ABSTRACT
The maize (
Zea mays
L.) endosperm specific transcription factor, encoded by the
Opaque-2
(
O2
) locus, functions
in vivo
to activate transcription from its target promoters.
O2
regulates the expression of a major storage protein class, the 22 kDa zeins,
and of a type I ribosome inactivating protein, b-32, during maturation phase endosperm development. The coding sequence of
O2
, which indicates it to be a member of the basic region-leucine zipper (bZIP) class of DNA-binding proteins, contains a number of regions rich in either
proline or acidic residues which are candidates for activation domains. In
functional assays using tobacco mesophyll protoplasts, the level of
transactivation conferred by a series of O2-deletion constructs was tested using as a reporter a fusion of the b-32 target promoter to
[beta]
-glucuronidase (GUS). The results indicate that O2 has a single acidic
activation domain, located near the N-terminus of the protein (amino acids 41-91). The ability of a shorter part of this domain (amino acids 39-82) to confer transactivation was also demonstrated in
domain swapping experiments, using fusions of the O2 polypeptide sequence to
the DNA-binding domain of the parsley (
Petroselinum crispum
) transcription factor CPRF1.
An essential component of gene regulation is transcriptional activation. The
process is dependent on the formation on regulatory DNA elements of
multiprotein complexes, which include transcription factors responsible for
cell-type specific and temporal gene regulation. They direct the
transcriptional machinery, composed of RNA-polymerase II and of additional factors, to the start site of gene
transcription (
1
). The identification of distinct domains within the polypeptides of
transcription factors has allowed the investigation of interactions with other
proteins present in initiation complexes. Domains reported to be responsible
for transcriptional activation contain acidic, glutamine-rich or proline-rich stretches of amino acids (
2
). These activation domains interact with other factors including TFIID or
TFIIB, and enable or facilitate the binding of the RNA polymerase II at the
start site for mRNA synthesis.
The
Zea mays
(L.) protein Opaque-2 (O2) is a member of the bZIP transcription factor family. It is involved
in the regulation of seed storage protein synthesis, modulating the
transcription of the 22 kDa [alpha]-zeins (
3
,
4
) and of the b-32 albumin (
5
). The 22 kDa [alpha]-zeins are the most abundant endosperm storage proteins in maize.
The function of the b-32 albumin in endosperm development remains to be established, but the
protein shares homology to type I ribosome-inactivating proteins (
6
).
O2
, which is located on chromosome 7, has been cloned by transposon tagging (
7
,
8
). A number of distinct functional domains have been identified on the O2
protein. In addition to the basic region followed by a leucine zipper, which is
responsible for DNA-binding and dimerization (
9
,
10
), two nuclear localization signals have been identified on the polypeptide (
11
).
Our goal was to identify the transcriptional activation domain of O2 by deletion
experiments in a transient expression system. To confirm that the region
identified was sufficient to confer transcriptional activation in the absence
of other parts of the sequence, domain swapping experiments were carried out.
We demonstrate that O2 has an acidic activation domain located near the N-terminus of the peptide and show that the ability of this domain to
activate transcription is retained in a hybrid protein composed of the O2
acidic activation domain and the bZIP domain of the parsley (
Petroselinum crispum
Bernh. ex Rchb.) common plant regulatory factor 1 (CPRF1) (
12
).
Effector constructs
. Derivatives of O2 were expressed under the control of the cauliflower mosaic
virus (CaMV) 35S promoter using pRT100 (
13
). The O2 sequences were fused at the 3'-end to the polyadenylation site of CaMV. C-terminal deletions were prepared from the vector pCaMVO2 (
5
), a plasmid containing the full length O2 cDNA (
9
), by using the natural restriction enzyme sites within the coding sequence of
O2. pO21-412 was constructed by cutting with
Bam
HI and religation (Fig.
