The Plant Cell, Vol. 8, 847-857, May 1996 O 1996 American Society of Plant Physiologists

Transcription Factor Veracity: 1s GBF3 Responsible for ABA-Regulated Expression of Arabidopsis Adh?

Guihua Lu,' Anna-Lisa Paul, Dona ld R. McCarty, a n d Rober t J. Fer12

Program in Plant Molecular and Cellular Biology, Horticultural Sciences Department, 1143 Fifield Hall, University of Florida, Gainesville, Florida 32611

Assignment of particular transcription factors to specific roles in promoter elements can be problematic, especially in systems such as the G-box, where multiple factors of overlapping specificity exist. In the Arabidopsis alcohol dehydfogenase (Adh) promoter, the G-box regulates expression in response to cold and dehydration, presumably through the action of abscisic acid (ABA), and is bound by a nuclear protein complex in vivo during expression in cell cultures. In this report, we test the conventional wisdom of biochemical approaches used to identify DNA binding proteins and a s s e s their specific interactions by using the G-box and a nearby half G-box element of the Arabidopsis Adh promoter as a model system. Typical in vitro assays demonstrated specific interaction of G-box factor 3 (GBF3) with both the G-box and the half G-box element. Dimethyl sulfate footprint analysis confirmed that the in vitro binding signature of GBFB essentially matches the footprint signature detected in vivo at the G-box. Because RNA gel blot data indicated that GBFB is itself induced by ABA, we might have concluded that GBFB is indeed the GBF responsible in cell cultures for binding to the Adh G-box and is therefore responsible for ABA-regulated expression of Adh. Potential limitations of this conclusion are exposed by the fact that other GBFs bind the G-box with the same signature as GBF3, and subtle differences between in vivo and in vitro footprint signatures indicate that factors other than or in addition to GBF3 interact with the half G-box element.

INTRODUCTION

The G-box (5'-CCACGTGG-3') is a cis-acting element that is present in the promoters of many plant genes and is respon- sive to diverse environmental stimuli (Katagiri and Chua, 1992; Lu et al., 1992; Brunelle and Chua, 1993; Menkens and Cashmore, 1994). More than 20 cDNA clones encoding G-box binding factors (GBFs) have been isolated from severa1 plant species (Brunelle and Chua, 1993; lzawa et ai., 1993, 1994). GBFs are basic leucine zipper (bZIP) proteins that exhibit a relaxed DNA binding specificity for sequences containing an ACGT core (Schindler et al., 1992b; lzawa et al., 1993). The ACGT core was thought to be essential for GBF binding, al- though sequences flanking the ACGT core affect the binding specificity and affinity (Schindler et al., 1992b; lzawa et al., 1993) of GBF interactions. The bZlP domain of GBFs as well as the amino acid sequence adjacent to the basic domain ap- pear to affect the DNA binding affinity as well (Schindler et al., 1992b). Thus, there is a bewildering array of interaction possibilities created by these various GBFs and the variations on the flanking sequences surrounding the G-box core se- quence. Because of the limited availability of GBF mutants, a direct relationship between in vitro binding and in vivo acti-

Current address: Pioneer Hi-Bred lnternational Inc., Johnston, IA

To whom correspondence should be addressed. 50131.

vation exists only for the Opaque2 and RITA-1 bZlPs (Schmidt et al., 1990, 1992; Varagona et al., 1991; lzawa et al., 1994).

Four of the GBF cDNA clones, GBF1, GBF2, GBF3, and GBF4, were isolated from Arabidopsis by using the tomato small unit of ribulose bisphosphate carboxylase (RbcS) G-box as the oligonucleotide probe for protein interaction screening and then by using GBFl as a hybridization probe (Schindler et al., 1992a; Menkens and Cashmore, 1994). The N-terminal proline- rich domain of Arabidopsis GBFl activates gene transcription in vivo (Schindler et al., 1992b). GBF4 has similarities to the Fos oncoprotein in that it can bind to G-box elements only as a heterodimer with GBF2 or GBF3 (Menkens and Cashmore, 1994). Arabidopsis GBF3 is highly expressed in roots but not in leaves (Schindler et al., 1992a), making it a candidate for tissue-specific regulation of the Arabidopsis alcohol dehydro- genase (Adh) gene (McKendree et al., 1990), which contains a dyad G-box element (McKendree and Ferl, 1992) whose se- quence falls within the range of G-box sequences strongly bound by GBF3.

The G-box at position -214 in the Arabidopsis Adh promoter is one of the functional cis-acting elements that is required for constitutive expression in cultured cells and responsible for abscisic acid (ABA)-related cold and dehydration induc- tion in seedlings (McKendree and Ferl, 1992; Dolferus et al., 1994). In vivo and in vitro DNA-protein interaction analyses have demonstrated that the Arabidopsis Adh G-box interacts

848 The Plant Cell

with a nuclear protein complex (Ferl and Laughner, 1989; DeLisle and Ferl, 1990; McKendree et al., 1990; Lu et al., 1992). One part of the complex is composed of GF14s, which are 14- 3-3 proteins that can be phosphorylated and can bind calcium (Lu et al., 1992,1994). The GF14s do not bind the G-box directly (Lu et al., 1992). Instead, the GF14s appear to interact physi- cally with GBFs in a multiprotein complex. Although all of the known Arabidopsis GBFs can bind to sequences similar to the Adh G-box in vitro, GBFl does not bind with high affinity in a random sequence pool (Schindler et al., 1992b), apparently indicating that GBFl may not be the factor that binds to the Adh G-box element in vivo. Whether GFB2 and GBF3 or other GBF proteins are the in vivo Adh G-box binding factors remains to be determined.

The -214 G-box in the Arabidopsis Adh promoter is similar to the ABA response element present in the wheat Em gene (Guiltinan et al., 1990). Adh is induced by ABA, and because both cold and dehydration stresses have been correlated with increased levels of ABA in plant tissues (Guy, 1990; Hetherington and Quatrano, 1990; Jackson, 1991), a logical conclusion might be that the G-box serves as the final receptor of ABA signals involved in Adh gene expression. Thus, correct identification of which GBF(s) binds to the Adh -214 G-box would lead to increased understanding of the specificity of transcription fac- tors involved in ABA signal transduction.

Within the context of the Arabidopsis Adh promoter, the -214 G-box element lies in proximity to a half-G-box element (Fig- ure 1). The half G-box (5'-CCAAGTGG-3'; located at position -190, just downstream of the G-box) lacks the ACGT core thought to be necessary for bZlP binding but is necessary for

-214 G-bOx -190 half G-box -220 -210 -200 -1 90 -180

I I I I ,__._._....___.__________

GAATACTAGCAA+CCAAGTGG+GAG CTTATGATCGTTFGGTTCACCTTTCTC

. ........................ .

4 3 i 1 d o 1 ;3 4 Position Designation

AAATGCCACGTGGACGAA ATCTTCCACGTGGCATTA

ACAAATGCCAC ACGAATA ACAAATG GTGGACGAATA

AAATGAAACGTGGACGAA AAATGCCAATTGGACGAA AAATGCCACGGGGACGAA

CTAGCAACGCCAAGTGGAAAGAG

Adh G-box a t -214 RbcS G-box G-R G-L GM1 GM2 GM3 Adh half G-box a t - ,190

Figure 1. Arabidopsis Adh G-Box Region.

