ABSTRACT
The C
2
-H
2
zinc-finger is a widely occurring DNA binding motif, usually present as tandem
repeats. The majority of C
2
-H
2
zinc-finger proteins that have been studied are derived from animals. Here, we characterize a member of a distinct class of plant C
2
-H
2
zinc-finger proteins in detail. A cDNA clone encoding a DNA binding protein
from
Arabidopsis
was isolated by SouthWestern screening. The protein, termed ZAP1 (Zinc-dependent Activator Protein-1), is encoded by a single copy gene, which is expressed to similar
levels in root and flower, to a somewhat lower level in stem and to low levels
in leaf and siliques. The optimal binding site was determined by random binding
site selection, and the consensus sequence found is CGTTGACCGAG. The homology
between ZAP1 and other DNA binding proteins is restricted to a repeated region
of a stretch of 24 highly conserved amino acids followed by a zinc-finger motif (C-X
4
-C-X
22-23
-H-X
1
-H). The C-terminal zinc-finger region is essential for DNA binding, whereas deletion of the N-terminal one resulted in 2.5-fold reduced binding affinity. Binding of ZAP1 to
DNA was abolished by metal-chelating agents. The activation domain as determined in yeast is adjacent
to and possibly overlapping with the DNA binding domain. Particle bombardment
experiments with plant cells showed that ZAP1 increases expression of a
gus
A reporter gene that is under control of ZAP1 binding sites. We conclude that
ZAP1 is a plant transcriptional activator with a C
2
-H
2
zinc-finger DNA binding domain.
Transcriptional regulation of gene expression is determined by the interaction of transcription factors with sequences in the promoter region.
Besides general transcription factors that assemble at the TATA-box (
1
), a variety of sequence-specific DNA binding proteins have been studied, that are necessary for
inducible or high levels of transcription. Transcription factors can be classified according to their DNA binding domain (
2
). A widely occurring DNA binding motif is the so-called zinc-finger (
3
). The term zinc-finger applies to a rather diverse set of protein motifs. These motifs
have in common that zinc-ions interact with histidines and/or cysteines to stabilize a small
functional protein domain or `finger'. In addition to DNA binding, zinc-fingers may play a role in protein-protein interactions, as observed for the LIM domain (
4
).
Several cDNA clones encoding different classes of zinc-finger DNA binding proteins have been isolated from plants (
5
). Recently a novel class of DNA binding proteins has been found, that contain a DNA binding domain predicted to form a special type of C
2
-H
2
zinc-finger structure. This motif, only found in plants so far, is present in
SPF1 from sweet potato (
6
) and ABF1 and ABF2 from wild oat (
7
). A stretch of 40 amino acids of the DNA binding domain is duplicated in the N-terminal part of SPF1 and ABF1. SPF1 binds to promoter sequences of the
sporamin and [beta]-amylase genes and is thought to play a role in sucrose-inducible gene expression (
6
). ABF1 and ABF2 bind to [alpha]-amylase 2 promoters and may play a role in the induction of [alpha]-amylase expression during germination (
7
).
For many different plant transcription factors the exact sequence requirement for DNA binding have been determined. The most extensively studied
binding sites are for bZIP proteins (
8
). They contain a core ACGT (
8
) or AC/GT (
9
) sequence with different flanking sequences. For one group of bZIP proteins,
the C-box binding proteins, the binding sites contain TGAC half sites. This
sequence can also be found in binding sites for an unrelated plant zinc-finger protein (
10
) and for animal nuclear hormone receptors (
11
). Thus, the TGAC sequence is present in binding sites for different classes of
DNA binding proteins and forms part of DNA elements that are likely to confer
different expression patterns.
For transcriptional activation, the activation domain of a sequence-specific DNA binding protein needs to interact directly or indirectly via co-activators with the general transcription machinery (
12
). Several different types of activation domains have been identified and are classified as acidic, glutamine rich or proline rich.
The mechanisms by which these different activation domains function involve
protein-protein interaction with one or more components of the basal transcription machinery, thereby recruiting the basal machinery to the promoter or increasing the
rate of a kinetically slow step (
13
,
14
). Several plant DNA binding proteins were shown to act as transcriptional
activators (
5
). Only for very few of these proteins have the activation domains been
localized. The transcriptional activator C1 from maize (
15
), regulating anthocyanin biosynthesis, and a myb protein from snapdragon (
16
), regulating flavonoid biosynthesis, contain acidic activation domains.
