Download The maize ID1 flowering time regulator is a zinc finger protein with

Published online March 12, 2004
1710±1720 Nucleic Acids Research, 2004, Vol. 32, No. 5
DOI: 10.1093/nar/gkh337
The maize ID1 ¯owering time regulator is a zinc
®nger protein with novel DNA binding properties
Akiko Kozaki1, Sarah Hake1 and Joseph Colasanti1,2,*
1
Department of Plant and Microbial Biology, University of California, Berkeley, and the Plant Gene Expression
Center, Albany, CA 94720, USA and 2Department of Molecular Biology and Genetics, University of Guelph,
Guelph, Ontario N1G 2W1, Canada
Received December 20, 2003; Revised February 10, 2004; Accepted February 19, 2004
ABSTRACT
INTRODUCTION
The C2H2-type zinc ®ngers are among the best characterized
DNA binding proteins of eukaryotic transcription factors and
have been shown to play important roles in speci®c development processes (1,2). The canonical C2H2-type zinc ®nger
motif is typi®ed by the amino acid sequence F/Y-X-C-X2-5-CX3-F/Y-X5-Y-X2-H-X3-5-H, where X represents any amino
acid and Y is a hydrophobic residue. Two cysteine (C) and
two histidine (H) residues tetrahedrally coordinate a zinc ion
to form a compact structure containing two b-sheets and an ahelix (bba) that interacts with the major groove of DNA in a
sequence-speci®c manner (3,4). Each C2H2-type zinc ®nger
motif consists of a module of approximately 30 amino acids
that recognizes a triplet of double-stranded DNA (dsDNA). In
*To whom correspondence should be addressed. Tel: +1 519 824 4120 ext. 58052; Fax: +1 519 837 2075; Email: [email protected]
Nucleic Acids Research, Vol. 32 No. 5 ã Oxford University Press 2004; all rights reserved
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
The INDETERMINATE protein, ID1, plays a key role
in regulating the transition to ¯owering in maize. ID1
is the founding member of a plant-speci®c zinc
®nger protein family that is de®ned by a highly conserved amino sequence called the ID domain. The ID
domain includes a cluster of three different types of
zinc ®ngers separated from a fourth C2H2 ®nger by
a long spacer; ID1 is distinct from other ID domain
proteins by having a much longer spacer. In vitro
DNA selection and ampli®cation binding assays and
DNA binding experiments showed that ID1 binds
selectively to an 11 bp consensus motif via the ID
domain. Unexpectedly, site-directed mutagenesis of
the ID1 protein showed that zinc ®ngers located at
each end of the ID domain are not required for binding to the consensus motif despite the fact that one
of these zinc ®ngers is a canonical C2H2 DNA binding domain. In addition, an ID1 in vitro deletion
mutant that lacks the extra spacer between zinc
®ngers binds the same 11 bp motif as normal ID1,
suggesting that all ID domain-containing proteins
recognize the same DNA target sequence. Our
results demonstrate that maize ID1 and ID domain
proteins have novel zinc ®nger con®gurations with
unique DNA binding properties.
animals, multiple zinc ®nger modules are linked in tandem
arrays, with each ®nger separated by a conserved short
sequence of seven amino acids known as the H/C link. In these
cluster-type zinc ®nger proteins key amino acid residues in the
a-helix of each ®nger interact with a triplet of base pairs (5,6).
In plants, the C2H2 class of zinc ®nger proteins is one of the
largest families of transcription factors to be identi®ed,
although they appear to not be as prevalent as C2H2 proteins
in animals (7,8). Genetic and functional analyses of various
genes that encode plant zinc ®nger proteins show that many
are involved in diverse biological processes. Among plant
C2H2-type zinc ®ngers, the DNA binding activity of the
petunia EPF protein family has been characterized to the
greatest extent. The EPF family differs from most C2H2-type
zinc ®ngers found in animal and yeast cells by the presence of
amino acid spacers of various lengths between zinc ®ngers
(1±65 amino acids) and by the signature QALGGH motif (9).
EPF proteins were reported to recognize two separate AGT
core sequences; presumably, each separate ®nger binds an
AGT triplet. The distance between the zinc ®nger domains
appears to dictate the amount of space between the binding
site and, therefore, the identity of the core target sequence
(10,11). Mutagenesis analysis has demonstrated that the
conserved QALGGH sequence is critical for DNA binding
activity of EPF proteins and that at least two zinc ®nger
domains are necessary for this interaction (11,12). Recently,
the QALGGH-containing EPF-like protein with a single
C2H2-type zinc ®nger encoded by SUPERMAN, an
Arabidopsis ¯ower development gene, was reported to have
speci®c DNA binding activity (13).
Another class of plant C2H2-type zinc ®nger proteins that
lacks the QALGGH motif was identi®ed and shown to
regulate important plant processes such as ¯owering time,
gametophyte formation and leaf development. The maize
INDETERMINATE1 gene, id1 (14), and the Arabidopsis
TRANSPARENT TESTA1 gene, TT1 (15), each contain at
least two zinc ®ngers. Other Arabidopsis proteins,
FERTILIZATION-INDEPENDENT SEED 2, FIS2 (16),
EMBRYONIC
FLOWER2,
EMF2
(17)
and
VERNALIZATION2, VRN2 (18), each have single C2H2 zinc
®ngers that show signi®cant homology with Su(z)12, a
Drosophila polycomb group gene (19). Similarly, the
Arabidopsis SERRATE gene (SE1) also contains a
single zinc ®nger that may function through chromatin
modi®cation (20).
Nucleic Acids Research, 2004, Vol. 32, No. 5
1711
Cloning of deletion mutants of ID1, VEG7 and VEG9
Construction of expression vectors
Fragments of deletion mutants of ID1 were ampli®ed by PCR
using full-length id1 cDNA in pBluescriptSK± vector (pBSK±;
Stratagene) as a template. The conserved region of the ID1
protein (ID domain; amino acid residues 67±254)
was ampli®ed by PCR using forward primer IDKRKRF (5¢and
GCGGATCCAAGAGGAAGAGAAGCCAGCCG-3¢)
reverse primer IDSART7 (5¢-GCCTCGAGTCAACCCATTTGCTGTCCACCAGTCATGCTAGCCATCCTCGCGCTCTCCTC-3¢). The deletion fragment encoding the N-terminal
232 amino acid residues (Z4del) was ampli®ed using forward
primer IDEcoRIF (5¢-GCGAATTCATGCAGATGATGATGCTCTC-3¢) and reverse primer Z4delT7 (5¢-CTCGAGTCAACCCATTTGCTGTCCACCAGTCATGCTAGCCATGAAGAAGAGGATGCCGCA-3¢). Primers IDKRKRF and
IDEcoRIF have BamHI and EcoRI sites, respectively (underlined), and IDSART7 and Z4delT7 have XhoI sites (underlined), a stop codon (bold) and a T7-tag sequence (italics).