1
), whereas pO21-284 and pO21-265 were prepared by cutting with
Sal
I/
Xba
I and
Eco
RI respectively, followed by religation. The leucine zipper of these two
constructs is truncated: pO21-265 contains two leucine repeats, pO21-284 four repeats instead of seven. During the construction of N-terminal deletions of O2, the translation initiation start codon was lost
and replaced by the ATG codon of the
Nco
I site of the pRT100 vector. For the construction of pO2192-460, pCaMVO2 was digested with
Xba
I and
Stu
I, and the resulting 1 kb fragment was inserted in the filled-in
Bam
HI site of pRT100. pO2125-460 was derived from inserting the 510 bp O2 fragment of
Bst
EII/
Sal
I digested pCaMVO2 in the
Sal
I/
Sma
I restriction sites of pO2192-460. In order to delete acidic region 3, a
Kpn
I site between this domain and the bZIP region was inserted by directed
in vitro
mutagenesis at aa 226. To introduce the site, the 300 bp
Pst
I/
Sal
I fragment of O2 was cloned in M13mp19 (
38
) and the mutagenesis was carried out using the oligonucleotide 5'-TCTTTCCTCGGTA
The internal deletions of O2 were made as follows. pO2[Delta]92-192 and pO2[Delta]142-192 were prepared by cutting pCaMVO2 with
Kpn
I/
Stu
I thereby removing the N-terminal part of O2 and religating with the 200 bp O2-fragment of
Kpn
I/
Nar
I digested pCaMVO2 or the 400 bp O2 fragment of
Kpn
I/
Xho
I-digested pCaMVO2. pO2[Delta]40-192 was made by religation of
Bst
EII/
Stu
I digested pCaMVO2. By digesting pO2[Delta]40-192 with
Bst
EII and
Kpn
I and religating the deletion construct pO2[Delta]1-40 [Delta]92-192 was generated. All deletion constructs were checked by sequencing to verify
fusion junctions.
The various O2 domains used to construct the O2-CPRF1 hybrids were cloned by PCR from pCaMVO2. The oligonucleotides for
the amplification reaction (Table
2
) carried an additional restriction site, an
Asp
718 site for 5' primers and a
Bsp
EI site for 3' primers with the exception of #4352, which carried a
Bst
EII restriction site. The PCR products were gel-purified and cloned via the appropriate restriction sites into the
effector plasmid pF1bZIP (kindly provided by M. Sprenger MPI f. Züchtungsforschung, Cologne, Germany). This vector is designed for the
expression in plants of polypeptide fusions of the CPRF1 DNA-binding domain [amino acids (aa) 258-337;
15
]. The gene fusion is flanked by the CaMV 35S promoter and the
nos
terminator (Fig.
3
). The O2 peptide sequence is cloned downstream of the nuclear localization
signal (NLS) of the simian virus 40 large-T antigen (SV40 NLS) and upstream of the DNA-binding domain of the parsley transcription factor CPRF-1. The plasmids constructed were designated pO212-228F1bZIP (oligonucleotides #1832 and #2402), pO212-197F1bZIP (#1832 and #1829), pO212-123F1bZIP (#1832 and #1693), pO239-228F1bZIP (#1398 and #2402), pO239-197F1bZIP (#1398 and #1829), pO239-147F1bZIP (#1398 and #3009), pO239-123F1bZIP (#1398 and #1693), pO239-82F1bZIP (#1398 and #4355), pO283-228F1bZIP (#4356 and #2402), pO2124-228F1bZIP (#2703 and #2402) and pO2203-228F1bZIP (#2401 and #2402). The effector pO239-82/124-228F1bZIP was made by inserting the amplification product of #1398 and #4352 in the
Asp
718 and
Bst
EII restriction sites of pO39-228F1bZIP. The numbers in subscript are the amino acid coordinates of the O2-ORF (
9
).
Table 1
For expression of the SV40 NLS-O2 region-CPRF1 bZIP triple fusion hybrid in
E.coli
, the plasmid pO2 39-197F1bZIP was used. The portion of O2 was selected to omit the proline-rich segment at the N-terminus of O2 as the instability of fusion polypeptides containing
this region suggested it was acting as a PEST site (
40
,
41
). The coding sequences were amplified by PCR for cloning in the
E.coli
expression plasmid pTrxFus (Invitrogen, Netherlands) using the primers #3099 (5'-GTGGTCGACCCACCGGTTGCATTTTTC-3'; containing a
Sal
I site) and #3100 (5'-CGTGGATCCTATGACGCACAATCCCAC-3'; containing a
Bam
HI site). The 5' primer starts at the SV40 NLS, the 3' primer at the end of the bZIP domain of CPRF1 in pF1. The PCR
product was gel-purified and after digestion with
Bam
HI and
Sal
I, integrated in the
Bam
HI/
Sal
I sites of pTrxFus and designated pTrxFusO239-197- F1bZIP (Fig.