Here, we describe the characterization of a DNA binding protein from
Arabidopsis,
termed ZAP1, that is homologous with SPF1 from sweet potato and ABF1 and ABF2
from wild oat. The optimal binding site was determined and the activation and
DNA binding domains were mapped. For binding to its optimal sequence, CGTTGACCGAG, ZAP1 requires metal-ions. Since the DNA binding domain contains a C
2
-H
2
motif, we postulate that it forms a zinc-finger. Furthermore, we show that ZAP1 can function as a sequence-specific transcriptional activator in plant cells.
A cDNA expression library in [lambda]ZAP (Stratagene), made on flowers and 0 to 4-day-old siliques of
Arabidopsis thaliana
(ecotype C24), was screened with a mixture of three probes: 3W1 (
17
), W2 (
9
) and IWT (
18
) (Table
1
). Lambda phages (20 000 p.f.u./150 mm plate) were grown on
E.coli
XL-1 blue for 4 h. Nitrocellulose filters saturated with 10 mM IPTG were
placed on the plates and phages were allowed to grow for another 4 h. Filters
were blocked for 1 h in binding buffer [20 mM HEPES/KOH (pH 7.2), 40 mM KCl, 1
mM EDTA, 0.5 mM DTT, 10% glycerol] supplemented with 5% non-fat dry milk at room temperature. After washing twice in binding buffer
the filters were incubated in binding buffer containing 2 ng/ml probe and 5 [mu]g/ml sonicated calf thymus DNA for 1 h. Probes were end-labeled using the Klenow fragment of DNA polymerase I and [[alpha]-
32
P]dNTPs. Finally, the filters were quickly washed 3 times and dried before autoradiography.
Table 1
The
Zap1
cDNA insert was subcloned and sequenced with a Pharmacia T7 sequencing kit. Nucleotide sequence data were collected, assembled and analysed with a VAX computer using the Genetics Computer Group Sequence Analysis Software Package (
19
).
For genomic Southern hybridization 1 [mu]g DNA was digested, electrophoresed on a 0.8% agarose gel, blotted and
hybridized as described (
20
). For Northern hybridization 10 [mu]g total RNA were electrophoresed on a 1.5% formaldehyde gel, blotted and
hybridized as described (
20
). As probes randomly labeled cDNA inserts were used. Blots were washed with 0.1* SSPE, 0.1% SDS at 42oC.
Gel shifts were performed essentially as described (
9
). DNA (10-20 fmol), 3'-end labeled using the Klenow fragment of DNA polymerase I,
was incubated with 50 ng of crude extract from
E.coli
expressing ZAP1.
Random binding site selection was essentially done as described (
21
). A pool of oligonucleotides containing a stretch of 20 random nucleotides (TCTAGAACTAGTGGATCC-N
20
-CGATACCGTCGACCTCG) were annealed with KS primer (CGAGGTCGACGGTATCG) and
extended with the Klenow fragment of DNA polymerase I. One [mu]g ds-probe and 5 [mu]g crude extract from
E.coli
expressing ZAP1 was allowed to form complexes. Five ng of [[alpha]-
32
P]dCTP-labeled probe was included in the binding reaction in order to follow the
complex during further procedures. After binding, the complex was separated
from the free probe on a 5% non-denaturing polyacrylamide gel. The gel was dried and exposed for 2 h to X-ray film. Complexes were cut out from the dried gel and DNA was
eluted overnight in 0.5 M NH
4
Ac (pH 8.0), 1 mM EDTA. DNA was phenol/chloroform extracted and 0.1 ng was used
for PCR amplification for 15 cycles in the presence of [[alpha]-
32
P]dCTP using SK (TCTAGAACTAGTGGATC) and KS primers. In the following rounds of binding 25, 2.5 and 0.5 ng PCR amplified DNA were
used, respectively. After the fourth round DNA was amplified, digested with
Bam
HI and
Sal
I and cloned in pBluescript SKII. The inserts of 96 different plasmids were
amplified by PCR using SK and KS primers and were used as competitors in gel
shift assays. About 50% of the analysed inserts were found to compete and were
used as gel shift probes after removal of most of the flanking sequences by
digestion with
Bam
HI and
Sal
I.