A portion of the Sflanking region of the Arabidopsis Adh gene isshown. The G-box at position -214 is indicated by the solid-line box. The half G-box at position -190 is indicated by the dotted-line box. The posi- tions of the nucleotides in the G-box are indicated, as designated by lzawa et al. (1993). The oligonucleotides used in this study are indi- cated below the sequence and aligned with their homologous position of the Adh sequence, with blanks indicating deletions in the right half site (G-R) and in the left half site (G-L). Dolferus et al. (1994) refer to the -214 G-box as G-box 1 and the -190 half G-box as G-box 2.

promoter activity in seedlings (Dolferus et al., 1994) and is bound by nuclear proteins in vivo (Ferl and Laughner, 1989). Given the proximity and sequence similarity of the half G-box at position -190 and the dyad G-box at position -214, the re- gion of the Adh promoter encompassing these elements constitutes an interesting sequence for testing assumptions and conclusions regarding biochemical assays for the clon- ing, identification, and characterization of factors associated with regulating gene activity through G-box and G-box-like elements.

This study addresses several questions about GBFs and the abilities of biochemical techniques to distinguish among them. First, do studies with the Adh G-box sequence identify any new GBFs or provide further insights into the existing GBFs that were recovered directly or indirectly by screening with the tomato RbcS G-box element (Schindler et al., 1992a; Menkens and Cashmore, 1994)? Second, do standard biochemical as- says of protein-DNA interactions strengthen the case for specificity between any of the GBFs and the Adh G-box? Third, are GBFs involved in the interactions with the half-G-box ele- ment at position -190, even though that element lacks an ACGT core? Finally, what are the limitations of existing biochemical assays to identify unambiguously proteins that interact with G-box-like sequences, and in particular, which GBF might be involved in ABA signal transduction to the Adh promoter?

RESULTS

Cloning of the GBF3 cDNA

Although many cDNA clones encoding G-box DNA binding bZlP proteins have been isolated from a wide array of plant species, none have been isolated specifically with the Arabidop- sis Adh G-box as a probe. By using the ligated Arabidopsis Adh G-box oligonucleotide to screen a cDNA expression library derived from cell suspension cultures, several positive cDNA clones were identified from among 6 x 106 phage plaques. Preliminary sequence analysis indicated that all were various Ytruncations of the same cDNA. To recover potential full-length cDNAs, the same library was rescreened by hybridization with a fragment of one of the initial clones.

There are four stop codons in the 5' untranslated region of GBF3, halting all three reading frames and resulting in the separation of the truncated 0-glactosidase encoded by the hgt l l vector from the predicted bZlP protein. Similar situations are present in GBF1, GBF2, and GBF4 cDNAs (Schindler et al., 1992a; Menkens and Cashmore, 1994). This may be why only 5' truncated clones were recovered during the interac- tion screen, whereas full-length clones were recovered by hybridization. The longest clone contains 1610 bp of cDNA and is nearly full length, as indicated by RNA gel blot analysis show- ing an mRNA of 4 . 6 kb. A single extended open reading frame encodes a 41-kD polypeptide of 378 amino acids (Figure 2). Length heterogeneity exists in the 3'ends of the clones, where

Transcription Factor Veracity 849

ATTTGAATTTCTGGGTTTCTCTCTGTTTAAGCTTCTTCTTCTTCATCTTCTGCTTACGTT 60TCTTCTTCAAGGAGCTTTCGGATTCTTGTAGAAAGAGTCATTGTTCTCTTGAGTGGGAAA 120CCTTGAAACCATTCCTATGGGAAATAGCAGCGAGGAACCAAAGCCTCCTACCAAATCAGA 180

M G N S S E E P K P _ P T K S DTAAACCATCTTCACCCCCGGTGGATCAAACAAATGTTCATGTCTACCCTGATTGGGCAGC 240K P S S P P - V D Q T N V H V Y P D W A A

TATGCAGGCATATTATGGTCCAAGAGTAGCAATGCCTCCTTATTACAATTCAGCTATGGC 300H Q A Y Y G P R V A M P P Y Y N S A M A

TGCATCTGGTCATCCTCCTCCTCCTTACATGTGGAATCCTCAGCATATGATGTCACCATC 360A S G H P P P P V M W N P Q H M M S P S

TGGAGCACCCTATGCTGCTGTTTATCCTCATGGAGGAGGAGTTTACGCTCATCCCGGTAT 420G A P Y _ A A V Y P H G G G V Y A H P G I

TCCCATGGGATCACTGCCTCAAGGTCAAAAGGATCCACCTTTAACAACTCCGGGGACGCT 480P M G 3 L P Q G Q K D I P P L T T P G T L

TTTGAGCATCGACACTCCTACTAAATCTACAGGGAACACAGACAATGGAITGATGAAGAA 540L S I D T P T K S T G N T D N G L M K K

GCTGAAAGAGTTTGATGGGCTTGCTATGTCTCTAGGAAATGGGAATCCTGAAAATGGTGC 600L K E F D G L A M 3 L G N G N P C N G A

AGATGAACATAAACGATCACGGAACAGCTCAGAAACTGATGGTTCTACTGATGGAAGTGA 660D E H K R S R N S S E T D G 5 T D G S D

TGGGAATACAACTGGGGCAGATGAACCGAAACTTAAAAGAAGTCGAGAGGGAACTCCAAC 720G N T T G A D E P K L K R S R E G T P T

AAAAGATGGGAAACAATTGGTTCAAGCTAGCTCATTTCATTCTGTTTCTCCGTCAAGTGG 780K D G K Q L V Q A S S F H S V S P S S G

TGATACCGGCGTAAAACTCATTCAAGGATCTGGAGCTATACTCTCTCCTGGTGTAAGTGC 840D T G V K L I Q G S G A I L S P G V S A

AAATTCCAACCCCTTCATGTCACAATCTTTAGCCATGGTTCCTCCTGAAACTTGGCTTCA 900N S N P F M S Q S L A M V P P E T W L O

GAACGACAGAGAACTGAAACGGGAGCGAAGGAAACAGTCTAATAGAGAATCTGCTAGAAG 960E L

GTCAAGATTAAGGAAACAGGC C[ ? R t R

R E R R K Q S,GGAAACACGCCGACACACAAGAACJC.TCCR K | Q A E T E E ( L ) AJGGCATTAAGATCTGAACTAAACCAACI

R E S A R RjJGCTAGGAAAGTGGAAGCCTIGAC 1020

_________ A R K V E A ( L ) TAGCCGAAAACATGGCATJI^GATCTGAACTAAACCAACJTAATGAGAAATCTGATAXACT 1080

A E N M A ( L ) R ) S E L N Q ( L ) N E K S D K ( L )AAGAGGAGCAAATGCAACCTIGTTGGACAAACTGAAATSCTCGGAACCCGAAAAGAGAST 1140

R G A N A T U O L D K L K C S E P E K R VCCCCGCAAATATGTTGTCTAGAGTTAAGAACTCAGGAGCTGGAGATAAGAACAAGAACCA 1200

P A N M L S R V K N S G A G D K N K N QAGGAGACAATGATTCTAACTCTACAAGCAAATTCCATCAACTGCTCGATACGAAGCCTCG 1260

G D N D S N S T S K F H O L L D T K P RAGCTAAAGCAGTAGCTGCAGGCTGAATCGATGGTAATTCATGTCGATTTCTACTTAATTT 1320A K A V A A G *

GTCGACATAAACAAAGAAAATAAGTGCTACTAATTTCAGAAAAACTTGATAGATAGATAG 1380TATAGTAGAGAGAGAGAGAGAGAGAGAGGTGTGATGATTATTGATCTATAAATTTTCGGA 1440GAGAGAGAGGGAGAAAGAGAAACTTTTCCTCCAGATCAAAATTTGGTGTTATGGTTTGTT 1500ACTGTTAATATAGAGAGGCTTTTCTTTTTTTATAAAATGGCTTCCTTTGTTGC^TTTCCT 1560TGTTTTAGACCTGATGTAATTTTATGAAATCGGTGTTATTGCTlfSCGTJf* 1611

Figure 2. Arabidopsis GBF3.