Three PCR steps were used to amplify the fragment of
IDdel1 (the deletion mutant lacking the insertion between zinc
®ngers). Forward primer IDEcoRIF and reverse primer
IDdel1R (5¢-GACGCGCTTCGGGACGACGAGGCTGCT3¢) were used in the ®rst PCR ampli®cation. In the second
PCR, forward primer IDdel1F (5¢-GTCGTCCGGAAGCGCGTCTACGTC-3¢) and reverse primer IDXhoR (5¢-
To construct the plasmid for the expression of glutathione
S-transferase (GST)±ID1 fusion protein the BamHI±XhoI
fragment from id1 cDNA in pBSK± was inserted into the
BamHI±XhoI site of pGEX-4T (Pharmacia). The BamHI site
is inside the id1 coding region (52 bp from the translation
start) and the XhoI site is in the multi-cloning site of the
pBSK± vector. The expression vectors for other GST fusion
proteins, except the vector for VEG9, Z3M and Z4M
expression, were constructed by inserting BamHI±XhoI
fragments into the BamHI±XhoI site of pGEX-4T. The
expression vector for GST±VEG9 NZ was constructed by
insertion of the BamHI±SalI fragment into the BamHI±SalI
site of pGEX-4T and EcoRI±XhoI fragments of Z3M
(C164A,C169A) and Z4M (C226A,C228A) were inserted
into the EcoRI±XhoI site of pGEX-4T.
Expression and puri®cation of recombinant proteins
Escherichia coli BL21 CodonPlus-RP cells (Stratagene) were
transformed with the expression vectors and grown in LB
medium at 37°C until the absorbance A600 reached 0.6; at this
point isopropyl b-D-thiogalactopyranoside (IPTG) inducer
was added to a ®nal concentration of 0.3 mM and the culture
was grown for an additional 14 h at 20°C. The cells were
harvested and suspended in a 1/50 to 1/100 culture volume of
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
MATERIALS AND METHODS
GCCTCGAGCTAGAAGTTGTGGCTCCAGGTC-3¢) were
used. IDXhoR contains an XhoI site (underlined) and stop
codon (bold). In the ®rst and second PCR steps full-length id1
cDNA was used as a template. The ®rst and second PCR
products were puri®ed by electrophoresis on 1% agarose gel
and extracted from the gel using QIAEX II (Qiagen). The
mixture containing the ®rst and second PCR products was
used as a template for the ®nal PCR, which was performed
using forward primer IDEcoRIF and reverse primer
IDSART7.
Full-length cDNA clones for VEG7 or VEG9, cloned into
the pBSK± vector, were used as templates for ampli®cation of
VEG7 and VEG9 fragments. The VEG7 fragment was
ampli®ed using forward primer VEG7KKKRBF (5¢GCGGATCCAAGAAGAAGAGGAACCAG-3¢) and reverse
primer VEG7SAQT7 (5¢-GCCTCGAGTCAACCCATTTGCTGTCCACCAGTCATGCTAGCCATCTGCGCGCTCTCACG-3¢), and the VEG9 fragment was ampli®ed using forward
primer VEG9BF (5¢-GCGGATCCATGGCATCGAATTCATCG-3¢) and reverse primer VEG9SART7 (5¢-GTCGACTCAACCCATTTGCTGTCCACCAGTCATGCTAGCCATGCGCGCGCTTTCCTG-3¢). VEG7KKKRBF and VEG9BF
have a BamHI site (underlined). VEG7SAQT7 and
VEG9SART7 have XhoI and SalI, respectively (underlined),
a stop codon (bold) and T7-tag sequence (italics). The
fragment for VEG7 contains the ID domain only and the
fragments for IDdel1 and VEG9 contain the ID domain with
49 extra amino acids of the N-terminal region because
expression of the ID domain alone in Escherichia coli was
not successful.
PCR was performed for 35 cycles (10 s at 96°C, 30 s at 55°C
and 2 min at 72°C) using Pfu turbo or Herculase Enhanced
DNA polymerase (Stratagene). The ®nal PCR products were
puri®ed and cloned into the pCR-Blunt vector (Invitrogen) and
sequenced.
Genetic analysis and expression studies demonstrated that
the id1 gene plays a key role in regulating the transition from
vegetative to reproductive growth in maize by controlling the
production or transmission of leaf-derived ¯oral inductive
signals (14,21). The ID1 protein de®nes a new family of zinc
®nger proteins, termed ID-like proteins, which are found in all
higher plants. The ID domain, a highly conserved stretch of
amino acids common to all ID-like proteins, was originally
described as having two typical C2H2 and C2HC zinc ®nger
motifs (14). Visual inspection of the deduced ID1 amino acid
sequence reveals two additional atypical zinc ®ngers in the ID
domain.
As a prelude to the identi®cation of target genes regulated
by ID1 we present here a detailed study of the target binding
site speci®city and the DNA binding activity of ID1. Using
in vitro analysis techniques we show that the ID domain
recognizes and speci®cally interacts with a contiguous 11 bp
sequence. We also describe an unusual feature of ID1
interaction with DNA in which an additional T residue at
the ±1 position of the consensus facilitates ID1 interaction
with sequences that differ slightly from the de®ned 11 bp
consensus motif. Mutation analysis of the ID1 protein revealed
that only the C2HC ®nger domain and an adjacent putative
zinc ®nger are essential for recognizing the target sequence.
The other zinc ®nger domains, although highly conserved in
all ID domain proteins, are not required for interaction with
the consensus motif. We also show that variations in spacing
between zinc ®ngers in ID1 and ID-like family proteins do not
alter DNA binding speci®city. These results reveal that ID
domain proteins exhibit novel DNA binding characteristics
that do not conform entirely to standard rules for zinc ®nger±
DNA interaction.
1712
Nucleic Acids Research, 2004, Vol. 32, No. 5
13 PBS (40 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and
1.8 mM KH2PO4, pH 7.3) with 5 mM DTT and protease
inhibitor cocktail (Sigma). The cells were disrupted by
sonication and the lysate was centrifuged at 16 000 g for
20 min at 4°C. The fusion protein was allowed to bind to the
glutathione±Sepharose 4B (Pharmacia) for 1 h at 4°C. The
glutathione±Sepharose was washed four times with 13 PBS
and proteins were eluted with elution buffer (50 mM Tris±
HCl, pH 8, 10 mM reduced glutathione and 100 mM NaCl).
Selection and ampli®cation binding (SAAB) assay
Electromobility shift assays (EMSAs)
For EMSA experiments, 5¢-end biotin-labeled probes were
made with biotin-labeled SAABHAND-F and -R primers. For
the binding test of the SAAB-selected sequence, insertions in
pCR-blunt vector were ampli®ed using these primers. To
produce mutant probes, oligonucleotides of the desired
sequence were synthesized and dsDNAs were synthesized
and ampli®ed by PCR using biotin-labeled SAABHAND-F
and -R primers. In the reaction, 100 ng of partially puri®ed
protein was incubated for 30 min on ice with 30±50 fmol of the
labeled probes in binding buffer, as used in the SAAB assay.
After incubation, the mixture was loaded on a 6% polyacrylamide gel (acrylamide/bisacrylamide, 29:1) and run
in 0.253 Tris-borate±EDTA buffer at 4°C. DNA±protein
complexes were transferred to a Hybond-N+ membrane
(Amersham) and detected with a LightShift Chemiluminescent EMSA kit (Pierce). Approximate quanti®cation of DNA±
protein interactions were assessed by intensity of the shifted
bands determined with the NIH Image analysis program.