3
). The DNA of all constructs was sequenced to verify fusion junctions.
The transient expression of the plasmids in tobacco protoplasts cv. Petit Havana
SR1 (
41
) was performed as described by Negrutiu
et al
. (
42
). Ten [mu]g each of the effector and reporter plasmid DNA was used for 3 * 105 protoplasts. The transformed protoplasts were harvested and extracts were
prepared for measurement of GUS activity as described by Jefferson (
39
). GUS activity in the extracts was measured using the fluorogenic substrate 4-methyl-umbelliferylglucuronide (4-MUG; Sigma) and measured in a luminescence spectrometer LS30
(Perkin Elmer). Each assay was carried out with extracts containing 5 [mu]g protein, estimated according to Bradford (
44
). The relative GUS activity were calculated for equimolar concentrations of the
constructs.
The ThioFusiontm expression system (Invitrogen, Netherlands) was used for expressing an
O2-CPRF1 hybrid as a C-terminal fusion to mammalian thioredoxin. The
E.coli
expression plasmid with the integrated O2-CPRF1 domains, pTrxFusO2 39-197F1bZIP, was transformed in
E.coli
GI724 by electroporation [25 [mu]F, 1600 V, 200 [Omega] using the GenePulsertm (BioRad)]. The thioredoxin fusion protein located to the
periplasmic space, allowing purification by an osmotic shock procedure (
45
).
SDS-PAGE was carried out as described (
46
). For immunoblot analysis the proteins were transferred electrophoretically
onto nitrocellulose overnight at 4oC using an immersed sandwich setup and the filter was blocked for 1 h in T-TBS [0.5% (v/v) Tween 20, 20 mM Tris-HCl pH 7.6, 137 mM NaCl] at room temperature containing 5%
(w/v) non-fat dried milk. Incubation with rabbit anti-O2 serum (1:5000) in T-TBS for 1 h at room temperature was followed after three
washing steps by incubation with horseradish peroxidase coupled anti-rabbit antibodies (1:5000). After washing for three times in T-TBS, bound antibody was detected using enhanced chemiluminescence
(ECL Western Blotting Detection System, Amersham) and the resulting image
recorded on Kodak Xomat-AR X-ray films. Rabbit polyclonal anti-O2 antibodies were raised to the
E.coli
-expressed O2 protein and kindly provided by Dr J. Muth (MPI f. Züchtungsforschung, Cologne, Germany).
Binding of pTrxFusO239-197F1bZIP proteins to a b-32 promoter fragment was carried out using 1 ng end-labelled DNA (5 * 104 c.p.m.) and the fusion protein (amounts as stated in legend of Fig.
4
C) for 15 min at room temperature in 20 [mu]l binding buffer (20 mM Tris-HCl pH 8.0, 1 mM EDTA, 50 mM KCl, 10% glycerol, 2 [mu]g sonicated salmon sperm carrier DNA, 0.1 [mu]g BSA). The binding assay was loaded on a native 3.5% PAA gel
in 0.25* TAE and the electrophoresis was carried out at 10 V/cm. The gel was
dried on Whatman paper and exposed on autoradiographic films (Kodak Xomat-AR). The b-32 fragment from -265 to -16 upstream of the TATA-box was isolated as previously described (
5
).
To characterise the transcriptional activation domain of O2 a transient
expression system in tobacco mesophyll protoplasts was used. The b-32 promoter is highly responsive to expression
in trans
of Opaque-2 protein in this assay, thus permitting a fine structure analysis which
would not have been practicable in the homologous system (
5
). Different regions of O2, expressed from the CaMV 35S promoter, were tested
for their ability to transactivate a reporter plasmid, consisting of the b-32 promoter fused to [beta]-glucuronidase (GUS), upon cotransfection. Every O2 construct
(Fig.
2
) retained the bZIP domain to ensure specific DNA-binding and nuclear localization properties were always present in the
protein (
9
,
11
).
Assays of O2 activity in a number of heterologous systems in yeast and tobacco,
and transient expression in the homologous system, maize, show that O2 is able
to confer transcriptional activation in a variety of environments (
10
,
22
-
24
). O2 has also been shown to partially substitute GCN4 function in yeast cells (
25
). These observations suggest that an activation domain is an integral function
of the O2 protein.