The sequence of the longest GBF3 cDNA is indicated, with the nucleo-tides numbered to the right. The derived amino acid sequence isindicated below the nucleotides in the single-letter designation.The proline-rich domain is underlined, the basic region is boxed, andthe leucine residues of the zipper region are circled. The locationsof the various 3' ends of the cDNAs are indicated by arrows near theend of the sequence. The position of the N-terminal deletion of GBF3aand GBF3b is indicated by a bracket at nucleotide 455. The positionof the C-terminal deletion of GBF3b is indicated by a bracket at nucleo-tide 1041. The stop codon is indicated by an asterisk. The GenBankaccession number is U151850.

poly(A) tails of up to 150 bases are found at three different sitesin the cDNA clones. All of the positive clones are homologsof GBF3, indicating a relative abundance of GBF3 cDNA inthe suspension culture cells, which also express Adh (Ferl andLaughner, 1989).

The derived primary amino acid sequence of these clonesis, for the most part, identical with that of GBF3 (Schindler etal., 1992a). This full-length version of GBF3, however, contrib-utes unique information on GBF structure in that it apparentlycompletes the proline-rich N terminus of the protein and pro-vides the 5' nontranslated region.

Expression of GBF3 mRNA

Both cold and dehydration stresses are apparently related toincreased levels of ABA (Guy, 1990; Hetherington and Quatrano,1990; Jackson, 1991). To address the possible relationshipsamong Adh expression, the G-boxes, GBF3, and ABA, we ex-amined the effects of various treatments on the expressionof GBF3 in cell suspension cultures. As shown in Figure 3,treatment with additional ABA resulted in a fourfold inductionof GBF3 transcript accumulation.

At this point, the data indicated that we had cloned a knownGBF by using the G-box found within the context of the Adhpromoter as a probe in a standard protein-DNA interactionscreen. Although the cDNA clones extended existing informa-tion on GBF3, we failed to find any new or unique GBFs dueto any novel features that the Adh G-box might have. However,by conducting the screen under stringent interaction condi-tions and thereby finding only one species of interacting clone,combined with the data on induction characteristics of GBFSmRNA, we made the tentative inference that GBF3 is in factthe G-box factor that interacts with the Adh G-box in suspen-sion cells. Because GBF3 is itself induced by ABA, the extendedconclusion would be that GBF3 delivers ABA-related signalsto the Adh regulatory apparatus through binding to the G-box.

C IAA ABA minus cold2,4-D

Figure 3. Accumulation of GBF3 mRNA in Suspension Cultures.

An RNA gel blot hybridized with GBF3 is shown, together with a bargraph of the quantitative data derived from dot blot hybridizations ofthree replicated experiments. The data were normalized against actinmRNA (solid bars) and rRNA (open bars). The error bars indicate thestandard deviation for the three replicate experiments in each set. Cellsuspension cultures were either untreated (C) or treated with 20 nMindolacetic acid (IAA) or 50 nM (ABA) for 24 hr. Additional treatmentsincluded removing 2,4-D from the media (minus 2,4-D) or placing thecultures at 4°C (cold) for 24 hr before isolation of total RNA.

Adh expression is regulated by cold and dehydration medi-ated through the -214 G-box element (Dolferus et al., 1994).

850 The Plant Cell

Purification from Escherichia coli and BiochemicalCharacterization of GBF3

To characterize more fully the interactions between GBF3 andthe Adh G-box, two forms of the G BF3 protein were expressedand purified from E. coli. The full-length GBF3 protein waslargely insoluble and presented several purification problems.However, N-terminal truncation at amino acid 106 greatly en-hanced the solubility of recombinant GBF3. Two truncationswere produced and are shown in Figure 4. GBF3a containsamino acids 106 to the C terminus at 382, whereas GBFSbcontains amino acids 106 to 302 and is thus additionally trun-cated at the C terminus in the middle of the leucine zipperdomain. The GBF3a and GBFSb truncations were expressedin the pET-15 system and purified from the soluble lysate byimmobilized metal affinity chromatography made possible bythe short N-terminal histidine tract provided by the vector.The truncated GBF proteins were purified further by usingSuperdex-75 column chromatography. GBF3a and GBF3b werehighly purified as determined by staining after SDS-PAGE (Fig-ure 4).

The final purification using Superdex-75 column chroma-tography confirmed the dimeric structure of both GBF3a andGBF3b, even though GBFSb is missing much of its leucinezipper domain. GBFSa resolved as a native protein peak at~69 kD (data not shown), whereas the protein in the peak frac-tions consisted of a band at ~34 kD, as determined by

M GBFSa GBFSb

46kD

30kD

21kD

f

GBF3

GBFSa

GBFSb

Figure 4. Purification and Characterization of Recombinant GBF3.

Two peak fractions from Superdex-75 chromatography of recombinantGBF3a and GBF3b were analyzed by using SDS-PAGE with the fol-lowing markers in lane M: ovalbumin at 46 kD, carbonic anhydraseat 30 kD, and trypsin inhibitor at 21 kD. Intervening lanes between thoseof GBFSa and GBFSb were removed. Diagrammatic representationsof the final GBF recombinant proteins are shown below the gel.

SDS-PAGE (Figure 4). The truncated GBF3b protein resolvedas peak at ~50 kD on the Superdex-75 column and produceda protein doublet at ̂ 26 kD after SDS-PAGE, indicating thatGBFSb also exists as a homodimer despite its truncated leu-cine zipper. Formation of the doublet band is a consistent butpoorly understood phenomenon in the production of GBFSb.Additional confirmation of native dimer formation was obtainedby the treatment of native recombinant GBF3 with 10 mMglutaraldehyde, which resulted in dimer-sized peptides whenanalyzed by SDS-PAGE (data not shown).