For functional analysis of ZAP1, deletion derivatives were produced using
selected restriction sites present in the cDNA (Fig.
8
). In D2 the
Nco
I site (accession number X92976, position 819) was fused to the
Sac
I site (position 1301) after generation of blunt ends by the Klenow fragment of
DNA polymerase I and T4 DNA polymerase, respectively. In D3 the
Hin
fI site (position 344) was fused to the
Bam
HI site (position 776), separated by polylinker sequence. In D4 the
Hin
fI site (position 344) was fused to the
Sac
I site (position 1301) after generation of blunt ends by the Klenow fragment of
DNA polymerase I and T4 DNA polymerase, respectively. In D5 the
Kpn
I site (position 1047) was blunt-ended with T4 DNA polymerase and fused to the
Rsa
I site (position 1639) present in the 3' non-coding region, in order to retain the polyA tail. The D5 encoded
protein contains nine additional amino acids at its C-terminus, from the
Rsa
I site to the first stop codon. In D6 the region 5' to the
Kpn
I site was deleted.
35
S-labeled ZAP1 and deletion derivatives were synthesized using the TnT system of Promega. Products were analysed on a 10% SDS-polyacrylamide gel and quantified with a phosphorimager (Molecular Dynamics). Equimolar amounts of protein were used for
gel shifts.
N-terminal deletions of
Zap1
were cloned in pAS1[Delta]Nco, a derivative of pAS1 (
22
), and introduced into yeast strain Y153 (
22
). Yeast transformants were grown on nitrocellulose filters, frozen in liquid N
2
and stained for [beta]-gal activity by incubating the filters on paper impregnated with 0.3 mg/ml X-gal in 0.1 M Na-phosphate-NaOH (pH 7), 10 mM KCl, 1 mM MgSO
4
, 50 mM [beta]-mercaptoethanol.
The synthetic binding site 65 (BS 65) was multimerized. The monomer, dimer,
trimer and hexamer were cloned in plasmid GusSH-47 (
23
).
Zap1
cDNA missing the first 150 bp of the leader sequence was cloned under control
of the CaMV 35S promoter and the
Rbc
S-3C terminator.
Catharanthus roseus
cell suspensions were grown in LS medium containing 2 mg/l NAA and 0.2 mg/l kinetin at 25oC with light/dark intervals of 16/10 h and subcultured weekly. Four days after
subculture, 1 ml of suspension was collected on Whatman filter paper using a Büchner funnel. Paper discs were transferred to Petri dishes containing
solidified medium and left overnight in the growth chamber.
Particles were prepared using a procedure modified from (
24
). The precipitation mixture included 2 mg gold particles (Aldrich), 25 [mu]g plasmid DNA, 8% PEG 1500, 5 mM spermidine in a total volume of 575 [mu]l. After each addition the suspension was mixed by sonication. After 10
min whirl mixing at 4oC, gold particles were collected by centrifugation at 1000 r.p.m. for 5 min
at 4oC and resuspended in 20 [mu]l supernatant. The suspension was sonicated for 2 s and 2 [mu]l was pipetted on the support screen of the loader of a home-made helium gun. A steel mesh was used at a distance of 3 cm
below the loader. Petri dishes containing the paper filter with the cells were
placed 9.5 cm below the loader and bombardment was performed at 35 mbar with
2,5 bar helium pressure. After 16-20 h the cells were overlaid with 250 [mu]l GUS staining solution (
25
) containing 6 mM 5-bromo-4-chloro-3-indolyl glucuronide (X-gluc) and incubated for 1 h at 37oC, followed by an overnight incubation
at 28oC.
For each bombardment experiment constructs were delivered in triplicate. This was repeated once or twice depending on the construct. GUS
activities obtained by one representative bombardment experiment were
quantified on black and white video images of the plates by measuring dark staining of a fixed area, using Image Quant
tm
software. The value obtained after bombardment with the control plasmid GusSH-47 was subtracted from all other values. The value obtained after bombardment with the GusSH-47 derivative containing the BS65 monomer was set at 100.