It is known from comparisons of different bZIP proteins (
9
) and from the analysis of o2 mutants (
10
) that the DNA-binding region of O2 is located in the basic region between aa 228 and
247. One of the two nuclear localization signals (NLS) of O2 lies in the same
region (aa 223-254), while the second NLS is located between aa 101 and 135 (
11
). In this study, the region responsible for transcriptional activation has been
located using deletion constructs of O2 retaining the bZIP region, and with the
aid of domain swapping analysis. C-terminal deletions which did not affect the leucine-zipper (pO21-284 and pO21-412) were able to transactivate as well as O2 WT (pO21-460). The deletion derivative pO21-284, which lacks the entire region C-terminal to the bZIP domain, transactivated 2-fold higher than the WT protein. The truncated O2 polypeptide may be
sterically improved for efficient transactivation, or it may lack a domain
capable of repressing transcription. A significant reduction in transactivation
was seen only with either the removal of the leucine repeat or with deletions
of the N-terminal part of O2. In the N-terminal region, one proline-rich and three acidic-rich putative activation domains were identified on the
basis of sequence comparisons (
2
). The proline-rich sequence (8/19 consecutive aa are proline), has similarities to the
proline-rich activation domain of CTF/NF1(
26
), and is located at the N-terminus, between aa 13 and 31 (Fig.
2
). Acidic domain S1, covering aa 42-71, has a net negative charge of -8. Acidic domain S2, between aa 102 and 122, has net charge of -4. The last acidic domain, S3, is just upstream of the bZIP
domain at aa 204-227 and has a net charge of -8). S3 does not function as an activation domain when present as the only
acidic domain in the O2 polypeptide (constructs pO2[Delta]125-460 and pO2[Delta]192-460 in Fig.
2
). Similarly, the proline-rich domain alone did not confer on the O2 fusion the ability to
transactivate (construct pO2[Delta]40-192 in Fig.
2
). Addition to the basal construct (pO2[Delta]193-460) of a fragment containing both acidic domains 1 and 2 (pO2[Delta]142-192), however, increased transactivation to 70% of the wild-type activity. Further deletions demonstrated that addition of acidic
domain 1 alone to the basal construct raises the transactivation seen to 48% of
the WT activity.
Interestingly, acidic domain 1 has sequence homology to a region of the maize
transcription factor C1 which has also been shown to function as an activation
domain (
14
). The homology of these acidic regions between O2 and C1 proteins, which lack
further homology elsewhere in the sequences, suggests structural conservation,
although acidic activation domains in general share no significant sequence
homology. The addition of acidic domain 2 to the acidic domain 1-containing effector construct resulted in an increase of the activation
from 48 to 70% of WT level. The increase could result from improved spacing
between acidic region 1 and the DNA-binding domain which might allow a more productive interaction with
proteins of the transcription machinery. The results obtained are summarized in
the modular structure of O2 shown in Figure
6
.
The promoter of the maize b-32 gene is inactive in tobacco leaf protoplasts in the absence of
exogenous transactivator (
5
). EMSA assays were performed to test whether the parsley CPRF1 DNA-binding domain could bind to the b-32 promoter, since if this were the case, fusions containing this
DNA-binding domain could also be assayed with this system. The experiments
showed an
in vitro
binding of the O2-CPRF1 hybrid to a 300 bp b-32 promoter fragment which contains an ACGT core element (5'-TGACGTGA-3') (
9
). This ACGT-element is not perfectly palindromic, but binding of bZIP proteins to non-palindromic motifs has been demonstrated, for example for the
Arabidopsis transcription factor GBF1 (
27
), or the wheat bZIP protein EmBP-1 (
28
). The ACGT-core element could therefore function as a target for CPRF-1 which has been shown to bind to different ACGT-elements (
19
,
20
). The DNA-binding domain of CPRF1 may also be capable of binding to non-ACGT elements, as has been demonstrated for O2 (
5
,
29
) and Em-BP1 (
28
). The assumption that the DNA-binding domain of CPRF1 binds to the O2 target sequence(s) within the b-32 promoter was supported by its inability to transactivate the
reporter pb32*-GUS, in which the O2 target sites have been mutated (Fig.
5
).
The N-terminal segment of O2 can stimulate transcription independent of its own
DNA-binding domain (Fig.