Specific Binding of GBF3 to the G-Box and Half G-Box

Standard electrophoretic mobility shift assays were used todetermine the binding parameters of GBF3 to the G-box at po-sition -214 and the half G-box at position -190, as found inthe context of the Adh promoter. The in vivo dimethyl sulfate(DMS) footprinting assay had demonstrated that both the G-boxat -214 and the half G-box at -190 were bound by nuclearproteins (Ferl and Laughner, 1989). As shown in Figure 5A,both labeled the -214 G-box element, and the -190 half-G-box element formed single retarded protein-DNA complexeswith purified GBFSa. The more severely truncated GBFSb alsoformed a single shifted complex with both sites. These com-plexes were stable in the presence of 1000-fold excess poly-(dl-dC) in the binding reaction as a nonspecific competitor, andthe binding was dependent on the concentration of the addedGBF protein for all four protein/site combinations.

Competition assays further defined the specificities and af-finities of the GBF3 proteins for G-box elements. As shownin Figure 5B, the -214 G-box-GBF3a complex was reducedby competition with the 100- to 200-fold unlabeled flbcS G-boxelement, indicating that the complex is quite specific for the5'-GCCACGTGGA-3' extended core G-box sequence (see alsoFigure 1). However, neither the right half site (G-R) nor the lefthalf site (G-L) of the Adh G-box competed for GBFSa binding(Figure 5B), initially suggesting that the intact dyad G-box motifis essential for the binding of GBFSa and that half-G-box sitesare not sufficient to create a binding site for GBF3. To charac-terize further the important nucleotides in the central G-boxcore motif, three additional G-box mutants (GM1, GM2, andGM3) and the -190 half G-box (see also Figure 1) were usedin the competition assay. None of the mutants could competewith labeled G-box oligonucleotide for binding to GBFSa, evenat 200-fold excess of the competitor (Figure 5B, top). Thesedata suggest that (1) the two C residues at positions -3 and-2 of the G-box cannot both be mutated and still maintainstrong binding competency (see GM1); (2) the central CG pairat positions -0 and 0 of the dyad cannot be mutated as a pairand still maintain strong binding competency (see GM2); (3)the T residue at position 1 of the G-box is essential for strongbinding competency (see GM3); and (4) the C residue at posi-tion -0 of the G-box also is essential (see -190 half G-box).

The truncated GBFSb protein demonstrated a similar set ofbinding competitions, although the apparent binding was

Transcription Factor Veracity 851

-214G-box -190 half G-boxGBF3b GBF3a - GBF3b GBF3a

B

0)

g0.

X

26

to .a"6

COMPETITIONS

2Q. S O Q O QQ Q O O O O O OO

O O O O O O O O O O O O OT— fsj T— CM T— C M » — C N J » — C M * — C N l T — CM

I*

Figure 5. Electrophoretic Mobility Shift Assays of GBF3.(A) The binding of GBF3a and GBF3b to the G-box and the half G-box in the presence of poly(dl-dC) as nonspecific competitor was analyzedby using electrophoretic mobility shift assays. At left and right, the first lanes contain no GBF protein, and the next three lanes contain 1, 3, and5 ^L of recombinant GBF protein extract. Horizontal triangles indicate the increasing protein.(B) Electrophoretic mobility shift assays were conducted with the addition of 100- and 200-fold excess of specific competitor sequences, as indi-cated at the top. The designations of the competitors are detailed in Figure 1. In the top row, the -214 G-box is the labeled probe and GBFSais the protein. In the middle row, the G-box is the probe and GBFSb is the protein. In the bottom row, the -190 half G-box is the probe and GBF3ais the protein.

generally weaker, revealing some slight competition by GM1,GM2, and GM3 (Figure 58, middle). Nonetheless, similar clearconclusions about the binding specificities of GBF3 were ob-tained from both GBF3a and the additionally truncated formGBF3b. Thus, binding specificity appears to be determinedby the bZIP region of the GBFs, and truncation of the leucinezipper may slightly reduce overall DNA binding capacity butdoes not affect sequence specificity.

All of the competition data presented here are consistentwith previous data on GBF binding to G-box-like sequencesand, in the context of our current work, immediately call intoquestion the earlier conclusion that GBF3 specifically inter-acts with the -190 half-G-box element of Adh. However,competition assays with the -190 half G-box as the labeledprobe revealed a second layer of specificity in GBF3-DNA in-teractions. When the -190 half G-box was used as the probe(Figure 5B, bottom), it was quickly reduced by competition withthe RbcS G-box. This is in keeping with the failure of the halfG-boxes to compete for binding to the full G-box. However, bind-ing to the -190 half G-box was not competed by the G-boxdeletions (left half site and right half site) or GM2.

These data are consistent with a conclusion that the -190half G-box is indeed a specific binding site for GBFS. How-ever, the susceptibility of the -190 half G-box to competitionsuggests an overall affinity for GBF3 much lower than that ofthe complete dyad -214 G-box. In addition, the competitiondata with the -190 half G-box as a probe indicate a hierarchyof importance to binding for several of the internal residuesof the G-box: (1) the central CG pair at positions -0 and 0 can-not both be mutated and maintain binding to GBF3, because

GM2 fails to compete even with the -190 half G-box; (2) theCC pair at positions -3 and -2 may both be mutated andmaintain some binding activity, because mild competition wasseen with GM1; and (3) the T residue at position 1 is the leastrequired of the central elements tested, because GM3 nearlyfully competed for binding. The main conclusion from thesedata is that both the -214 dyad G-box and the -190 half G-boxare specific target sites for GBFS binding.

DNase I Footprints

The ability of GBF3 to bind specifically to the -214 G-box andthe -190 half G-box was further examined by DNase I footprintanalysis by using GBF3a and 5'end-labeled double-strandedDNA extending from positions -107 to -300 of the Adh 5' flank-ing region. Figure 6 shows the footprints generated byincreasing concentrations of GBF3a. The bottom strand foot-prints of a 20-fold range of concentration of GBF3a areexamined in Figure 6A. A clear and distinct pattern of DNaseI footprints is evident, with each individual footprint character-ized by a dramatic hypersensitivity at its 3' terminus. At lowGBFSa concentrations, a DNase I footprint can be observedover the -214 G-box, whereas the -190 half G-box is largelyunoccupied. However, at higher concentrations, footprintingis evident over both the -214 G-box and the -190 half G-box.Examination of the increase in footprint strength as a functionof increasing GBFSa concentration indicates an ~10-fold differ-ence in binding affinity for GBFSa to the dyad -214 G-boxcompared with the -190 half G-box (compare Figure 6A, lanes

852 The Plant Cell

BBottom Strand Top Strand

.1 .5 1 5 0 G [GBF3a] G 0 1 2 4 8 0

*****- - - - t

« , * * * •fttttt

Figure 6. In Vitro DNase I Footprint Analysis of GBF3 in the AdhPromoter.

(A) The binding of GBF3a to the bottom strand of the Adh promoteris demonstrated by DNase I footprint analysis.(B) The binding of GBF3a to the top strand of the Adh promoter is shown.For both (A) and (B), the marker lane (G) and positions of selectedG residues are indicated, and the positions of the -214 dyad G-boxand the -190 half G-box are indicated by solid and dotted boxes, respec-tively. The relative concentration of GBF3a is indicated above each lane.

.1 and 1 as well as lanes .5 and 5). This conclusion regardingbinding affinities of the G-box and the half G-box is consistentwith the earlier conclusion drawn from competition data.

