), which represses
the transcriptional activity of the TH gene promoter (35).
To address whether the CRE activates transcription in a
distance-dependent manner, an oligonucleotide encompassing
the TH CRE was inserted at different loci 5 or 3
to the reporter gene in the mutant 42A38G construct (Fig. 4). Strikingly, this
reconstruction experiment demonstrates that the CRE can support basal
and cAMP-inducible transcription in either orientation only through a
window of approximately 200 bp upstream of the transcription initiation
site. When three or four copies of the CRE are inserted at
503 and
2400 bp, respectively, they restore the forskolin inducibility but
still do not support basal transcription of the reporter gene.
The CRE is generally known to have properties of an enhancer (39). This
notion is based on earlier studies indicating that the CRE of the
proenkephalin gene can activate transcription when located in positions
relatively distant from the transcription initiation site or at 3 to
the gene (40). Our results thus contrast to the general notion that the
CRE can work in a distance- and position-independent manner. The
structure and function of the CRE of the proenkephalin gene differ from
those of the TH gene in that it is heptanucleotide with the sequence
5
-TGCGTCA-3
and mediates both cAMP and phorbol ester inducibility.
Thus, the mechanisms underlying the function of the octamer CRE and the heptamer CRE may be significantly different. It is to be noted that
most known CREs are located within the proximal 170 bp of upstream
region and have the octamer structure (Ref. 1 and references therein).
One obvious exception is the tyrosine aminotransferase gene, which has
a functional CRE at the
3650-bp position (41). Here, the CRE has the
same haptamer motif as that of the proenkephalin gene and is one of the
functionally interdependent components of a hepatocyte-specific
enhancer (41). In this context, one interesting possibility is that the
CRE activates TH transcription synergistically with other
cis-element(s), and this synergism works in a distance- and/or promoter
context-dependent manner. In support of this, we recently
found that mutation of a contiguous cis-regulatory element, Sp1, at
120 to
113 bp (11) diminished the basal promoter activity by 85%,
suggesting a synergistic activation of TH transcription by the CRE and
Sp1.2 This potential mechanism underlying the
distance-dependent activation of TH transcription is
currently being investigated.
Finally, we determined whether the location of the CRE relative to the TATA box is important for transcriptional activity (Fig. 5). For many genes that are inducible by hormone, the optimal inducibility and synergism often require stereospecific alignments (42, 43). Furthermore, stereospecific alignment of several promoter elements of eukaryotic genes relative to basic elements, e.g. TATA box, are known to be critical for full promoter activity (44-46). The CRE and the TATA elements are essential for TH gene transcription (1, 2, 11) and are separated from each other by 8 nucleotides. Insertion of either 5 or 10 nucleotides between these elements resulted in an increase of the basal promoter activity by approximately 3-fold, indicating that phasic interactions between the cognate binding proteins are not required for supporting basal transcription. This observation is thus consistent with a model that the CRE-binding protein(s) may activate TH transcription through a co-factor. CREB-binding protein, a recently identified transcriptional coactivator of CREB (47), is one such candidate molecule. CREB-binding protein, recruited to cAMP-dependent protein kinase-phosphorylated CREB by protein-protein interaction (48), may activate transcription of the TH gene, thus obviating the requirement for stereospecific alignment between the CRE and the TATA elements.
In summary, we defined the structure/function relationship of the CRE
in TH transcription by performing the saturated mutagenesis, reconstruction experiments, and insertional mutation analyses and by
examining their effects on the promoter activity in the context of the
2.4-kb upstream sequence. Our results for the first time show that
every nucleotide of the octamer of the CRE as well as the G base
immediately 3 are important for full basal and cAMP-inducible
transcriptional activity of the TH promoter. In addition, the CRE needs
to be located within a certain proximity in the upstream region for
exhibiting full transcriptional activation function, indicating that
the CRE supports transcriptional activation of the TH gene in a
distance-dependent manner. Furthermore, insertional mutation analyses suggest that the native spatial arrangement of the
CRE and TATA elements may contribute to determining the level of basal
TH gene expression in vivo. In support of these notions, the
nucleotide sequences and the location of the CRE in relation to the
TATA box are well conserved in all the TH genes identified from
different species (Fig. 6). Specifically, 12 nucleotides encompassing the CRE are strictly conserved in the TH upstream sequences of all six species thus far identified.
We thank Dr. M. Montminy (The Salk Institute) for providing the bacterially expressed CREB. We also thank Dr. T. S. Nowak, Jr., in the Department of Neurology for critically reading this manuscript.