Site-directed mutagenesis was introduced by three-step PCR,
using the same protocol as described above. Full-length id1
cDNA was used as template in the ®rst and second PCR
ampli®cations. In the ®rst PCR, the forward primer IDEcoRIF
and reverse primers containing the desired mutation were
used. In the second PCR, the forward primers, the complement
sequence of the reverse primer, and reverse primer ID XhoR
were used. The ®rst and second PCR products were puri®ed
and this mixture was used as a template for the ®nal PCR.
Final PCR for Z3M (C164A,C169A) and Z4M
(C226A,C228A) was performed using forward primer
6HKRKRF (5¢-GCGAATTCATGCACCACCACCACCACCACAAGAGGAAGAGAAGCCAG-3¢) and reverse primer
SARstop (5¢-GCCTCGAGCTACCTGGCGCTCTCCTCTGC-3¢). EcoRI and XhoI, respectively, are underlined,
6HKRKF has 63 His (italics) and SARstop has a stop
codon (bold). For ampli®cation of Z1M (C98A,C101A) and
Z2M (C199A,C202A), primers IDEcoRIF and SARstop were
used because GST fusion proteins of these mutants with the ID
domains alone could not be expressed in E.coli. PCR was
performed as described above and the ®nal products were
inserted into the pCR-Blunt vector and con®rmed by
sequences analysis.
RESULTS
The maize ID1 protein binds to a speci®c DNA sequence
BLAST analysis of the maize id1 gene indicated that the
deduced ID1 protein contains two zinc ®ngers similar to the
Drosophila KruÈppel-like zinc ®ngersÐa C2H2 ®nger and the
less prevalent C2HC zinc ®nger (14). The presence of zinc
®nger motifs and the localization of ID1 protein to the nucleus
(J. Colasanti, unpublished results) suggest that ID1 binds DNA
and acts as a transcriptional regulator. Full-length ID1 was
fused with GST and the recombinant protein was used for a
SAAB assay to determine whether ID1 protein recognizes and
interacts with a particular DNA sequence. The GST±ID1
fusion protein was immobilized and used to select DNA
sequences that speci®cally bind ID1 from a population of
dsDNA oligonucleotides containing 20 bp of random sequences. After several rounds of capture by the protein complex
followed by PCR ampli®cation, 38 clones were isolated and
sequenced (see Materials and Methods).
The MEME motif identi®cation program (Multiple Em for
Motif Elicitation; http://meme.sdsc.edu/meme/website/meme.html) identi®ed an 11 bp consensus motif that was prominent
in 19 out of the 38 SAAB-selected clones (Fig. 1A). To
determine whether ID1 interacts with this 11 bp motif an
EMSA was performed with each of the 19 sequences using
GST±ID1 fusion protein. The ID1 protein bound to 16 out of
the 19 MEME-selected sequences with varying af®nities.
Three clones (clones 21, 42 and 45) did not bind to the GST±
ID1 fusion protein (Fig. 1A). No shifted bands were detected
in reactions with GST protein alone (data not shown). A
consensus motif based on the binding activity and frequency
of nucleotide utilization was derived (Fig. 1B).
The remaining 19 sequences derived from the SAAB assay,
but not selected by MEME analysis, were screened individually by EMSA; ®ve of these sequences were found to bind to
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
The GST±ID1 fusion protein (~10 mg) was immobilized on
100 ml of glutathione±Sepharose 4B resin or 20 ul of
Dynabeads ProteinG (DYNAL) coupled with anti-ID1
protein-speci®c antibody (5 mg). Af®nity-puri®ed anti-ID1
antibody was generated in rabbits immunized with a 20 amino
acid peptide that corresponds to the C-terminus of the deduced
ID1 protein coupled to Keyhole Limpet Hemaocyanin. Resins
were washed with 13 PBS three times, and suspended in
300 ml of binding buffer P {10 mM Tris±HCl (pH 7.5), 75 mM
NaCl, 1 mM DTT, 6% glycerol, 1% BSA, 1% Nonidet P-40,
poly[d(I,C)] and 10 mM ZnCl}. The suspension was divided
into six portions and one portion was used for each round of
selection. In the ®rst round, one portion of protein mixture was
incubated with 20 pmol of randomly synthesized DNA for 3 h
at 4°C.
A library of randomly synthesized DNA, a gift from Dr
Harley Smith (University of California, Berkeley), was used
as described to identify binding sites for the maize KNOTTED
protein (22). Bound DNA was eluted in 50 ml of H2O by
boiling for 10 min after washing the resin six times with
washing buffer (15 mM Tris±HCL, pH7.5, 75 mM NaCl,
0.15% Triton X-100 and 1% BSA) and then twice with
washing buffer without BSA. For each step, 10 ml of eluted
DNA was ampli®ed by 15 cycles of PCR (10 s at 96°C, 30 s at
52°C and 1 min at 72°C) using forward primer SAABHANDF (5¢-GAGAGGATCCAGTCAGCATG-3¢) and reverse primer SAABHAND-R (5¢-TGTCGAATTCTCGAGGCTGAG3¢). The 10 ml of PCR products were subjected to the second
selection. After six cycles of selection, populations of DNA
were cloned into the pCR Blunt vector and sequenced.
Site-directed mutagenesis
Nucleic Acids Research, 2004, Vol. 32, No. 5
1713
replacement of any single base in the 11 bp consensus
sequence by a non-consensus base completely abolished, or
severely reduced, binding. Competition experiments with
unlabeled mutant probes versus labeled wild-type probe
further supported these results (data not shown). These
®ndings con®rm the binding test of selected sequences
(Fig. 1A), where a perfect consensus showed very strong
binding, and any variation from the consensus showed weaker
binding, or the inability to bind, to ID1 protein.
Sequences outside the 11 bp DNA consensus motif affect
ID1 binding
ID1 (Fig. 1C). These oligonucleotide probes all contained the
sequence, TTTTGTCG/C, which matches 7 bp of the 11 bp
consensus motif. A common feature of this motif is a T residue
in the ±1 position; the signi®cance of this ®nding is discussed
below.
To further de®ne the binding speci®city of the ID1 protein
we identi®ed the essential core motif by scanning mutation of
the 11 bp consensus sequence. As shown in Figure 2A,
Both typical and putative novel zinc ®ngers in the ID
domain interact with DNA
ID1 de®nes a family of zinc ®nger proteins that are unique to
plants (J. Colasanti, K. Briggs, M. Muszynski, R.Tremblay
and A. Kozaki, unpublished results). The Arabidopsis genome
has 16 id-like genes, and we have isolated 4 id-like cDNAs
from maize, in addition to the id1 gene (14). The ID domain, a
stretch of approximately 170 amino acids, begins with a
putative nuclear localization signal and contains two BLASTidenti®ed zinc ®ngers, Z1 and Z2 (Fig. 3A). The highly
conserved ID domain is common to all ID-like proteins and is
a de®ning feature of the ID-like protein family. Figure 3 shows
an alignment of the ID domains of ID1 and two maize ID-like
proteins, VEG7 and VEG9. Note that an additional 23±25
amino acids between zinc ®ngers distinguishes maize ID1
from most ID family proteins examined so far (see below).
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
Figure 1. Determination of DNA sequences bound by ID1 protein.