Footprinting of the top strand was conducted over a tighter,eightfold range of GBF3a concentrations to address the pos-sibility of cooperativity in the binding of GBFs to the nearlyadjacent -214 and -190 sites. As shown in Figure 6B, a distinctpair of GBFSa footprints is evident again, with the character-istic hypersensitivity at the 3' terminus of each footprint. The

-190 half-G-box footprint is evident only at relatively high GBF3concentrations. No obvious evidence for cooperativity in bind-ing at the two sites was observed, because the linear increasein GBFSa concentration produced an approximate linear in-crease in the -190 half-G-box footprint (as measured by thedisappearance or gain of bands within the footprints of Figure6B), even while the -214 G-box was fully occupied. Thesefootprinting data strengthen an emerging model in which GBF3binds to both the -214 G-box and the -190 half-G-box ele-ments in cultured cells, and the -190 half G-box representsa specific but lower affinity site for GBFS binding.

DNA Binding Signatures of GBF3 at the G-Boxand Half G-Box

As a final test of specificity and identity, the in vitro DMS foot-printing signature of GBFSa was compared directly with theDMS signature obtained by in vivo footprinting of Arabidopsissuspension cells. The in vitro and in vivo DMS footprint signa-tures shown in Figure 7 were obtained from separateexperiments. Therefore, the run lengths are different but thepatterns of G-residues are directly comparable. Also, becausethe in vivo footprinting procedure involves hybridization ofgenomic DNA electrotransferred from sequencing gels, thein vivo footprints are somewhat less well resolved.

Bottom Strandvitro vivo

Top Strandvivo vitro

-226-217/168

-214°

-193,-192*

-179

- -182S -186/7

-189

g-210/11°-213•=-218

-225

Figure 7. Comparison of in Vitro and in Vivo DMS Footprint Signatures.

DMS footprint signatures for the G-box region of the Adh promoter areshown for the bottom strand (left) and top strand (right). The in vivofootprints are from McKendree et al. (1990); the top strand in vivo laneshave been electronically inverted to correspond to the direction of theother lanes. Lanes designated (-) indicate naked DNA analyses thatdemonstrate the reactivities of G residues in the absence of proteins.Lanes designated ( + ) indicate that GBFSa was added to the reactionduring the in vitro analyses or that intact cells were treated in the invivo analyses. The regions of the -214 dyad G-box and the -190 halfG-box are indicated by solid and dotted boxes, respectively, and spe-cific interactions are indicated by filled ovals for enhancements andopen ovals for protections. In vivo interactions at positions -182 and-194 are italicized and indicated by dotted ovals.

Transcription Factor Veracity 853

As shown in Figure 7, the DMS footprint of in vitro GBF3ainteractions at the -214 G-box is very similar to the in vivofootprint for both the bottom and top strands. At the -214 G-box,there is a bottom strand enhancement at position -217 andprotections at positions -216 and -214 visible in the in vitroGBF3a footprint. The same pattern can be observed in thein vivo footprint, and although the doublet at positions -217and -216 is difficult to resolve in the in vivo footprint, the pat-tern is clearly consistent with the in vitro footprint of GBF3a.On the top strand, protections at positions -218, -213, and-211 and a slight enhancement at position -210 character-ize the GBF3a footprint in vitro as well in vivo.

A set of qualitatively similar interactions also characterizesthe -190 half-G-box footprints. In both the in vivo and the invitro footprints, there is protection of position -192 and en-hancement of position -193 on the bottom strand, as well asprotection of positions -187 and -189 and slight enhance-ment of position -186 on the top strand. In this case, however,differences exist between the in vitro and in vivo footprints onthe top strand. The first involves a prominent protection at po-sition -182 that is observed in vivo but not in vitro. The seconddifference involves a more subtle protection at position -194that is fairly well observed in vivo but difficult to establish in vitro.

Thus, the comparison of in vivo and in vitro DMS bindingsignatures is consistent with the conclusion that GBF3 is theGBF responsible for protein interactions with the -214 G-boxelement of the Adh promoter region. However, the pattern ofinteractions observed in vivo at the -190 half G-box and thepattern produced by GBF3 in vitro are clearly not identical,indicating the possibility that factors other than or in additionto GBF3 may be responsible for in vivo interactions with thatelement.

DNA Binding Signatures of Truncated GBF3and Maize GBF1

_-186°-187o-189

• -210

'-213

o-218

Top Strand

in •»

,-217'-216

'-214

-193o-192

Bottom Strand

Figure 8. In Vitro Footprint Signatures of GBF3a, GBFSb, and MaizeGBF1.

The G-box (-210 to -218) and half G-box (-186 to -193) regions ofDMS footprinting gels are shown, with the top strand at the top andthe bottom strand at the bottom. In vitro interactions are indicated byfilled and open ovals, as given in Figure 7. mGBFl, maize GBF1.

DISCUSSION

To test further the veracity of these conclusions and as an ad-ditional approach to defining the discrimination limits of theDMS footprint signature, the DMS footprints of GBF3b andmaize GBF1 on the Arabidopsis Adh promoter were deter-mined. GBFSb lacks the entire C-terminal portion of GBF3 andmore than half of the leucine zipper, and thus GBFSb teststhe effect of the additional truncation on the DMS footprint sig-nature of GBF3. Maize GBF1 is quite diverged from ArabidopsisGBF3 in overall amino acid sequence but is similar within thebasic DNA binding region (deVetten and Ferl, 1995). Thus, thebinding signature of maize GBF1 essentially tests the effectof substituting the entire protein outside of the basic region.As shown in Figure 8, the DMS footprint signatures of maizeGBF1 and the truncated GBFSb are similar to that of GBFSa.It appears then that the DMS footprint signature is a productonly of the intimate contacts that occur between DNA and thebasic region of GBFs and is insensitive to massive changesin GBF structure outside of the basic region.

Studies of the G-box and G-box-like elements of the Arabidop-sis Adh gene have led to questions regarding the signaltransduction pathways involving cold, dehydration, and ABA.As the apparent DNA binding termini of these signaling path-ways, GBFs play a major role in transducing signals to the Adhpromoter, and a clear understanding of which GBF interactswith which Adh element becomes a key issue. However, thesesorts of fundamental issues surrounding the identification ofrrans-acting factors extend well beyond the current study andpresent questions that could be addressed in any element-fac-tor relationship.

Search for GBFs That Interact with Adh FindsOnly GBF3

Only GBF3 was recovered in the interaction screen of the li-brary by using commonly accepted methods and stringencies.

854 The Plant Cell

Although some new information about GBF3 with regard to its N-terminal sequences and its multiple 3’ends was obtained, no new GBFs were recovered from the screen. Thus, it would appear initially that the Adh G-box selectively identifies GBF3 from among the many bZlPs that are likely to be encoded within the cDNA library. At this point in our study, GBF3 becomes a candidate to be the GBF responsible for carrying signals to Adh through specific interactions with the G-box elements.