5
), a feature common to most activation domains described so far. The same
transactivation as that obtained for O2 is seen for the truncated version pO212-228F1bZIP. The proline-rich domain of O2 is not required for transcriptional activation in
the CPRF1 fusion (construct pO2 39-228), confirming the result of the O2-deletion series. Together, the results obtained rule out a major role in
transcriptional activation of either the acidic domain 3 (construct pO2 203-228) or the proline-rich domain.
In contrast, constructs containing acidic domain 1 were able to activate target
promoters to give 50% of the activity elicited by the complete O2 region when
fused N-terminal to the bZIP domain of CPRF1. A C-terminal extension of this O2-fragment stimulated the transactivation further: the O2-region with the acidic domains 1 and 2 (pO2 39-123F1bZIP) conferred 60% and pO2 39-147F1bZIP 70% of the activity of the control construct. Although the fragment from acidic
domain 2 through acidic domain 3 increased transactivation when fused with
acidic domain 1, this fragment alone could not stimulate the reporter
significantly over background activity, and therefore does not represent a
further autonomous activation domain. The deletion of the acidic domain 2 alone
reduced transactivation only slightly to 90% of that observed with the control
construct. The disparity of 50% activity seen for pO2 39-82F1bZIP, which contains only the S1 domain, compared to 100% for the control construct
could be due either to the absence of additional domains which function only in
the presence of the 39-82 region, or to suboptimal spacing or presentation of the activation
domain to the transcriptional machinery in the O2[Delta]39-82 protein. However, as no effector constructs containing regions outside of S1
were able to activate measurably, the existence of additional independent
activation domains is unlikely. We have not seen large additive effects for
segments beside S1, making it unlikely that O2 has an extended activation
domain analogous to that reported for GCN-4 (
30
), which consists of the N-terminal one-third of the protein. Bobb
et al
. (
31
) have also reported a 243 aa acidic activation domain for a plant transcription
factor, PvAlf; however, a more detailed dissection of this region may yield a
smaller fragment sufficient for function. In contrast, Goff
et al.
(
14
) show that a 10 aa peptide is sufficient to activate transcription in plants
when fused to the GAL4 DNA-binding domain. The 80 aa acidic domain from VP16 and a proline-rich segment from the plant transcription factor GBF1 have both also
been shown to activate transcription in tobacco protoplasts (
27
,
32
).
The spacing between the DNA-binding domain and the activation domain seems to be critical for the
level of activation conferred by a transcription factor (
33
,
34
). Support for a spacing effect for S1 is supplied by construct pO2[Delta]39-82/124-228F1bZIP, which does not contain the acidic region 2. It increased the relative activity
of this construct by 40% over that of pO2[Delta]39-82F1bZIP, though the spacing fragment from 124 to 228 alone was not able to
transactivate over background (pO2[Delta]124-228F1bZIP ).
The domain swapping experiments (Fig.
5
) confirm the findings of the O2 deletion series that O2 activates transcription
by an acidic domain located at the N-terminus, and that this activation does not depend on the presence of a
particular O2
cis
-acting sequence. The significance of this lies in the ability of O2 to
form heterodimers with other transcription factors (
35
), which may give rise to transcriptional activation from promoters not
recognized by O2 alone. The ability of O2 to transactivate when located at
varying distances to the TATA box has been demonstrated by altering the
position of the binding site in the reporter construct (
36
). O2 polypeptide can be phosphorylated
in vitro
by nuclear extracts of maize endosperm or recombinant casein kinase II of
Arabidopsis (
37
, our unpublished results). The major activation domain identified in O2
contains a potential casein kinase II phosphorylation site at Ser45, which may
play a role in modulating activation from this protein.
We thank Drs M. Sprenger and B. Weisshaar (MPI f.Züchtungsforschung, Köln) for the gift of the vectors pF1bZIP, pBTACEchsII-4GUS and pF1bZIPVP16AD. The pRT100 vector and the O2 antibody were provided by Drs R. Töpfer and J. Muth. Dr K. Shinozaki (RIKEN, Tsukuba, Japan) generously
supplied the Arabidopsis Casein kinase II-expression construct. The research was supported by the Deutsche
Forschungsgemeinschaft (SFB 274) and the European Communities BIOTECH programme
as part of the project of technological priority (PTP) 1993-1996.
REFERENCES
Return