(A) Sequences of oligonucleotides selected by SAAB and identi®ed by
MEME analysis. (B) Consensus sequence determined from the binding
sequences in (A). (C) SAAB-selected sequences that were not selected by
MEME analysis but which showed binding to ID1. Sequences are aligned
and oriented to match the consensus sequences. Binding ability was tested
by EMSA in which a constant amount of protein was used in each binding
reaction. The approximate percentage of bound probe in each reaction was
used to characterize the relative binding af®nities, as indicated with +++,
++, + and (+) corresponding to reactions in which >70, 40±70, 10±40 and
<10% of the probe was bound, respectively. Flanking primer sequences are
indicated with lower case letters.
As described above, an exception to the binding speci®city of
the 11 bp consensus was the 8 bp motif TTTTGTCG/C found
in ®ve probe sequences but not identi®ed by MEME (Fig. 1C).
A common feature of these ®ve probes was an additional T
residue at the 5¢ end (the ±1 position). SAAB-selected
sequences 16, 13 and 19, which contain a non-consensus G
at positions 9, 11 and 8, respectively, showed relatively strong
binding to ID1. In contrast, scanning mutation experiments
with the 11 bp consensus motif without the T at ±1 showed that
base changes in these same positions severely reduced
binding.
To test the possibility that an extra T residue 5¢ of the 11 bp
consensus can alter ID1 binding speci®city, the scanning
mutation experiment was repeated with mutant probes containing Ts in the ±1 and ±2 positions and a series of single
nucleotide replacements at positions 3±8 (Fig. 2B).
Surprisingly, most mutant probes showed binding af®nity
that was comparable to, or slightly lower than, binding to the
11 bp consensus motif. Only mutations in consensus positions
3, 4 and 6 (TT-M3, TT-M4 and TT-M6) showed weak or no
binding to ID1 (Fig. 2B). Probes with mutations at positions 1
and 2 also showed comparable binding to the consensus
sequence (data not shown). Similar base changes in the 11 bp
consensus sequence alone showed very weak or no binding to
ID1 (Fig. 2A). Competition experiments showed that a T
residue at the ±1 position of the 11 bp consensus sequence
does not simply increase the binding af®nity of ID1 for this
sequence above that of the consensus motif without a T
residue in the 5¢ position (data not shown). Rather, a T residue
at the ±1 position speci®cally enables ID1 to bind variants of
the consensus motif with less stringent requirements for
interaction.
1714
Nucleic Acids Research, 2004, Vol. 32, No. 5
In addition to Z1 and Z2, visual examination of the ID
domain revealed two other potential zinc ®nger domains that
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
Figure 2. Ability of GST±ID1 fusion protein to bind mutant variants of the
consensus motif. (A) EMSA using single base mutant probes. (B) EMSA
using single base mutant probes that have additional Ts at position ±1 and
±2. Mutated bases are shown as boxed lower case letters. The sequences of
the probes are shown under the panels. An arrowhead indicates the
ID1/DNA complex. Less intense lower bands are not associated with the
speci®c shifted bands. Relative binding af®nity of each probe was compared
with that of consensus sequence (WT: TTTGTCGTATT) and indicated with
+++, ++ + and (+) corresponding to >70, 40±70, 10±40 and <10% of the
binding to the WT consensus sequence.
were not identi®ed by BLAST similarity searches. One
putative zinc ®nger, designated Z3, is located between Z1
and Z2, and another potential zinc ®nger motif, Z4, is
downstream of Z2 (Figs 4 and 5A). The Z3 sequence contains
two conserved cysteines (C164 and C169) and two histidines
(H187 and H192), however, the 17 amino acids between C169
and H187 differ from the 12 residues found in typical C2H2
zinc ®ngers (5). The six amino acid sequence immediately
upstream of C199 in Z2 is similar to the H/C link motif, an
amino acid spacer that separates modules in cluster-type zinc
®nger proteins. H/C links often contain the sequence TGEKP,
therefore the sequence GEKRWC between Z3 and Z2 could
be a putative H/C link (23).
The other candidate, Z4, is a potential C2HC zinc ®nger that
contains cysteines (Cys226, Cys228 and Cys245), a histidine
(His241) and hydrophobic residues in positions that conform
to the rules for C2H2 zinc ®ngers (Fig. 3A). Moreover, the
structure of Z4 is similar to the third zinc ®nger of the yeast
SWI5 transcription factor which also has the unusual feature
of a single amino acid separating Cys226 and Cys228 (Fig. 3B)
(24). Overall, the four putative zinc ®nger domains are highly
conserved among all ID-like family members analyzed so far,
with >80% amino acid identity (Fig. 3A and J. Colasanti,
K. Briggs, M. Muszynski, R.Tremblay and A. Kozaki,
unpublished results).
The addition of metal chelators such as EDTA and 1,10phenanthroline abolished the shifted band ID1±DNA complex
in EMSA experiments, indicating that the DNA binding
ability of ID1 protein is zinc dependent (data not shown). In
most cluster-type C2H2 zinc ®ngers in animals, and EPF-type
zinc ®ngers in plants, each ®nger domain has been shown to
recognize a triplet of base pairs (9,23). If the general rule of
one zinc ®nger module binding a triplet of DNA is applied
here, then ID1 would require three or four zinc ®ngers in its
DNA binding domain to recognize an 11 bp target site.
To further characterize the mode of DNA binding by ID1,
we determined whether the ID domain alone mediates speci®c
DNA binding. A recombinant, partially puri®ed GST±ID
domain fusion protein was used for EMSA. As shown in
Figure 4B, the ID domain of ID1 binds to the 11 bp consensus
motif and showed the same speci®city in binding to mutant
probes as does the full-length ID1 protein, suggesting that the
ID domain is suf®cient for mediating speci®c DNA binding.
The contribution to DNA binding made by zinc ®ngers Z1
and Z2, as well as Z3 and Z4 motifs, was tested by disrupting
each putative zinc ®nger and determining if the mutant ID
domains could recognize the consensus sequence. Zinc ®ngers
were disrupted by replacing the ®rst cysteine pair of each
module with two alanine residues; i.e. Z1M (C98A,C101A),
Z2M (C199A,C202A), Z3M (C164A,C169A) and Z4M
(C226A,C228A) represent ID domain proteins with mutant
versions of Z1, Z2, Z3 and Z4, respectively (Fig. 4A). Mutant
expression constructs consisted of the ID domain (amino acids
67±254) fused to GST, except for Z1M and Z2M, which could
not be expressed in E.coli. Expression of these mutant proteins
in E.coli required that a slightly longer region (amino acids 18±
254) be used for adequate levels of expression (Fig. 4C). This
longer protein shows the same binding properties as full-length
wild-type ID1 or ID domain only protein (data not shown).
The results in Figure 4B show that mutating the ®rst
two cysteines of zinc ®ngers Z2 or Z3 (Z2M and Z3M,
Nucleic Acids Research, 2004, Vol. 32, No. 5
1715
respectively) abolished the binding activity of the ID domain,
indicating that functional versions of each of these zinc ®ngers
are required for binding the consensus motif. Unexpectedly,
ID domain proteins with mutant Z1 or Z4 zinc ®ngers (Z1M
and Z4M, respectively) retained DNA binding activity that
was comparable with the wild-type ID domain (Fig. 4B).
Moreover, scanning mutation experiments showed that loss of
functional Z1 or Z4 zinc ®ngers does not change the binding
speci®city to mutated consensus probes (data not shown),
further demonstrating that Z1 and Z4 do not contribute to ID
domain DNA binding speci®city. Overall, our data suggest
that only two of the four possible zinc ®ngers in the ID
domain, Z2 and Z3, are required for interacting with the DNA
consensus motif.