GBF3 1s lnduced by ABA and Binds Specifically to G-Box Sequences

GBF3 mRNA accumulation is induced by ABA. The subse- quent potential increase in the GBFB concentration supports the idea that GBFB plays a role in conducting ABA-related sig- nals to the Adh G-box. The electrophoretic mobility shift assay and DNase I footprint analysis confirmed that GBF3 binds spe- cifically to the dyad G-box at position -214 of the Adb promoter. The leve1 of specificity (as demonstrated by competition as- says using the Adh G-box as a probe and DNase I footprinting of the Adh promoter) supports the conclusion that GBF3 is the GBF that interacts with the Adh G-box in vivo and is consonant with expectations based strictly on in vitro binding data (Izawa et al., 1993). GBF3 also is capable of specifically interacting with the half G-box at position -190, albeit at a lower overall affinity. This conclusion is surprising primarily because the -190 half G-box site lacks the ACGT central core thought to be absolutely required for bZIP/GBF binding (Izawaet al., 1993).

Does GBF3 Bind to the G-Box in Vivo?

The Arabidopsis Adh 5‘flanking region is one of the few plant promoters to be footprinted in vivo. The intrinsic value of in vivo DMS footprinting is twofold. First, clusters of interactive G residues can identify sequences as potential cis-acting ele- ments, thereby providing target sequences for mutational analysis. In fact, the,G-box at position -214 and the half G-box at position -190 both were identified initially by in vivo foot- printing and then subsequently confirmed as cis-acting elements by transgenic analyses (Ferl and Laughner, 1989; McKendree and Ferl, 1992; Dolferus et al., 1994). Second, in vivo DMS footprinting provides a DNA binding signature for the protein-DNA interaction that occurs in the living cell. This binding signature is defined by the pattern of enhancements and protections observed in the footprinted region and is the result of the presumably unique interactions that occur between DNA binding proteins and their recognition sequences. An im- mediate application of in vivo footprinting would be the ability to identify uniquely the protein-DNA interaction in vitro and thereby confirm the identity of the factor responsible for the in vivo footprint and the regulatory effect of the target cis-acting element.

GBF3 meets all of the generally accepted in vitro criteria for DNA binding specificity for the Adh G-box. In addition, the

in vitro DMS footprint signature of GBFB at the -214 G-box matches the in vivo footprints within the limits of resolution of the technique. Therefore, the primary conclusion might be that GBF3 is responsible for binding to the -214 G-box in sus- pension cells, thereby bringing cold, dehydration, and perhaps other ABA-related signals to the Adh promoter via interactions with the -214 G-box.

The half G-box at position -190 appears to be a relatively low-affinity site for GBF3 binding as defined by in vitro DNA binding studies. These data alone might lead to the inference that the GBF3 binding to the half G-box is weak. And because the -190 half G-box lacks the ACGT core sequence, we might speculate that a heterologous, non-GBF pairing partner is re- quired to create a high-affinity interaction with the -190 half G-box. Careful comparison of in vivo and in vitro footprints sup- ports this conjecture. An obvious interaction at position -182 that was observed in vivo was not observed in vitro with GBF3, and a more subtle interaction at postion -194 appeared to be similarly absent in the in vitro footprint. Although position -182 is outside the generally accepted limits of the G-box sequence, it does lie within the DNase I footprint of GBF. Do these obser- vations mean that homodimeric GBFB is not the factor bound to the -190 half G-box in vivo? At the very least, recombinant GBF3 alone is not capable of reconstructing the entire foot- print signature in the -190 half G-box.

The Value and Limits of Current Biochemical Methods in General and DMS Footprinting Signatures in Particular

Although supporting the possibility that GBF3 interacts with the G-box and perhaps even the half G-box, DMS footprinting itself cannot exclude other GBFs from a similar role. DMS foot- printing cannot distinguish GBF3a from its more severely truncated form, GBF3b. Therefore, it seems that protein do- mains outside of the bZlP region have little influence on the DMS signature. The in vitro binding signature of maize GBFl also is identical to that of GBF3. Because maize GBF1 shares extensive homology with GBFB only in the basic region, it would seem that any bZlP with sequence conservation to GBF3 in the basic domain might leave the same footprint signature. Although this has yet to be tested experimentally, there are many bZlP proteins with basic region sequences quite simi- lar to those of GBF3 and maize GBF1. These include at least GBFl and GBF2 from Arabidopsis as well as EmBP-1, TAF-1, CPRF-1, and CPRF-3 from other species (Izawa et al., 1993).

Can we determine which GBF is responsible for carrying ABA-related signals to Adh? Given the apparent discrimina- tion limits of DMS footprinting, the answer is clearly “no, not yet.” At the least, all related GBFs should be tested to see whether they bind this G-box with the same signature as GBF3. We are now pursuing this experiment. It may be that all type I GBF interactions (Izawa et al., 1993) will leave the same DMS footprint signature on the Adh dyad G-box at position -214. Therefore, it is possible that current biochemical methods that

Transcription Factor Veracity 855

allow comparison of in vivo and in vitro binding data would not be able to distinguish among the GBFs and other bZlP homologs likely to exist in any cell type.

Additional in vivo and in vitro methods are needed if the bio- chemical approach is to be relied on to identify uniquely a specific protein-DNA interaction out of a family of possible in- teractions. Clearly, some in vitro methods can and do distinguish among GBFs: Arabidopsis GBF2 can be distinguished from GBFB by methylation interference (Schindler et al., 1992b), and tomato GBF9 can be distinguished from GBF12 and GBF4 by DNase I footprint analysis (Meir and Gruissem, 1994). How- ever, neither methylation interference nor DNase I footprinting can yet be done in vivo, hence limiting their application for the purpose of correlating specific factors with specific re- sponses and elements. Additional correlating methods, such as assaying the effect of expression of a transcription factor on reporter gene expression or the characterization of the phenotypes of mutant transcription factors, would certainly help to solidify the potential role(s) of transcription factors, but that approach also may be limited in many cases by overlapping expression patterns and activities of related family members. Therefore, until the biochemical approaches are strengthened or combined with clear mutant data, assignment of any tran- scription factor, particularly any GBF-like factor, to a unique or specific function will remain a tentative and inexact correla- tive science.

Model for GBF lnteraction wi th Adh

We propose the following model for transcription factor inter- actions in the -190 to -214 region of the Arabidopsis Adh promoter (see Figure 9). For the dyad G-box at position -214, a GBF is responsible for bringing ABA-related signals to the promoter. GBF3 is a likely candidate, because it leaves the proper footprint, is expressed in the proper tissues to control Adh, and is itself induced by ABA. However, this conclusion must be tempered by the observation that other GBFs may leave the same signature, and until all GBFs and related bZlPs are fully characterized with respect to expression patterns, in- duction by ABA, and DMS footprints, there is a possibility that other factors may exhibit an equal or better claim to the role of ABA-regulated expression of Adh. GBF3 alone also can spe- cifically bind the half G-box at position -190 in vitro, but differences in the in vivo and in vitro binding signatures sug- gest that factors other than or in addition to GBF3 deliver signals to the -190 half-G-box element.

METHODS

Library Screening and DNA Sequencing

An Arabidopsis thaliana suspension cell cDNA expression library was screened for interaction with the ligated alcohol dehydrogenase (Adh)

GBF3 delivers Factors other than ABA-related signals

to the G-box or in addition to GBF3

interact ai me half Gbox

-220 -210 -zoa - 1 9 0 -180

c;! .......... ee.~..i , I GAATACTAGCAA CCAA T AAAGRGCGTT CTTATGATCGTT+:TTCC$TTCTCGCAA

A A

ma DNAse-l protection o DMS protection A DNAse-l enhancement I DMS enhancement 7 DMS orotection observed in vivo but not in vitro

Figure 9. Summary of GBF lnteraction with the Arabidopsis Adh G-Box Region.