In a SouthWestern screening of an
Arabidopsis thaliana
(ecotype C24) cDNA expression library with TGAC half site-containing binding sites (C-boxes and G/A box hybrids), different classes of DNA binding
proteins were obtained. The sequences of these clones were compared with databases. One clone was homologous to previously isolated cDNAs from sweet potato (
6
), wild oat (
7
) and cucumber (accession number L44134) and to EST sequences of
Arabidopsis
(accession numbers T44598 and T45479) and rice (accession number D38923). The
clone from sweet potato, termed
SPF1
, was described to encode a novel type of DNA binding protein. The
Arabidopsis
cDNA clone is 1888 bp long and contains a polyA-tail of 62 bp. It contains an open reading frame encoding a protein of 463
amino acids in length with a predicted molecular mass of 51.2 kDa. The most
striking feature of the putative polypeptide is the presence of two regions of
58 amino acids, which are 56% identical. The homology with SPF1 is highest in
these repeated regions (Fig.
1
). They are 67 and 75% identical, with the highest homology in the N-terminal part of each conserved sequence. The protein encoded by the cDNA
clone isolated from sweet potato is 99 amino acids longer at the N-terminus. The
Arabidopsis
cDNA is full-length since there are in frame stop codons preceding the predicted start
codon. We have designated the
Arabidopsis
protein ZAP1 (Zinc-dependent Activator Protein 1).
To explore the expression of
Zap1
in different
Arabidopsis
organs, we performed Northern blot analysis. Hybridization with total RNA from
root, stem, leaf, flower and siliques showed expression mainly in root and
flower (Fig.
2
A). The length of the detected transcript is ~2.0 kb. The expression in stem was somewhat lower and in leaf and siliques
the expression was hardly detectable. The blot was reprobed with ubiquitin cDNA
to show the presence of intact RNA in all lanes (Fig.
2
B).
The cDNA clone encoding ZAP1 was isolated by SouthWestern screening with a
mixture of odd base C-boxes and G/A-boxes [3W1 (
17
), W2 (
9
) and IWT (
18
)], which are medium to high affinity binding sites for bZIP proteins. To
determine the DNA binding specificity of ZAP1, these probes and some related
ones (Table
1
) were used in a gel shift experiment, using crude extract from
E.coli
expressing ZAP1. Figure
4
shows that ZAP1 binds to 3W1 and the shorter derivative W2. These probes both
contain an odd base C-box (
9
). Lower affinity was observed for 4A1 (
26
), a tetramer of the as-1 sequence of the CaMV 35S promoter. No binding to IWT was observed, but
the affinity for its mutant derivative IMU (
18
) was relatively high. The sequences that show binding to ZAP1 have the C-box half site sequence TGAC in common.
ZAP1 protein contains a repeated sequence (56% identity), that is also conserved
in ABF1 and ABF2 of wild oat (
7
) and in SPF1 from sweet potato (
6
). The C-terminal part of these repeated regions contain putative zinc-finger motifs (C-X
4
-C-X
22-23
-H-X
1
-H) somewhat different from previously described zinc-fingers. Binding of ABF1 and ABF2 is abolished by the metal-chelating agent 1,10-
o
-phenanthroline and by EDTA (
7
). We tested whether these agents also abolish binding of ZAP1 (Fig.
7
). No complex formation was observed after addition of 10 mM 1,10-
o
-phenanthroline (in 10% ethanol final concentration) or 50 mM EDTA to the binding
reaction, whereas addition of 10% ethanol or 50 mM EGTA (calcium-chelating agent) did not abolish ZAP1 binding. As a control we tested the
effect of these metal-chelating agents on binding of proteins that do not require metal-ions. Binding of the bZIP proteins RITA-1 (
27
) to 3W1 and TGA2 (
28
) to 4A1 was not affected by 1,10-
o
-phenanthroline or EDTA (results not shown). These results show that ZAP1
requires metal-ions for its binding activity. The presence of zinc-finger motifs in ZAP1 suggests that this metal is zinc.
The C
2
-H
2
zinc-finger motif is usually present as tandem repeats (
2
). Most of the members of the class of DNA-binding proteins to which ZAP1 belongs also contain a repeated zinc-finger motif. ABF2 from wild oat, however, contains only one zinc-finger motif (
7
). SPF1 contains two motifs but only the second one is required for DNA binding
(
6
). We tested whether both motifs of ZAP1 are necessary for DNA binding, employing
in vitro
generated proteins. The wild-type and deletion constructs (D2-D6) shown in Figure
8
were introduced in an
in vitro
transcription/translation system and labeled proteins were analysed on a 10%
SDS-polyacrylamide gel (Fig.
9
A). In all cases proteins of the correct size were produced. With the shortest
construct (D6) an additional polypeptide with lower mobility was visible, which
was also produced in low amounts in the transcription/translation system
programmed with the other
Zap1
constructs and with unrelated genes and is not encoded by the cDNA insert.
Equimolar amounts of these proteins were used in a gel shift assay. The optimal
binding site for ZAP1 (BS65; Fig.
5
) was used as probe. The amount of complex formed with the full-length protein (Fig.
9
B) was set at 100% (Fig.
8
). The only other protein that gives a visible complex is encoded by D3. It
forms 42% of the wild-type amount of complex. In D3 the N-terminal conserved sequence is deleted and the C-terminal one is intact. The D2 protein, in which the C-terminal repeated sequence is deleted but the N-terminal one is intact, forms only 3% of the wild-type complex and D5 protein, also missing
the C-terminal repeated sequence, forms 2% of the wild-type complex. These values are in the range of background levels.
The other constructs (D4 and D6), which are missing both conserved regions also
do not produce proteins that are able to bind to the probe. From these results
we conclude that the C-terminal conserved region is essential for DNA binding and that the N-terminal one increases the binding 2.5-fold. The N-terminal conserved region alone does not allow complex
formation above background level.
ZAP1 binds DNA
in vitro
and activates gene expression
in vivo
in yeast. To test whether ZAP1 can activate gene expression in plant cells,
particle bombardment experiments were done with
C.roseus
cell suspensions. The reporter constructs (Fig.
10
A) contained a monomer or multimers of BS65 (Fig.
5
) coupled to the TATA box of the CaMV 35S promoter and the
gus
A reporter gene (GusSH-47). GusSH-47 and 35S-GUS were used as negative and positive control constructs, respectively. The reporter
constructs were introduced in
C.roseus
cells together with
ZAP1
cDNA under control of the CaMV 35S promoter or together with an empty vector in
order to keep the amount of DNA constant.
This paper describes the characterization of the DNA binding protein ZAP1 from
Arabidopsis
. The
Zap1
cDNA clone was isolated by SouthWestern screening using a mixture of bZIP
protein binding sites. ZAP1 is not a bZIP protein but a putative member of a
distinct class of DNA binding proteins: the C
2
-H
2
zinc-finger proteins. The optimal binding site for ZAP1 has a 4 bp homology with the probes used for the screening of the library. Low
affinity of ZAP1 for these probes made it possible to isolate the
Zap1
cDNA during this SouthWestern screening experiment.
The sequence of the ZAP1 DNA binding domain and the requirement of transition
metal ions for DNA binding suggests that a zinc-finger is formed. Two zinc-finger motifs are found in ZAP1 and several other members of the
class of proteins to which ZAP1 belongs. Nevertheless, only one motif is
required for binding [Fig.
8
and (
6
)]. DNA binding activity was also observed for a truncated derivative of the
Petunia
EPF1 protein containing only the C-terminal zinc-finger (
10
). In this respect, these proteins differ from other zinc-finger proteins, in which single base-pair mutations in one of three zinc-finger modules present in the DNA-binding domain abolished binding to their cognate
binding sites (
29
,
30
,
31
). Whereas zinc-fingers generally are adjacent to each other, in ZAP1, SPF1 and EPF1 they are separated by long stretches of 140, 145 and 61 amino acids, respectively, and
therefore possibly function as independent DNA binding units. The regions N-terminal of the zinc-finger motifs of ZAP1 and SPF1 are highly conserved and are probably
also involved in DNA binding.
For both SPF1 and ZAP1 the C-terminal conserved region is essential for DNA binding, whereas deletion
of the N-terminal one only has a minor effect. Apparently, the C-terminal region or its flanking regions contain some extra
information that is missing in the other one. Examination of the structural
features of the conserved regions according to the Chou-Fasman (
32
) and Garnier-Osguthorpe-Robson (
33
) methods reveal similar patterns. Both contain a stretch of amino acids that probably forms an [alpha]-helix followed by a stretch that probably forms a [beta]-sheet. These structures are conserved in the repeats of
SPF1. However, outside the repeat the three dimensional structure diverges. Thus the difference might reside in the flanking regions of the repeats. On the other hand a single
amino acid difference might cause the first conserved sequence to be non-functional. This has also been found for DNA-binding regions of bZIP proteins (
34
). Exchanging (parts of) both conserved sequences may answer the question whether the functional difference is located in the flanking regions or within the conserved sequences themselves.
The region between 210 and 285 contains (part of) the activation domain, since its deletion from a GAL4-ZAP fusion resulted in loss of activation (Fig.
8
). The activation domain of ZAP1 is not acidic (pI 7.9), proline-rich or glutamine-rich, as is often observed for activation domains of other proteins.
Possibly the secondary structure determines its activation ability. It has been
suggested for the activation domain of GAL4 that it forms a [beta]-sheet structure (
35
). This putative structure rather than the negative charge is thought to be
important for its function (
36
). The region in ZAP1 that activates transcription contains a stretch of 13
amino acids (270-282), which might form a [beta]-sheet (
32
,
33
). In analogy with the model for GAL4, these [beta]-sheet structures in ZAP1 could form the activation domain.
The homology between the DNA binding domains of ZAP1 and related proteins suggests that the proteins bind to similar DNA sequences.
The optimal binding site for ZAP1 is CGTTGACCGAG. This sequence shows
similarity to the binding site for ABF1 and ABF2 in the [alpha]-amylase promoter (A
Although ZAP1 and related proteins recognize similar DNA sequences, they may be involved in different biological processes, since the parts of the proteins outside the DNA-binding domains are not homologous and may have different functions. Furthermore, their expression patterns are different.
ZAP1
is mainly expressed in root and flower (Fig.
2
).
ABF1
and
ABF2
mRNAs are accumulating in germinating seed (
7
), where the encoded proteins are thought to control [alpha]-amylase expression. Although SPF1 binds to the promoters of sweet potato tuber-specific genes encoding [beta]-amylase and sporamin genes, it is ubiquitously
expressed (
6
).
Comparison of the ZAP1 consensus binding sequence with plant promoters revealed a striking homology with an elicitor-responsive element (AATTGACC) of a maize
PRms
gene (
37
). A similar sequence is present in promoters of several other genes that are expressed in response to fungal infection and elicitor treatment (
37
). This could mean that ZAP1 or a related protein is involved in the expression
of elicitor inducible genes. An indication about the biological process in
which ZAP1 plays a role might be obtained by ectopic expression or inactivation
of the gene.
The characterization of ZAP1 described in this paper contributes to the knowledge of this novel class of zinc-dependent DNA binding proteins. Since
Arabidopsis
contains at least two other genes with regions homologous to the conserved zinc-finger motif (EST sequences) it would be interesting to study these as
well as homologous genes from other organisms in more detail.
Note added in proof
While this paper was in press, three cDNAs from parsley were reported encoding
members of the class of zinc-finger DNA binding proteins described in this paper. The parsley proteins
bind to elicitor response elements in the promoters of PR1 genes and their mRNA
levels show rapid changes upon elicitor treatment [Rushton
et al
. (1996)
EMBO J.
15
, 5690-5700].
We thank Annemarie Meijer for pAS1[Delta]Nco and other control plasmids and Peter Hock and the Biological Photographic Service, Leiden University, for preparation of the figures. SdP and KP were supported by the Netherlands Organization for Applied Scientific Research
TNO, VG by a fellowship of the Consiglio Nazionale delle Ricerche (CNR) and JM
by a Fulbright fellowship.
*To whom correspondence should be addressed. Tel: +31 71 527 4803; Fax: +31 71
527 4863; Email: sbtnbp@rulsfb.leidenuniv.nl
+
Present address: Istituto di Ricerca sulle Biotecnologie Agroalimentari, CNR,
Via Prov. le Lecce-Monteroni, 73100 Lecce, Italy
Name
Repeat sequence
Number
of repeats
3W1
TCGACGTGA
3
W2
GTGA
4
IWT
G
4
IMU
TGAC
TGTTCT
TGAC
TGTTCT
2
4A1
C
TGAC
GTAAGGGA
TGAC
GCAC
4
REFERENCES
Return