Additional amino acids between ID1 zinc ®ngers do not
alter DNA recognition
A unique feature of maize ID1, relative to other ID domain
proteins, is an additional 23±25 amino acids between the ®rst
two zinc ®ngers (Fig. 3A). Assuming that Z3 is a true zinc
®nger, the distance between Z1 and Z3 in all other ID-like
proteins identi®ed so far ranges from 20 to 22 amino acids,
whereas 45 amino acids separate these motifs in ID1.
Although no other ID domain protein has such a long spacer,
a putative id1 ortholog from rice, idOs, has an eight amino
acid extension (A. Kozaki and J. Colasanti, unpublished
results). For most animal C2H2-type proteins, the zinc ®nger
modules are grouped into clusters, with spacing between
®ngers invariant for any particular family (23). In contrast,
many plant zinc ®nger proteins appear to be more ¯exible with
respect to spacing between ®ngers. In the petunia EPF-like
proteins, where family members are classi®ed based on the
distance between ®ngers, studies suggest that the spacing
between the ®ngers dictates the speci®city of the DNA binding
site (12,25). In the case of ID1, our results suggest that zinc
®nger Z1 is not essential for speci®c DNA binding and the
function of the long spacer between Z1 and the Z3±Z2±Z4
cluster is different from that of the inter-®nger spacers found
in EPF proteins.
We compared the DNA binding ability of the ID domain
with and without the 25 amino acid spacer to test whether the
extra distance between Z1 and Z3 alters the ID1 binding site
preference such that it recognizes target sites that are signi®cantly different from sequences recognized by other ID-like
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
Figure 3. Sequence alignment of ID domain and comparison of structure of Z4 to the third zinc ®nger of SWI5. (A) Amino acid sequence alignment of ID
domains of ID1, VEG7 and VEG9. (B) The structure of the Z4 region in (A) is compared with that of the third zinc ®nger of SWI5. Conserved C and H
residues are boxed and conserved amino acids are shaded. Numbers on the sequence indicate the amino acid number of ID1 protein; the entire ID1 protein is
436 amino acids. Asterisks indicate the conserved hydrophobic residues in zinc ®ngers. Dots indicate the sequence homologous to the H/C link and the double
underline indicates putative nuclear localization signal. Each zinc ®nger region or zinc ®nger-like region is underlined.
1716
Nucleic Acids Research, 2004, Vol. 32, No. 5
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
Figure 4. Analysis of the DNA binding function of each putative zinc ®nger motif. (A) Schematic diagram of mutant proteins used in EMSA experiments.
Structure of the ID domain common to all ID-like family proteins is shown on top. Zinc ®ngers (Z1 and Z2) and putative zinc ®ngers (Z3 and Z4) are
indicated as hatched boxes. C and H indicate cysteines and histidines that de®ne putative zinc ®ngers. Numbers indicate the amino acid position of C and H
residues of the ID1 protein. A black box indicates the position of a putative nuclear localization signal. (B) DNA binding of wild-type and mutant proteins in
which each putative zinc ®nger was disrupted. Mutant proteins (as GST fusions) were incubated with biotin-labeled consensus sequences (WT:
TTTGTCGTATT). Arrowheads indicate the complex of proteins and probes. (C) Preparation of the partially puri®ed mutant proteins used in (B). Proteins
were separated by SDS±PAGE and stained with Coomassie Brilliant Blue. Arrows show the positions of full-length GST fusion proteins, indicating that they
essentially are intact and present in similar concentrations. Molecular-weight markers (M) are shown on the left (kDa).
family proteins. The ID domain of ID1 that lacked the insertion
(IDdel1; Fig. 4A) was fused to GST and partially puri®ed
recombinant protein was used for EMSA. As shown in Figure 5,
the IDdel1 protein bound to the 11 bp consensus motif and
showed the same binding properties to mutant probes as did
full-length ID1, although binding to the probe with the
mutation at position 11 (M11) is slightly stronger. This ®nding
suggests that, unlike EPF-like proteins, the presence of
additional amino acids between zinc ®ngers does not affect
binding speci®city relative to other ID domain proteins.
Nucleic Acids Research, 2004, Vol. 32, No. 5
1717
sensitive to the base change at this position (position 11) than
proteins without insertion between zinc ®ngers. These results
suggest that sequences recognized by VEG7 or VEG9 are
similar to that of ID1 even though they lack the long spacer
between zinc ®ngers present in the ID1 protein.
DISCUSSION
ID1 protein has unique structural features
Figure 5. DNA binding of deletion mutants of ID and maize ID-like
proteins VEG7 and VEG9. (A) Summary of binding of deletion mutant and
ID-like proteins. Relative binding af®nity of each probe to each protein was
compared with binding to the consensus motif (WT: TTTGTCGTATT) and
indicated with +++, ++, + and (+), corresponding to >70, 40±70, 10±40 and
<10% of the binding to WT (m19 sequence). (B) EMSA showing some of
the DNA/protein complexes with several of the probes in (A).
ID1 and ID-like proteins recognize the same consensus
sequence
Since IDdel1 has DNA binding properties similar to wild-type
ID1, we tested whether other ID-like proteins could interact
with the 11 bp consensus motif. We used EMSA to test the
DNA binding capabilities of maize ID domain proteins VEG7
and VEG9 (Fig. 3). Figure 5 shows that both VEG7 and VEG9
bind to oligonucleotides containing the 11 bp consensus motif;
i.e., they exhibit binding properties similar to ID1 and the
IDdel1 mutant.
As shown in Figure 5A, ID1, IDdel1 and the ID-like
proteins VEG7 and VEG9 did not bind probes with mutations
at positions 3±8, indicating that bases at these positions are
important for protein±DNA interaction. On the other hand,
these proteins showed weak binding to mutant probes with
mutations at positions 1, 2, 9, 10 and 11. The binding of VEG7
and VEG9 to the M11 probe was comparable to their
interaction with the consensus motif, and the binding of
IDdel1 to M11 was slightly stronger than that of ID1. These
observations may re¯ect differences in binding sequence
speci®city of ID-like family proteins or ID1 may be more
ID1 is a zinc ®nger protein with several unique characteristics.