The Arabidopsis Adh 5'flanking region from positions -180 to -220 is presented, along with the matching in vitro GBF3a and in vivo DMS binding signatures at the -214 G-box (solid-line box) and -190 half G-box (dotted-line box). The model indicates that the binding of a sin- gle GBF type can accommodate all of the current in vitro and in vivo DNA binding data for the -214 G-box. GBF3a is a likely candidate and meets all of the criteria for binding, but the current technology cannot prove that binding to the G-box is accomplished exclusively by GBF3. Other GBFs may meet the current DNA binding criteria, so suggestions as to which GBF is responsible for a particular signal re- main based on correlative data. The binding at the -190 half G-box may be accomplished in vivo by GBF3, but differences between in vivo and in vitro interactions at positions -194 and -182 suggest that GBF3 binding alone cannot account for the in vivo interaction data. Thus, factors other than or in addition to GBF3 are likely to be involved.

-214 G-box oligonucleotide (5'-AGAAATGCCACGTGGACGAATA-3 see Figure I)(Miskimins et al., 1985; Sambrooket al., 1989). Thedouble- stranded oligonucleotide was 5' end labeled with Y-~~P-ATP and T4 polynucleotide kinase. The binding buffer was 10 mM Hepes buffer, pH 8.0, containing 50 mM NaCI, 10 mM MgCIZ, 0.1 mM EDTA, 1 mM DTT, and 0.25% nonfat dry milk. Filters were washed in three changes of binding buffer containing 100 mM NaCl for ~ 9 0 min. After screen- ing 6 x 106 bacteriophage plaques, positive clones were isolated and plaque purified. One of the positive cDNA clones (G-box factor 3 IGBF3J) is 1494 bp long (Figure 2, positions 155 to 1649). Four positive clones were obtained from 2 x 105 phages of the same library when rescreened by hybridization with the EcoRI-BamHI fragment of the 5'region of the GBFB cDNA (Figure 2, positions 155 to 450). The clon- ing and sequencing of the DNA inserts were as described previously (Lu et al., 1992).

Expression and Characterization of Recombinant GBF3

The full-length GBF3 and truncated GBF3a and GBF3b expression constructs were made as follows. The EcoRl fragment of the GBF3 cDNA was subcloned into the EcoRl site of the pET3C vector (Lu et al., 1992) to express the full-length GBF protein. The 1221-bp GBFB cDNA BamHl fragment (from position 450 to the BamHl site in the pUC18 vector) and the BamHI-Bglll fragment (from position 450 topo- sition 1039) were subcloned into the BamHl site of pET15b (Novagen, Inc., Madison, WI) to express truncated GBF3a and GBF3b. Correct

856 The Plant Cell

orientations were confirmed by restriction analysis and DhJA sequenc- ing. When expressed in fscherichia coli, the full-length GBF3 protein mainly pelleted in inclusion bodies (data not shown). However, the trun- cated GBF3a (from residues 106 to 382) and GBF3b (residues 106 to 302) are largely soluble. To characterize the GBF3 binding activity, the truncated GBF3a and GBF3b forms were expressed and purified.

The pET15b vector provides the expressed GBF3 proteins with a histidine tract at their N termini, allowing purification by Ni+-charged immobilized metal-affinity chromatography acccjrding to the manufac- turer’s protocol (Novagen). The peak fractions were combined and further purified by Superdex-75 (Pharmacia) column chromatography in NEBD buffer (20 mM Hepes, pH 7.6, 100 mM KCI, 0.1 mM EDTA, 5 mM P-mercaptoethanol, and 10% glycerol). Superdex-75 chroma- tography and SDS-PAGE of purified GBF proteins were performed as described previously (Lu et al., 1994).

Electrophoretic Mobility Shift Assay

The electrophoretic mobility shift assays were performed as described previously (McKendree et al., 1990). DNA oligonucleotides were 5’end labeled with Y-~~P-ATP and T4 polynucleotide kinase (Sambrook et al., 1989). A total of 15 pL of binding reaction contained 0.5 ng of 32P-labeled probe, 1 pg of poly(d1-dC), 2 pg of tRNA, I pg of BSA, and GBF3a, or GBF3b in NEBD buffer. After incubating for 15 min at room temper- ature, the samples were separated by electrophoresis in a nondenaturing 5% polyacrylamide gel. When used, the unlabeled competitors were added to the reaction mixture before the labeled probe.

Probes for Dimethyl Sulfate and DNase I Footprinting

Double-stranded probes for developing DNase I footprints and dimethyl sulfate (DMS) footprint signatures were produced by polymerase chain reaction, using oligonucleotides whose 5’ ends correspond to posi- tions -300 (on the top strand) and -107 (on the bottom strand). One of the oligonucleotides was end labeled with y3’P-ATP and T4 poly- nucleotide kinase (Sambrook et al., 1989) and then paired with the unlabeled opposing primer for amplification of plasmid templates of the Adh Yregion. The resulting strand-specific end-labeled probe was purified by agarose gel electrophoresis and recovered by filter cen- trifugation and precipitation.

DNase I Footprinting

The labeled probe was incubated with purified GBF in NEBD buffer for 15 min, and then DNase 1(20 nglpL) was added to the reaction. After digestion for 60 sec at room temperature, the reaction was stopped by adding one-tenth volume of 5% SDS and 100 mM EGTA and vor- texing. After two chloroform extractions, the DNA was precipitated by the addition of me-tenth volume of 3 M NaOAc containing 5 pg of tRNA and 2.5 volumes of ethanol. Samples were separated by electropho- resis in a 7 M urea-8Vo polyacrylamide sequencing gel along with the G reaction marker of the same fragment (Maxam and Gilbert, 1980).

then maintained for 2 min at room temperature before phenol-chloro- form extraction and precipitation with ethanol and tRNA. After the piperidine cleavage reaction, samples were separated by electropho- resis in 7 M urea-8% polyacrylamide sequencing gels (Maxam and Gilbert, 1980).

The in vivofootprinting lanes that appear in Figure 7 are taken from the same autoradiographs as in vivo DMS experiments of McKendree et al. (1990). The film was scanned, and the top strand in vivo lane was electronically inverted to align with the gel from in vitro treatments.

RNA Analysis

RNA was isolated from Arabidopsis suspension cells as described pre- viously (Lu et al., 1992). Four days after subculture, cellswere subjected to 4’C or were treated by the addition of 20 pM indolacetic acid or 50 pM abscisic acid (ABA), or by the removal of 2.3 pM 2,4-D for 24 hr with constant shaking. RNAwas isolated from the treated cells, sepa- rated on formaldehyde-agarose gels, and transferred onto a GeneScreen membrane (Du Pont-New England Nuclear). Three replicate experi- ments also were analyzed by dot blot analysis. The blots were probed by hybridization with the GBF3 cDNA, the Arabidopsis Atc4 actin gene (Nairn et al., 1988), and rRNA at high stringency (Paul and Ferl, 1991). Quantitative data of hybridization intensity were obtained from a Mo- lecular Dynamics (Sunnyvale, CA) Phosphorlmager. The amount of GBF3 mRNA was determined relative to the amount of rRNA and ac- tin mRNA and graphed along with the standard deviation.