First, the ID domain, the structural component of ID1 that
interacts with DNA and de®nes the ID-like protein family, is
composed of different types of zinc ®ngers (Fig. 3). Zinc
®nger Z1 has the hallmarks of canonical C2H2 zinc ®ngers
typi®ed by KruÈppel-like proteins and the general transcription
factor TFIIIA. Zinc ®nger Z2 encodes an atypical C2HC
domain that has the same structure as canonical C2H2 zinc
®ngers except for a histidine replacing the last cysteine. The
other putative zinc ®ngers, Z3 and Z4, were not recognized as
typical zinc ®ngers in genome-wide similarity searches
(J. Colasanti, unpublished results). However, alignment of
ID domains from several plant species reveals that the
sequences of Z3 and Z4 are highly conserved, suggesting
that they may have an important function, possibly by acting
as functional zinc ®ngers that mediate DNA recognition
(Fig. 4). Here we have shown that Z3, but not Z4, is required
for recognition of the 11 bp consensus motif. DNA binding
zinc ®ngers with the structure of Z3, YXCX4CX17H4H, have
not been reported. However, the baculoviral IAP BIR2 domain
has a primary structure, CX2CX16HX6C, that differs from
typical C2H2 zinc ®ngers, yet is able to chelate zinc and form
a three-dimensional structure similar to C2H2 zinc ®ngers
(26). Moreover, the six amino acid spacer separating the Z3
and Z2 domains resembles the conserved H/C link motifs that
often separate cluster-type zinc ®ngers (5,27), further supporting the possibility that Z3 functions as a DNA binding zinc
®nger module. The structure of Z4, YXCXCX3FX8HX3C,
satis®es the requirements of canonical C2H2 ®ngers, except
that a single amino acid separates the two cysteines that are
presumed to chelate zinc. In addition, the similarity of Z4 to
the third zinc ®nger of the yeast SWI5 transcription factor (24)
strongly suggests that the Z4 domain is a functional zinc ®nger
(Fig. 4B). Overall, ID1 and all ID domain family members
contain four distinct types of zinc ®nger modules.
The other unique structural feature of ID1 is that three of the
four putative zinc ®ngers, Z2, Z3 and Z4, are arrayed in a
tandem cluster, as is found with most animal zinc ®nger
proteins, but one ®nger, Z1, is separated from the cluster by a
long spacer. Separation of zinc ®nger modules by long,
variable length spacers appears to be a common characteristic
of plant zinc ®nger proteins (8,11,12,25).
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
We are investigating the biochemical function of the maize
ID1 protein by characterizing its DNA binding properties
in vitro, and by examining the structural role of the ID domain
zinc ®ngers in recognizing the speci®c DNA consensus motif.
The presence of zinc ®nger domains that interact with speci®c
DNA sequences is further evidence that maize ID1, and most
likely all ID domain proteins, function as transcriptional
regulators.
1718
Nucleic Acids Research, 2004, Vol. 32, No. 5
The ID1 protein binds to a speci®c DNA sequence via
zinc ®ngers of the ID domain
Different spacing between ID domain zinc ®ngers does
not alter DNA binding speci®city
The additional 25 amino acids between zinc ®ngers Z1
and Z3 distinguishes ID1 from all other known members
of the ID-like family (Fig. 3). We wanted to determine
whether this amino acid insertion contributes to the
speci®c ID1 function of controlling ¯owering time by
specifying interaction with particular target genes.
However, we found that an ID1 protein lacking this 25
amino acid insertion recognized the 11 bp binding motif
with the same speci®city as the complete ID1 protein
(Fig. 5). Furthermore, two ID family proteins from maize
that naturally lack the 25 amino acid addition, VEG7 and
VEG9, are able to recognize the ID1 consensus motif,
suggesting that they interact with the same 11 bp site or
they share an overlapping set of binding sequences. As of
yet, no function has been ascribed to any of the ID
domain family proteins. Although VEG7 and VEG9
mRNAs are expressed in the same tissues as ID1 protein
(i.e. immature leaves), genetic evidence does not suggest
that they share with ID1 the function of controlling
¯owering time (J. Colasanti, unpublished results).
The ®nding that changing the spacing between zinc
®ngers of the ID1 protein has little effect on its ability to
recognize the 11 bp binding motif contrasts with the
situation for EPF-type proteins, where the distance
between ®ngers dictates the spacing between core binding
motifs (11). However, our ®ndings might not be unexpected since the isolated Z1 zinc ®nger does not seem to
be required for recognition of the 11 bp consensus motif.
Therefore, it is likely that the additional amino acids
between domains Z1 and Z3 in the ID1 protein do not
participate in contacting the cognate DNA sequence and
might have another function, such as mediating protein±
protein interaction.
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
The role of ID1 as a regulator of gene expression was
supported by the ®nding that ID1 protein binds to a speci®c
11 bp DNA consensus motif, 5¢-T-T-T-G-T-C-G/C-T/C-T/aT/a-T-3¢. We showed that recognition of this consensus
sequence occurs via the ID domain, and therefore, is likely to
be mediated by the zinc ®nger modules located within this
structure. The 11 bp sequence recognized by ID1 is the
approximate length expected for a protein with four zinc
®ngers if one assumes that each module interacts with a DNA
triplet (6,11). However, by mutating each zinc ®nger in ID1
we have shown that not all the presumed zinc binding motifs
participate in recognition of the consensus sequence. An
unexpected ®nding was that Z1, which has the structure of a
typical C2H2 zinc ®nger, is not required for interaction with
the consensus motif. Likewise, binding experiments with
mutant Z4 proteins showed that Z4 is not required for speci®c
DNA binding to the consensus (Fig. 4B). In contrast, sitedirected mutagenesis of Z2 and Z3 showed that they are
essential for optimal DNA binding, presumably because they
form functional DNA binding zinc ®ngers. It is possible that
Z2 and Z3 are required for mediating protein±protein interaction, and that disruption of this interaction would cause loss
of DNA binding activity. At present we do not know if
dimerization of ID domain proteins is required for consensus
motif recognition. Although the proposed rules for zinc ®nger
interaction with DNA would require more than two zinc
®ngers to bind an 11 bp sequence, it is possible that motifs
outside the ID domain zinc ®ngers could interact with DNA.
For example, the Drosophila GAGA protein contains basic
amino acids that were shown to act in concert with a single
zinc ®nger module to bind a speci®c 7 bp DNA motif (28).
Although more biochemical and structural evidence is needed
to prove that the unorthodox Z3 motif forms a zinc ®nger,
the mutagenesis experiments presented here support the
possibility that Z3 does interact with DNA.
Our results suggest that the binding ability of each
putative zinc ®nger domain of ID1 could not be predicted
based on similarity to previously characterized DNA
binding proteins. All ID domain proteins examined to
date contain two putative C2HC-type zinc ®ngers, Z2 and
Z4, and we have shown that Z2 is essential for speci®c
DNA binding. The ability of a C2HC-type zinc ®nger to
bind DNA in vitro was shown recently (29), but, so far,
most studies of naturally occurring C2HC-type domains
report that this type of zinc ®nger more often has a role
in protein±protein interaction or RNA binding, rather than
DNA binding (30,31). For example, the C2HC domains of
the Friend of GATA (FOG) proteins are involved in
mediating interaction with GATA-1 proteins (32). Akhtar
and Becker (33) showed that a C2HC-type ®nger in the
MOF protein is involved in nucleosome binding and also
is essential for the H4 histone acetyltransferase activity of
this protein. On the other hand, the third zinc ®nger of the
yeast SWI5 protein, an unusual C2HC-type ®nger that is
similar to Z4, is reported to be involved in DNA binding
(34). We ®nd that the putative Z4 ®nger of ID1 is not
involved in binding to the 11 bp consensus motif,
therefore it may have another, undiscovered function.
The surprising result that Z1 does not interact with the
consensus motif suggests that the binding ability of even
canonical zinc ®nger motifs must be tested on a case-bycase basis. However, the position of Z1 within the ID
domain may contribute to this unusual ®nding. The Z3, Z2
and Z4 modules together form a zinc ®nger cluster, but
Z1 is separated from this cluster by a long spacer. In
animals, the REST protein, a repressor of type II sodium
channel genes, has a cluster of eight zinc ®ngers that is
involved in DNA binding, and one zinc ®nger, separated
from this cluster, that is reported to be essential for
suppressor activity, but not DNA binding (35). An
arrangement of zinc ®ngers similar to ID domain proteins
is observed in the maize TRM1 protein, a YY1-like
suppressor of rbc-m3 expression (36). The TRM1 protein,
which is reported to bind DNA, contains ®ve putative zinc
®ngers; four in a cluster and one ®nger separated from the
cluster. However, the function of each of these domains
has not been determined. Therefore, the high level of
conservation of the Z1 ®nger of ID1, as well as Z4,
suggest that these domains may have other functions, such
as mediating protein±protein interaction or RNA binding.
However, the ability of Z1 to recognize sequences
separated from the 11 bp motif remains a possibility.
Nucleic Acids Research, 2004, Vol. 32, No. 5
Sequences outside the consensus motif affect ID1
binding
Biological relevance of ID1 protein function
Our study of the DNA binding properties of ID1 and the ID
family proteins reveals novel DNA recognition properties for
zinc ®nger proteins containing both typical C2H2-type and
atypical C2HC-type zinc ®ngers. Recently, Sagasser et al.
reported that the Arabidopsis transparent testa 1 gene, TT1, is
a member of the WIP domain family of proteins, which also
have a combination of C2H2-type and C2HC-type zinc ®nger
motifs (15). Although WIP domain proteins show signi®cant
homology to ID domain proteins (40% identity), the id-like
genes form a distinct, highly conserved sub-family. The
authors point out that additional conserved cysteine and
histidine residues in the C-terminus of the WIP domain might
lead to the formation of two additional ®ngers (15). Therefore,
it is possible that the various types of zinc ®ngers in WIP
domain proteins, like ID1, may mediate DNA binding,
whereas others may have other functions.
Our ®nding that DNA recognition by ID1 does not conform
entirely to prescribed rules for zinc ®nger±DNA interaction is
of considerable interest in light of recent endeavors to
engineer `polydactyl' zinc ®nger proteins to bind speci®c
DNA targets for altering expression of particular genes in vivo
(39,40) or for targeting nucleases to speci®c sites in the
genome to enhance homologous recombination (41).
Investigation of the class of plant proteins represented by
ID1 may provide novel insights into how zinc ®nger proteins
work in general.
ACKNOWLEDGEMENTS
We thank Harley Smith and George Chuck for critical
comments on the manuscript and Ryan Geil for editorial
assistance. This research is supported by grants to J.C. from
the National Science Foundation (MCB-9982714), Pioneer
Hi-bred International, Inc., and the Natural Sciences and
Engineering Research Council of Canada.
REFERENCES
1. Bieker,J.J. (2001) Kruppel-like factors: three ®ngers in many pies. J. Biol.
Chem., 276, 34355±34358.
2. Klug,A. and Schwabe,J.W.R. (1995) Protein motifs. 5. Zinc ®ngers.
FASEB J., 9, 597±604.
3. Parraga,G., Horvath,S.J., Eisen,A., Taylor,W.E. and Hood,L. (1988)
Zinc-dependent structure of a single-®nger domain of yeast ADR1.
Science, 241, 1489±1492.
4. Lee,M.S., Gippert,G.P., Soman,K.V., Case,D.A. and Wright,P.E. (1989)
Three-dimensional solution structure of a single zinc ®nger DNA-binding
domain. Science, 245, 635±637.
5. Wolfe,S.A., Nekludova,L. and Pabo,C.O. (2000) DNA recognition by
Cys(2)His(2) zinc ®nger proteins. Annu. Rev. Biophys Biomol. Struct.,
29, 183±212.
6. Pavletich,N.P. and Pabo,C.O. (1991) Zinc ®nger DNA recognitionÐ
crystal-structure of a Zif268-DNA complex at 2.1-A. Science, 252,
809±817.
7. Riechmann,J.L., Heard,J., Martin,G., Reuber,L., Jiang,C.Z., Keddie,J.,
Adam,L., Pineda,O., Ratcliffe,O.J., Samaha,R.R. et al. (2000)
Arabidopsis transcription factors: genome-wide comparative analysis
among eukaryotes. Science, 290, 2105±2110.
8. Takatsuji,H. (1998) Zinc-®nger transcription factors in plants. Cell. Mol.
Life Sci., 54, 582±596.
9. Takatsuji,H. (1999) Zinc-®nger proteins: the classical zinc ®nger
emerges in contemporary plant science. Plant Mol. Biol., 39, 1073±1078.
10. Takatsuji,H., Nakamura,N. and Katsumoto,Y. (1994) A new family of
zinc-®nger proteins in petuniaÐstructure, DNA-sequence recognition
and ¯oral organ-speci®c expression. Plant Cell, 6, 947±958.
11. Takatsuji,H. and Matsumoto,T. (1996) Target sequence recognition by
separate type Cys(2)/His(2) zinc ®nger proteins in plants. J. Biol. Chem.,
271, 23368±23373.
12. Kubo,K., Sakamoto,A., Kobayashi,A., Rybka,Z., Kanno,Y.,
Nakagawa,H., Nishino,T. and Takatsuji,H. (1998) Cys(2)/His(2) zinc®nger protein family of petunia: evolution and general mechanism of
target-sequence recognition. Nucleic Acids Res., 26, 608±615.
13. Dathan,N., Zaccaro,L., Esposito,S., Isernia,C., Omichinski,J.G.,
Riccio,A., Pedone,C., Di Blasio,B., Fattorusso,R. and Pedone,P.V.
(2002) The Arabidopsis SUPERMAN protein is able to speci®cally bind
DNA through its single Cys(2)-His(2) zinc ®nger motif. Nucleic Acids
Res., 30, 4945±4951.
14. Colasanti,J., Yuan,Z. and Sundaresan,V. (1998) The indeterminate gene
encodes a zinc ®nger protein and regulates a leaf-generated signal
required for the transition to ¯owering in maize. Cell, 93, 593±603.
15. Sagasser,M., Lu,G.H., Hahlbrock,K. and Weisshaar,B. (2002) A-thaliana
TRANSPARENT TESTA 1 is involved in seed coat development and
de®nes the WIP subfamily of plant zinc ®nger proteins. Genes Dev., 16,
138±149.
16. Luo,M., Bilodeau,P., Koltunow,A., Dennis,E.S., Peacock,W.J. and
Chaudhury,A.M. (1999) Genes controlling fertilization-independent seed
development in Arabidopsis thaliana. Proc. Natl Acad. Sci. USA, 96,
296±301.
17. Yoshida,N., Yanai,Y., Chen,L.J., Kato,Y., Hiratsuka,J., Miwa,T.,
Sung,Z.R. and Takahashi,S. (2001) EMBRYONIC FLOWER2, a novel
polycomb group protein homolog, mediates shoot development and
¯owering in Arabidopsis. Plant Cell, 13, 2471±2481.
18. Gendall,A.R., Levy,Y.Y., Wilson,A. and Dean,C. (2001) The
VERNALIZATION 2 gene mediates the epigenetic regulation of
vernalization in Arabidopsis. Cell, 107, 525±535.
19. Birve,A., Sengupta,A.K., Beuchle,D., Larsson,J., Kennison,J.A.,
Rasmuson-Lestander,A. and Muller,J. (2001) Su(z)12, a novel
Drosophila Polycomb group gene that is conserved in vertebrates and
plants. Development, 128, 3371±3379.
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
Although ID1 interacts with a speci®c target sequence, an
unexpected ®nding was that the addition of a T residue at the
±1 position changes ID1 binding speci®city for this consensus
motif. Selectivity for sequences outside the core binding site
for some C2H2-type zinc ®ngers has been reported (37,38).
Nagaoka et al. (37) found that the zinc ®nger protein Sp1
selectivity binds a G triplet outside the Sp1 GC box core
binding motif, thus revealing a novel binding site outside the
de®ned Sp1 consensus motif. Sp1 was shown to recognize this
alternate binding sequence with higher af®nity than the GC
box. Therefore, the ability to recognize alternate sequence
motifs might be a feature of some zinc ®nger DNA binding
proteins.
It is possible that ID1 recognizes alternative sequences
when T is present at the ±1 position, as in the case of the
alternative GC box mentioned above. In fact, probes with
slight variations from the consensus motif that had an extra T
at the ±1 position were selected by the SAAB assay and
showed relatively high binding af®nity to ID1 (Fig. 1A).
Perhaps the T outside the consensus sequence stabilizes the
DNA±protein complex. Although we do not know whether T
at the ±1 position of the consensus is preferred in vivo or not,
this property might provide valuable information when
searching for ID1 target genes. Analysis of the Arabidopsis
and maize genomes reveals ID domain binding motifs, with
and without additional ±1 T residues, in upstream positions of
many different genes (A. Kozaki and J. Colasanti, unpublished
results). Experiments are in progress to determine if any of
these particular sequences are preferred in vivo targets.
1719
1720
Nucleic Acids Research, 2004, Vol. 32, No. 5
30. Schmeichel,K.L. and Beckerle,M.C. (1997) Molecular dissection of a
LIM domain. Mol. Biol. Cell, 8, 219±230.
31. Arrizabalaga,G. and Lehmann,I. (1999) A selective screen reveals
discrete functional domains in Drosophila Nanos. Genetics, 153,
1825±1838.
32. Tsang,A.P., Visvader,J.E., Turner,C.A., Fujiwara,Y., Yu,C.N.,
Weiss,M.J., Crossley,M. and Orkin,S.H. (1997) FOG, a multitype zinc
®nger protein, acts as a cofactor for transcription factor GATA-1 in
erythroid and megakaryocytic differentiation. Cell, 90, 109±119.
33. Akhtar,A. and Becker,P.B. (2001) The histone H4 acetyltransferase MOF
uses a C2HC zinc ®nger for substrate recognition. EMBO Rep., 2,
113±118.
34. Nakaseko,Y., Neuhaus,D., Klug,A. and Rhodes,D. (1992) Adjacent zinc®nger motifs in multiple zinc-®nger peptides from Swi5 form structurally
independent, ¯exibly linked domains. J. Mol. Biol., 228, 619±636.
35. TapiaRamirez,J., Eggen,B.J.L., PeralRubio,M.J., ToledoAral,J.J. and
Mandel,G. (1997) A single zinc ®nger motif in the silencing factor REST
represses the neural-speci®c type II sodium channel promoter. Proc. Natl
Acad. Sci. USA, 94, 1177±1182.
36. Xu,T., Purcell,M., Zucchi,P., Helentjaris,T. and Bogorad,L. (2001)
TRM1, a YY1-like suppressor of rbcS-m3 expression in maize mesophyll
cells. Proc. Natl Acad. Sci. USA, 98, 2295±2300.
37. Nagaoka,M., Shiraishi,Y. and Sugiura,Y. (2001) Selected base sequence
outside the target binding site of zinc ®nger protein Sp1. Nucleic Acids
Res., 29, 4920±4929.
38. Swirnoff,A.H. and Milbrandt,J. (1995) DNA-binding speci®city of Ng®-a
and related zinc-®nger transcription factors. Mol. Cell. Biol., 15,
2275±2287.
39. Segal,D.J. (2002) The use of zinc ®nger peptides to study the role of
speci®c factor binding sites in the chromatin environment. Methods, 26,
76±83.
40. Beerli,R.R. and Barbas,C.F. (2002) Engineering polydactyl zinc-®nger
transcription factors. Nat. Biotechnol., 20, 135±141.
41. Bibikova,M., Beumer,K., Trautman,J.K. and Carroll,D. (2003)
Enhancing gene targeting with designed zinc ®nger nucleases. Science,
300, 764.
Downloaded from http://nar.oxfordjournals.org/ at Acquisitions Section on May 8, 2012
20. Prigge,M.J. and Wagner,D.R. (2001) The Arabidopsis SERRATE gene
encodes a zinc-®nger protein required for normal shoot development.
Plant Cell, 13, 1263±1279.
21. Colasanti,J. and Sundaresan,V. (2000) `Florigen' enters the molecular
age: long-distance signals that cause plants to ¯ower. Trends Biochem.
Sci., 25, 236±240.
22. Smith,H.M.S., Boschke,I. and Hake,S. (2002) Selective interaction of
plant homeodomain proteins mediates high DNA-binding af®nity. Proc.
Natl Acad. Sci. USA, 99, 9579±9584.
23. Pabo,C.O., Peisach,E. and Grant,R.A. (2001) Design and selection of
novel Cys(2)His(2) zinc ®nger proteins. Annu. Rev. Biochem., 70,
313±340.
24. Stillman,D.J., Bankier,A.T., Seddon,A., Groenhout,E.G. and
Nasmyth,K.A. (1988) Characterization of a transcription factor involved
in mother cell speci®c transcription of the yeast HO gene. EMBO J., 7,
485±494.
25. Yoshioka,K., Fukushima,S., Yamazaki,T., Yoshida,M. and Takatsuji,H.
(2001) The plant zinc ®nger protein ZPT2-2 has a unique mode of DNA
interaction. J. Biol. Chem., 276, 35802±35807.
26. Sun,C.H., Cai,M.L., Gunasekera,A.H., Meadows,R.P., Wang,H., Chen,J.,
Zhang,H.C., Wu,W., Xu,N., Ng,S.C. et al. (1999) NMR structure and
mutagenesis of the inhibitor-of-apoptosis protein XIAP. Nature, 401,
818±822.
27. Schuh,R., Aicher,W., Gaul,U., Cote,S., Preiss,A., Maier,D., Seifert,E.,
Nauber,U., Schroder,C., Kemler,R. et al. (1986) A conserved family of
nuclear proteins containing structural elements of the ®nger protein
encoded by Kruppel, a Drosophila segmentation gene. Cell, 47,
1025±1032.
28. Pedone,P.V., Ghirlando,R., Clore,G.M., Gronenborn,A.M., Felsenfeld,G.
and Omichinski,J.G. (1996) The single Cys(2)-His(2) zinc ®nger domain
of the GAGA protein ¯anked by basic residues is suf®cient for highaf®nity speci®c DNA binding. Proc. Natl Acad. Sci. USA, 93,
2822±2826.
29. Simpson,R.J.Y., Cram,E.D., Czolij,R., Matthews,J.M., Crossley,M. and
Mackay,J.P. (2003) CCHX zinc ®nger derivatives retain the ability to
bind Zn(II) and mediate protein±DNA interactions. J. Biol. Chem., 278,
28011±28018.