ACKNOWLEDGMENTS

We are grateful to Beth Laughner for her excellent technical assistance and to Paul Sehnke for his advice on purification of recombinant pro- teins. We thank Ernie Almira in the Biotechnology Program DNA Sequencing Core Laboratory at the University of Florida for DNA se- quencing. This work was supported by National lnstitutes of Health Grant No. ROI GM40061 to R.J.F. This is manuscript No. R-05030 of the Florida Agricultura1 Experiment Station.

Received November 8, 1995; accepted March 13, 1996.

REFERENCES

Brunelle, A.N., and Chua, N.-H. (1993). Transcription regulatory pro- teins in higher plants. Curr. Opin. Gen. Dev. 3, 254-258.

DeLisle, A.J., and Ferl, R.J. (1990). Characterization of the Arabidopsis Adh G-box binding factor. Plant Cell 2, 547-557.

devetten, N.C., and Ferl, R.J. (1995). Characterization of a maize G-box binding factor that is induced by hypoxia. Plant J. 7, 589-601.

Dolterus, R., Jacobs, M., Peacock, W.J., and Dennis, E.S. (1994). Differential interactions of promoter elements in stress responses of Arabidopsis Adh gene. Plant Physiol. 105, 1075-1087.

Ferl, R.J., and Laughner, B.H. (1989). In vivo detection of regulatory factor binding sites of Arabidopsis thaliana Adh. Plant MOI. Biol. 12,

DMS Footprint Signature Analysis

After incubating the labeled probe with protein in NEBD buffer (total 50 pL), 1 pL of fourfold-diluted DMS was added into the mixture and 257-266.

Transcription Factor Veracity 857

Guiltinan, M.J., Marcotte, W.M., and Quatrano, R.S. (1990). A leu- cine zipper protein that recognizes an abscisic acid response element. Science 250, 267-271.

Guy, C.L. (1990). Cold acclimation and freezing stress tolerance: Role of protein metabolism. Annu. Rev. Eiochem. 41, 187-223.

Hetherington, A.M., and Quatrano, R.S. (1990). Mechanism of ac- tion of abscisic acid at the cellular level. New Phytol. 119, 9-32.

Izawa, T., Foster, R., and Chua, N.-H. (1993). Plant bZlP protein DNA binding specificity. J. MOI. Eiol. 230, 1131-1144.

Iam, T., Foster, R., Nakajlma, M., Shlmamoto, K., and Chua, N.-H. (1994). The rice bZlP transcriptional activator RITA-1 is highly ex- pressed during seed development. Plant Cell 6, 1277-1287.

Jackson, M.B. (1991). Regulation of water relationships in flooded plants by ABA from leaves, roots, and xylem sap. In Environmental Plant Physiology Series: Abscisic Acid, Physiology and Eiochemistry, W.J. Davies and H.G. Jones, eds (Oxford, UK: Eios Scientific Publishers),

Katagiri, E , and Chua, N.-H. (1992). Plant transcription factors: Pres- ent knowledge and future challenges. Trends Eiol. Sci. 8, 22-27.

Lu, G., DeLisle, A.J., devetten, N.C., and Ferl, R.J. (1992). Erain proteins in plants: An Arabidopsis homolog to neurotransmitter path- way activators is part of a DNA binding complex. Proc. Natl. Acad. Sci. USA 9, 11490-11494.

Lu, G., devetten, N.C., Sehnke, P.C., Isobe, T., Ichimura, T., Fu, H., van Heusden, G.P.H., and Ferl, R.J. (1994). A single Arabidopsis GF14 isoform possesses biochemical characteristics of diverse 14- 3-3 homologues. Plant MOI. Eiol. 25, 659-667.

Maxam, A.M., and Gilbert, W. (1980). Sequencing end-labeled DNA with base-specific chemical cleavage. Methods Enzymol. 65,497-560.

McKendree, W.L., and Ferl, R.J. (1992). Functional elements of the Arabidopsis Adh promoter include the G-box. Plant MOI. Biol. 19,

McKendree, W.L., Paul, A.-L., OeLisle, A.J., and Ferl, R.J. (1990). In vivo and in vitro characterization of protein interactions with the dyad G-box of the Arabidopsis Adh gene. Plant Cell 2, 207-214.

pp. 217-226.

859-862.

Meir, I., and Gruissem, W. (1994). Nove1 conserved sequence motifs in plant G-box binding proteins and implications for interactive do- mains. Nucleic Acids Res. 22, 470-478.

Menkens, A.E., and Cashmore, A.R. (1994). lsolation and charac- terization of a fourth Arabidopsis G-box binding factor which has similarities to FOS oncoprotein. Proc. Natl. Acad. Sci. USA 91,

Miskimins, W.K., Robert, M.D., McClelland, A., and Ruddle, F.H. (1985). Use of a protein-blotting procedure and specific DNA probe to identify nuclear proteins that recognize a region of the transferrin receptor gene. Proc. Natl. Acad. Sci. USA 82, 6741-6744.

Nairn, C.J., Winesett, L., and Ferl, R.J. (1988). Nucleotide sequence of an actin gene of Arabidopsis thalíana. Gene 65, 247-257.

Paul, A.-L., and Ferl, R.J. (1991). Adhl and AdhP regulation. Maydica

Sambrook, J., Fritsch, E.F., and Maniatis, 1. (1989). Molecular Clon- ing: A Laboratory Manual. (Cold Spring Harbor, NY Cold Spring Harbor Laboratory).

Schindler, U., Beckmann, H., and Cashmore, A.R. (1992a). TGAl and G-box binding factors: Two distinct classes of Arabidopsis leu- cine zipper proteins compete for the G-box-like element TGACGTGG. Plant Cell 4, 1309-1319.

Schindler, U., Menkens, A.E., Beckmann, H., Ecker, J.R., and Cashmore, A.R. (1992b). Heterodimerization between light-regulated and ubiquitously expressed Arabidopsis thaliana GEF bZlP proteins.

Schmidt, R.J., Burr, F.A., Aukerman, M.J., and Burr, B. (1990). Maize regulatory gene opaque-2 encodes a protein with a "leucine-zipper" matif that binds to zein DNA. Proc. Natl. Acad. Sci. USA 87,4640.

Schmidt, R.J., Ketudat, M., Aukerman, M.J., andHoschek, G. (1992). Opaque-2 is a transcriptional activator that recognizes a specific target site in 22-kD zein genes. Plant Cell 4, 689-700.

Varagona, M.J., Schmidt, R. J., and Raikhel, N.V. (1991). Monocot regulatory protein Opaque-2 is localized in the nucleus of maize endosperm and transformed tobacco plants. Plant Cell3, 105-113.

2522-2526.

36, 129-134.

EMBO J. 4, 1261-1273.

DOI 10.1105/tpc.8.5.847 1996;8;847-857Plant Cell

G Lu, A L Paul, D R McCarty and R J FerlAdh?

Transcription factor veracity: is GBF3 responsible for ABA-regulated expression of Arabidopsis

 This information is current as of February 5, 2021

 

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists