Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 569

_____________________________ _____________________________

Functional Analysis ofHomeodomain-Leucine Zipper

Transcription Factors inArabidopsis thaliana

BY

HENRIK JOHANNESSON

ACTA UNIVERSITATIS UPSALIENSISUPPSALA 2000

Dissertation for the Degree of Doctor of Philosophy in Physiological Botany presentedat Uppsala University in 2000

Abstract

Johannesson, H., 2000. Functional Analysis of Homeodomain-Leucine ZipperTranscription Factors in Arabidopsis thaliana. Acta Universitatis Upsaliensis.Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science andTechnology 569. 44 pp. Uppsala. ISBN 91-554-4816-X

Homeodomain-leucine zipper (HDZip) proteins constitute a large family oftranscription factors apparently unique to plants. To elucidate the function of thesefactors, the biochemical properties in vitro as well as the effects on transgenic plantswhen expressed at high levels were studied. The conclusion is that HDZip proteins arevery similar with respect to DNA-binding specificity in vitro but appear to be active indifferent aspects of plant development. Thus, functional specificity of HDZip proteinsis most likely determined by other aspects of proteins function, e.g. their capacity tointeract with other proteins.

High-level expression of the HDZip gene ATHB5 in transgenic plants results inhypersensitivity to the inhibitory effect of the plant hormone abscisic acid (ABA) onseed germination and seedling root growth. Furthermore, the expression of ATHB5 ingerminating seedlings is downregulated in the Arabidopsis ABA response mutants abi3and abi5. Together, these data suggest that ATHB5 acts as a regulator of seedgermination and postgerminative growth downstream of ABI3 and ABI5 in an ABAresponse-signaling pathway.

Enhanced levels of the HDZip gene ATHB13 in transgenic Arabidopsis confer asugar-dependent reduction of cotyledon width. In addition, a subset of known sugar-dependent genes was hyperinduced by sucrose in Arabidopsis seedlings overexpressingATHB13. These data suggest that ATHB13 affects both cotyledon morphology and generegulation as a component of a sucrose-signaling pathway.

Loss-of-function mutations in ATHB5 and ATHB13 did not result in any discernablemutant phenotypes, suggesting that these genes are only required under specificphysiological conditions or that they act in a redundant fashion in the plant.

Henrik Johannesson, Department of Physiological Botany, Evolutionary Biology Centre,Uppsala University, Villavägen 6, SE-752 36 Uppsala, Sweden

© Henrik Johannesson 2000

ISSN 1104-232XISBN 91-554-4816-X

Printed in Sweden by University Printers, Uppsala, 2000

This thesis is based on the following papers, which will be referred to in the text by

their Roman numerals:

I. Johannesson, H., Wang, Y. and Engström, P. 2000. DNA-binding and

dimerisation preferences of Arabidopsis homeodomain-leucine zipper

transcription factors in vitro. Plant Mol. Biol., in press.

II. Hanson, J., Johannesson, H. and Engström, P. 2000. Sugar-dependent alterations

in cotyledon and leaf development in transgenic plants expressing the HDZip

gene ATHB13. Plant Mol. Biol., in press.

III. Johannesson, H., Wang, Y. and Engström, P. 2000. High-level expression of the

homeodomain-leucine zipper gene ATHB5 in transgenic Arabidopsis results in

hypersensitivity to abscisic acid. In manuscript.

IV. Johannesson, H. Hanson, J., Söderman, E., Wang, Y. and Engström, P. 2000.

HDZip proteins in Arabidopsis thaliana: a case of functional conservation and

redundancy within a family of transcription factors. In manuscript.

Reprints of papers I and II were made with permissions from Kluwer Academic

Publishers.

CONTENTS

Page

ABBREVIATIONS 6

INTRODUCTION

Arabidopsis thaliana: the model plant 7

Transcriptional regulation 10

Homeodomain proteins 11

Homeodomain proteins in plants: KNOTTED-1 proteins 12

PHD finger proteins 12

HDZip proteins 12

DNA-binding properties of HDZip proteins 14

Dimerization properties of HDZip proteins 16

Function of HDZip proteins 17

RESULTS AND DISCUSSION

Identification of new HDZip genes 19

Identification of HDZip I target sequences in vitro 23

ATHB7 and 12 have alternative binding-site specificities 24

Heterodimer formation between HDZip I proteins 26

Expression analysis of ATHB5 27

A reverse genetics approach to understand the function of HDZip proteins 28

Overexpression of ATHB5 in transgenic Arabidopsis 30

High-level expression of ATHB13 in transgenic Arabidopsis 31

Interpretation of phenotypes resulting from constitutive expression 32

Isolation of T-DNA insertion lines in HDZip I genes 33

SUMMARY 35

ACKNOWLEDGEMENTS 36

REFERENCES 37

ABBREVIATIONS

ABA abscisic acid

abi ABA insensitive mutant

ACC 1-aminocyclopropane-1-carboxylic acid

ATHB Arabidopsis thaliana homeobox

BAC bacterial artificial chromosome

bZip basic leucine zipper

cDNA complementary DNA

HD homeodomain

HDZip homeodomain leucine zipper

HTH helix-turn-helix

NMR nuclear magnetic resonance spectroscopy

SAM shoot apical meristem

T-DNA transferred DNA

7

INTRODUCTION

Arabidopsis thaliana: the model plant

Arabidopsis thaliana, thale cress, belongs to the crucifer family (Brassicaceae). This

plant has a broad natural distribution throughout Europe, Asia and North America. In

the wild, Arabidopsis thaliana grows as a winter annual, that is, the seed germinates in

the autumn, pass the winter as a rosette and then flower early during the spring when

temperature and light conditions are favourable.

Historically, agriculturally important species such as maize, tomato, potato or

barley were the plants of choice for experimental studies. As the efforts of the research

community was spread out on several plant species, each considered as the ideal model

organism by the different researchers, the progress in the understanding of fundamental

plant processes was rather slow (Meinke et al., 1998).

Already in 1943, Friedrich Laibach, Professor at the University of Frankfurt am

Main, published an article describing the advantages of Arabidopsis thaliana as a

research tool. In his article, a number of favourable features of Arabidopsis were pointed

out, and since then, only a few additional advantages are acknowledged. The features of

this plant that make it suitable for experimental studies include several traits. One is the

short generation time of Arabidopsis; when grown under optimal conditions in the green

house it has a generation time of about six weeks. Another advantage is the small size of

the plant, which make it possible to grow a large number of plants in limited growth

facilities. The diameter of the rosette ranges from 2 to 10 cm in diameter depending on

growth conditions, the final height of the inflorescence stem reaches 20-40 and the

flowers are about 2 mm long. Furthermore, Arabidopsis flowers self-pollinate as the

buds open and each plant can produce hundreds of siliques, containing more than 5000

seeds. It is easy to perform genetic crosses of Arabidopsis and recover a fully fertile

progeny, and both induced and spontaneous mutations can be isolated with ease.

Although all these advantages of using Arabidopsis for experimental studies were

pointed out already in 1943, it would last another forty years before it really became

favoured as a model organism (Redei, 1992; Meinke et al., 1998).

8

Figure 1. Arabidopsis thaliana, thale cress, approximately four weeks old and

grown under long day conditions. (Photograph by Eva Söderman).

9

In 1983, Maarten Koorneef presented a detailed genetic map of Arabidopsis.

Furthermore, it was demonstrated that the genome of Arabidopsis is one of the smallest

genomes known among angiosperm species (Meyerowitz and Pruitt, 1985); a

characteristic that facilitates the isolation of genes through screening of genomic libraries,

which can be a time-consuming task when working with large genomes. Another

groundbreaking discovery was that it is possible to transform Arabidopsis with

Agrobacterium, making it possible to generate stable transgenic plants with ease. The

methods for Agrobacterium-mediated transformation of Arabidopsis have developed

from tissue culture techniques (Lloyd et al., 1986), through seed transformation

(Feldmann and Marks, 1987), to the whole-plant transformation methods (Bechtold et

al., 1993). With the latter method, it is possible to generate large number of

transformants with a reasonable effort.

Over the past 20 years, several thousand mutants of Arabidopsis, defective in

almost every aspect of plant growth and development, have been identified. The

development of efficient transformation methods for Arabidopsis has facilitated the

establishment of large mutant collections of T-DNA insertion lines, carrying randomly

inserted T-DNAs. Seed stocks and DNA-pools from these collections of lines are

available from public stock centers. The use of these lines for mutant screenings is

particularly advantageous as the genes that are disrupted in these lines are tagged by the

inserted T-DNA, which facilitate the isolation of the mutated gene.

Research using Arabidopsis as a model plant has provided insight into almost

every aspect of plant biology. For example, the understanding of phytohormone action

has improved drastically through genetic and molecular analysis of Arabidopsis. The

classical approach to understand the function of plant hormones has largely been based

on application of the hormone and recording the observed effect. However, this

approach has numerous limitations related to the uptake, transport and sequestration of

the exogenously applied hormone. Usually, an indirect, rather than a direct, link between

the hormone and its effect on a particular developmental process of the plant will be the

result. In contrast, the use of mutants, defective in either their ability to produce the

hormone, or in their ability to perceive and respond to the hormone, has been

particularly informative (Klee and Estelle, 1991). For example, mutant analysis of the

10

response to the gaseous plant hormone ethylene lead to the identification of the first

hormone receptor in plants (Chang et al., 1993). In addition, increasing knowledge about

the signal transduction and physiological responses to a range of different plant

hormones has emerged through the genetic analysis of hormone mutants in Arabidopsis

(reviewed in McCourt, 1999). The use of Arabidopsis for this purpose makes it

possible to isolate and characterize the mutated genes with a reasonable effort due to the

extensive amount of genetic information available for Arabidopsis.

The perhaps most important progress of Arabidopsis research is the sequencing

of the entire genome, which is performed as a collaborative effort from several

participants all over the world. This project was initiated in 1996 and estimated to be

completed in year 2004. However, this project has proceeded at a much higher speed

than expected; the completion of the genome is now scheduled to the fall, 2000. With

this information available, the field of Arabidopsis research will definitively change,

making it possible to analyze plants from the perspective of the whole genome.

Transcriptional regulation

The process of gene expression links the genetic information that resides in the DNA of

the nucleus with the activity of proteins that give the individual cell its characteristics.

The conversion of the genetic information into a protein product is to a large extent

dependent on the intermediate step, gene transcription. This process is the key-point

for the regulation of gene expression, controlling the cell type or timing-specific

expression of a certain gene, as well as expression in response to a certain signal

(Latchman, 1998). Transcription is regulated by the activity of transcription factors,

which interact with target genes and either activate or repress transcription of the gene.

In most cases, transcription factors are sequence-specific DNA-binding proteins that

bind to specific DNA target sequences and interact with the basal transcription

machinery.

11

Homeodomain proteins

Among transcription factors, several different types of DNA-binding protein motifs

have been identified and characterized. Homeodomain proteins, encoded by the

homeobox gene family, represent one class of transcription factors. Homeobox genes

have been shown to be important for the genetic control of development in range of

different eucaryotes, i.e. specification of the body plan, determination of cell fate and a

number of other developmental processes (for reviews, see e.g. Gehring, 1992; Lawrence

and Morata, 1994).

The first homeodomain protein to undergo a structural analysis was the

Drosophila homeodomain protein Antennapedia (AntP), whose three-dimensional

structure in solution was determined by nuclear magnetic resonance (NMR)

spectroscopy (Qian et al., 1989; Billeter et al., 1990). The analysis revealed that the

homeodomain of AntP is composed of three α-helices that are folded in a tight globular

structure. Helix 1 is separated from helix 2 by a loop whereas helix 2 and 3 are separated

by a turn. Sequence similarity between the homeodomain and the helix-turn-helix (HTH)

motif of prokaryotic regulatory proteins (Pabo and Sauer, 1984), in which the second

helix of the HTH motif makes specific contacts with DNA, suggested that

homeodomain proteins act as DNA-binding regulatory proteins. Subsequently, X-ray

crystallographic studies of homeodomain proteins complexed with DNA have revealed

how homeodomain proteins recognize DNA (reviewed in Gehring et al., 1994). Helix 3

are positioned in the major groove of the DNA and make sequence specific interactions

with the bases of the DNA. Amino acid residues of the N-terminal extension of the

homeodomain perform additional base-contacts in the minor groove of DNA, and amino

acid residues positioned in the loop between helix 1 and 2 and at the start of helix 2

make contacts with the DNA backbone. Structural studies show that various

homeodomain proteins from a range of different organisms interact with DNA in

essentially identical ways, demonstrating that the mode of DNA binding of

homeodomain proteins has been highly conserved during evolution.

12

Homeodomain proteins in plants: KNOTTED-1 proteins

The first homeodomain-encoding gene to be isolated from plants was KNOTTED-1 from

maize (Vollbrecht et al., 1991) corresponding to the Kn1 locus, which is defined by

several dominant gain-of-function mutations that disrupt leaf development, resulting in

the formation of knots on the leaf blade. In the Kn1 mutants, KNOTTED-1 is expressed

ectopically in vascular bundles within developing leaves, thus correlating with the

phenotypic alterations evident in the mutant (Smith et al., 1992). The wild-type

KNOTTED-1 gene is expressed in the apical meristem of vegetative and floral shoots but

expression is excluded from determinate organs such as leaves (Smith et al., 1992),

suggesting that this gene is important for shoot meristem function. Mutations in the

Arabidopsis KNOTTED-like gene SHOOTMERISTEMLESS (STM) result in seedlings

that are incapable of developing shoot apical meristems during embryogenesis (Barton

and Poethig, 1993; Long et al., 1996). These observations suggest that KNOTTED-like

genes have important roles in shoot apical meristem maintenance (Long et al., 1996).

PHD finger proteins

A different class of plant homeodomain proteins are defined by virtue of a second

conserved protein motif denoted PHD-finger. An example of a PHD-finger

homeodomain protein is HAT3.1 from Arabidopsis thaliana. This protein was isolated

based on its capability to interact with the light-induced cab-E promoter of tobacco

(Schindler et al., 1993). The PHD-finger class of homeodomain proteins also includes

the maize protein Zmhox1a, which binds to the 26 bp feedback control element of the of

the Shrunken gene promoter (Bellman and Werr, 1992), and PRHA/PRHP from

Arabidopsis and parsley, respectively, that both interact with a region of the pr2

promoter required for elicitor-mediated expression of this gene (Korfhage et al., 1994).

HDZip proteins

In an effort to isolate homeobox containing genes from Arabidopsis thaliana, a strategy

previously used to isolate homeobox genes from the nematode Caenorhabditis elegans

13

(Bürglin et al., 1989) was applied. Arabidopsis cDNA libraries were screened with

degenerate oligonucleotide probes, designed to match all possible codon combinations

corresponding to eight conserved amino acids in the helix 3 region of the homeodomain

(Ruberti et al., 1991; Mattsson et al., 1992; Schena and Davies, 1992). A number of

clones were recovered and showed extensive sequence similarity to known homeobox

genes and were designated Arabidopsis thaliana homeobox genes (ATHB). In addition,

these clones contained a second sequence element positioned 3´and closely linked to the

homeobox, encoding an α-helical stretch of aminoacids, with a periodic repetition of

leucine residues at every seventh position. This configuration of aminoacid residues

resembled that of the transcription factor C/EBP, in which the leucine residues are

arranged along one side of the helix and interact with the leucine residues projecting from

a second helix, forming a dimeric “leucine zipper” complex (Landschulz et al., 1988). As

the positions of putative DNA-binding units relative the leucine zipper domain are

similar in ATHB and bZIP proteins, respectively, it was proposed that proteins with a

contiguous homeodomain-leucine zipper architecture should be referred to as HDZip

proteins (Ruberti et al., 1991).

Additional members of the HDZip gene family have subsequently been isolated

from Arabidopsis (Carabelli et al., 1993; Söderman et al., 1994; Schena and Davis, 1994;

Baima et al., 1995; Lu et al., 1996; Di Cristina et al., 1996; Sessa et al., 1998; Lee and

Chun, 1998; Kubo et al., 1999; Zhong and Ye, 1999; Hanson (II); Johannesson et al.,

(IV). HDZip genes have also been identified from a range of other plant species; tomato

(Tornero et al., 1996; Meissner and Theres, 1995; Mayda et al., 1999, sunflower (Chan

and Gonzalez, 1994; Gonzalez et al., 1997), carrot (Kawahara et al., 1995; Mattsson et

al., 1995), soybean (Moon et al., 1996), Pimpinella brachycarpa (Moon et al., 1996),

resurrection plant (Frank et al., 1998) and ferns (Aso et al., 1999). HDZip genes have

been isolated from plants, but not from any other eukaryotes, suggesting that the

HDZip class of homeodomain proteins is specific to plants.

14

DNA-binding properties of HDZip proteins

Based on sequence criteria and supported by intron positions, HDZip proteins have

been grouped in four different families, HDZip I – IV (Sessa et al., 1994). Proteins that

belong to HDZip I and II are very similar to each other with respect to the architecture

of the HDZip domain. DNA-binding studies of ATHB1 (HDZip I) and 2 (HDZip II)

Figure 2. A hypothetical model of a HDZip protein binding to DNA, based on the

three-dimensional structures of the Drosophila engrailed homeodomain and the yeast

GCN4 leucine zipper motif, respectively (K. Johansson, H. Johannesson and E.

Söderaman, unpublished).

15

revealed that these two proteins bind as homodimers to pseudopalindromic binding

sites, CAAT(A/T)ATTG and CAAT(G/C)ATTG, respectively, consisting of two 5 bp

half-sites that overlap at the central position (Sessa et al., 1993). Furthermore, it was

obvious that the spacing between the homeodomain and the leucine zipper is critical for

the function of HDZip proteins as an insertion of two amino acids in this region of the

HDZip protein abolished its DNA-binding activity. Two HDZip II proteins from other

species other than Arabidopsis thaliana, Oshox1 from rice and CPHB-1 from the

resurrection plant Craterostigma plantagineum, bind DNA with specificities similar to

that of ATHB2 (Meijer et al., 1997; Frank et al., 1998).

The class III proteins ATHB8, 9, 14 have a different architecture of their HDZip

domains as compared with HDZip I and II proteins, with insertions of four amino acids

both between helix 2 and 3 of the homeodomain and between helix 3 and the leucine

zipper domain (Baima et al., 1995; Sessa et al., 1998). Thus, the spacing

between the homeodomain and leucine zipper domain is different in these proteins as

compared to that of HDZip I and II proteins, which suggests that class III proteins

might exhibit a different mode of DNA-binding. However, DNA binding studies of the

ATHB9 protein revealed that this protein interacts specifically with the DNA sequence

GTAAT(G/C)ATTAC, thus demonstrating that the HDZip domain of class III proteins

interacts with DNA in a similar fashion as that of class I and II protein.

ATHB10/GLABRA2, a class IV protein, is identical to class I and II proteins in

the architecture of the homeodomain and spacing between the homeodomain and leucine

zipper domain (Di Christina et al., 1996). However, the leucine zipper motif of

ATHB10/GLABRA2 is separated in two subdomains by a loop of 10 amino acid

residues. Replacement of the ATHB2 leucine zipper domain with that of

ATHB10/GLABRA2 resulted in a chimeric protein that was capable of interacting with

the ATHB2 recognition site. Hence, despite the different organization of the leucine

zipper domain of ATHB10/GLABRA2, this protein is active in dimerization and most

likely exhibits a similar mode of DNA-binding as class I and II proteins.

The DNA-binding activities in vivo have been analyzed for three HDZip

proteins: ATHB1 and 2 from Arabidopsis (Aoyama et al., 1995; Steindler et al., 1999)

and Oshox1 from rice (Meijer et al., 1997). ATHB1 was able to activate transcription

16

from a promoter containing the ATHB1 target sequence CAAT(A/T)ATTG upstream

of a reporter gene in tobacco cells. On the other hand, Oshox1 and ATHB2 were shown

to repress reporter gene activity in rice suspension cells and Arabidopsis leaves,

respectively. The transcriptional repression property of Oshox1 was lost upon deletion

of the N-terminal part of the protein, indicating that the repression function of Oshox1

resides in this part of the protein (Meijer et al., 1997). Thus, based on these

observations it is obvious that the HDZip family of transcription factors include both

activators and repressors of transcription.

In another study, it was shown that the N-terminal part of the protein, which

repress transcription in rice cells, functions as a transcriptional activator in yeast cells,

suggesting that Oshox1 might function as either a transcriptional activator or repressor

depending on the promoter context of the target gene (Meijer et al., 2000). To what

extent this dual function applies to other HDZip proteins awaits further testing.

Dimerization properties of HDZip proteins

Protein-protein interaction studies in vitro have shown the ATHB1 and ATHB2

proteins to exclusively form homodimers (Sessa et al., 1993). Similarly, studies on the

interaction between the sunflower HDZip proteins Hahb-1 and -10, which belong to the

class I and II, respectively, demonstrated that these proteins were only capable of

homodimerization (Gonzalez et al., 1997). Heterodimerization has been demonstrated

between two HDZip II proteins from different species; Oshox1 and ATHB2 (Meijer et

al., 1997) and between the two HDZip II proteins CPHB-1 and -2 from the resurrection

plant Craterostigma plantagineum (Frank et al., 1998). Together, these data suggest that

heterodimerisation is possible between members of the same class, but not between

proteins of different classes. The mechanism behind this apparent specificity in

dimerization is not clear. However, the organization of amino acid residues in the N-

terminal part of the leucine zipper is different between HDZip I and II proteins and is

probably of great importance (Gonzalez et al., 1997).

17

Function of HDZip proteins

The function of three Arabidopsis HDZip genes has been resolved through mutant

analyses. The glabra2 mutation results in defective trichome and root hair development

(Rerie et al., 1994; Masucci et al., 1996). The GLABRA2 gene was later shown to be

identical to the HDZip IV gene ATHB10 ( Di Christina et al., 1996). Mutations in the

ANL2 gene (HDZip IV) result in abnormal anthocyanin accumulation in subepidermal

tissues of rosette leaves and disturbed organization of primary root cells (Kubo et al.,

1999). Disruption of the IFL1 gene (HDZip III) abolishes the formation of normal

interfascicular fibers in Arabidopsis inflorescence stems (Zhong et al., 1999).

Functional information on HDZip class II genes is based on results from

expression analyses and analyses of transgenic plants with altered expression levels of

the genes. Expression of the ATHB2 (HDZip II) gene is induced by changes in the ratio

of red to far-red light (Carabelli et al., 1996) and the phenotypic analysis of transgenic

Arabidopsis plants with altered levels of ATHB2 suggests this gene to be a mediator of

shade avoidance responses (Steindler et al., 1999). Two dehydration-stress inducible

HDZip II genes, CPHB-1 and -2, were isolated from the resurrection plant,

Craterostigma plantagineum, using a differential display technique. Expression of both

genes are induced at early stages of dehydration suggesting that these genes may be

involved in the regulation of gene expression during dehydration (Frank et al., 1998).

Overexpression of the rice Oshox1 protein (HDZip II) protein in transgenic Arabidopsis

resulted in small and lancet-shaped leaves (Meijer et al., 1997). Oshox1 is predominately

expressed in leaves and together, these data are suggestive of a role for this gene in the

control of leaf morphogenesis.

Functional information on HDZip class I proteins is limited. ATHB6, -7 and -12

are induced by exogenous ABA, water deficit and osmotic stress. The induction of

ATHB6 and -7 by drought was strictly dependent on ABA as no increase in ATHB6 and

-7 transcripts was detected in the drought-treated seedlings of the ABA-deficient mutant

aba-3. Furthermore, the induction by ABA of ATHB7 was impaired exclusively in the

ABA-response mutant abi1 whereas the induction of ATHB6 was impaired in both abi1

and abi2. These data suggest that ATHB6, 7 and 12 probably act in an ABA-dependent

signal transduction pathway, mediating the growth response to drought in the plant

18

(Söderman et al., 1996; Lee and Chun, 1998; Söderman et al., 1999). Transgenic plants

with enhanced levels of ATHB1 are defective in the palisade parenchyma as cells of this

tissue in the transgenic plants are replaced by other cells, similar to spongy mesophyll

cells. In addition, the transgenic plants exhibited deetiolated phenotypes in the dark as

well as abnormal expansion of cotyledons. Together, these observations suggest that

ATHB1 is involved in the activation of target genes that are closely linked to plant

development (Aoyama et al., 1995). Antisense suppression of the tomato HDZip I gene

H52 in transgenic tomato plants results in necrosis and abscission of leaves (Mayda et

al., 1999). The accumulation of ethylene and salicylic acid is increased and several

defense-related genes are constitutively expressed in the transgenic plants. H52 gene

expression is induced upon infection with virulent pathogens. Thus, these data are

suggestive of a role of H52 as a transcriptional regulator of genes involved in

programmed cell death (PCD), acting to protect the cells against PCD (Mayda et al.,

1999).

19

RESULTS AND DISCUSSION

Identification of new HDZip genes (II, IV)

By searching available databases, we were able to identify 26 Arabidopsis sequences

that show similarity to HDZip class I and II genes. Of these, 17 corresponded to

previously characterized HDZip genes (ATHB1, 2, Ruberti et al., 1991; ATHB3,

Mattson et al., 1992; ATHB5, 6, 7, Söderman et al., 1994; ATHB12, Lee and Chun,

1998; ATHB13, Hanson et al., paper III; ATHB16, Y. Wang, in preparation; ATHB4,

Carabelli et al., 1993; HAT22, Schena and Davies, 1992; HAT1, 2, 3, 9, 14, Schena and

Davies, 1994, ATHB17, Ruberti et al., unpublished), whereas nine represented novel

sequences encoding putative HDZip proteins. These nine genes were designated

ATHB20, 21, 22, 23, 40, 51, 52, 53 and 54, respectively.

Biochemical characterization of HD-DNA complexes have shown a majority of

HD proteins to recognize TAAT-containing DNA-sequences (reviewed in Treisman et

al., 1992). The inherent similarities in DNA-binding specificity of all these HD proteins

implicate a structural conservation at the protein-DNA interface, which has been

confirmed by structural studies of several HD-DNA complexes (reviewed in e.g.

Gehring et al., 1994). A homeodomain consensus sequence has been defined based on a

compilation of 346 homeodomain sequences (Bürglin, 1994). The amino acids conserved

in this consensus sequence can be assigned to three different functions: contributing to

the hydrophobic core, to sugar-phosphate backbone recognition or specific interactions

with the bases in the TAAT core site. The HDs of all HDZip proteins, including

ATHB20, 21, 22, 23, 40, 51, 52, 53 and 54, contain the five invariant amino acid (L16,

W48, F49, N51, R53) and six out of seven of the highly conserved residues (F20, L26, L40,

V45, I47, L57) of this homeodomain consensus sequence. In addition to the conserved

homeodomain, these protein, including ATHB20, 21, 22, 23, 40, 51, 52, 53 and 54,

contain leucine zipper motifs in identical positions C-terminal to the homeodomain.

An initial alignment of the HDZip region of the deduced aminoacid sequences

revealed that the ATHB22 sequence did not align to the other HDZip sequences in the

region upstream of helix 2 (Figure 3). However, if 8 aminoacids corresponding to the

region between helix 1 and 2 were looped out, a perfect alignment of ATHB22 along the

20

21

entire HDZip domain was possible. A similar atypical architecture of the homeodomain

is evident for the yeast homeodomain MATα2, which has an insertion of three

aminoacids in the turn between helices 1 and 2 (Hall and Johnson, 1987). The structure

of MATα2 (Wolberger et al., 1991) is very similar to the Drosophila homeodomain

proteins engrailed (en) and AntP structures, indicating that the extra aminoacids in the

MATα2 homeodomain do not influence the overall structure of this domain. Thus,

despite the atypical architecture of the ATHB22 homeodomain, it is likely that this

protein adopts a similar structure and exhibits a similar DNA-binding property as those

of other HDZip proteins.

Based on sequence criteria, HDZip proteins have been grouped in four distinct

subfamilies, HDZip I – IV (Sessa et al., 1994) and the members of each subfamily are

also related with respect to intron positions. A number of positions separate HDZip I

and II proteins. More specifically, position 46 of helix 3 is invariant within each of the

classes but distinct between HDZip I and II. The same is true for position 58 in the

region between the homeodomain and the leucine zipper domain. The alignment shows

that, according to these criteria, the nine newly identified proteins belong to HDZip I,

since the amino acid residues in positions 46 and 58 are identical to those of all the

previously identified HDZip I proteins.

The evolutionary relationship between the entire set of HDZip I and II

aminoacid sequences was analyzed by parsimony, using PAUP (Swofford, 2000). The

phylogenetic analysis resulted in four equally parsimonious trees. Within the HDZip

domain, introns can be found in five different positions. The intron positions are

conserved within the different subclasses defined by the tree and thus support the major

branching pattern of the tree. The tree revealed two distinct classes, corresponding to

the HDZip I and II classes previously identified by Sessa et al., (1994), respectively.

All of the nine newly identified sequences were included in HDZip class I, confirming

that ATHB20, 21, 22, 23, 40, 51, 52, 53 and 54 belong to HDZip I.

The chromosomal locations of each of the HDZip genes were determined based

on the positions of corresponding BAC-clones on the physical map of Arabidopsis

thaliana. As shown in Figure 4, HDZip I and II genes are evenly distributed over the

entire length of the five chromosomes of Arabidopsis.

22

Figure 4. Locations of HDZip I and II genes on the five chromosomes of Arabidopsis

thaliana. The position of each gene is according to the physical map of Arabidopsis and

their locations are indicated relative a subset of mi-RFLP markers (Liu et al., 1996)

23

Identification of HDZip target sequences in vitro (I)

HDZip proteins contain leucine zipper domain located immediately C-terminal to the

homeodomain. As the leucine zipper motifs in other classes of transcription factors

mediate dimerization it was suggested that HDZip proteins are active as protein dimers

in the binding of DNA, such that the leucine zipper juxtaposes a pair of DNA

interacting units onto the DNA (Ruberti et al., 1991). This notion was confirmed when

analyzing the DNA-binding properties of ATHB1 and ATHB2, which interacted with

the distinct pseudopalindromic sequences CAAT(A/T)ATTG and CAAT(G/C)ATTG,

respectively, as homodimers (Sessa et al., 1993) . Thus, ATHB1 and ATHB2 interact

with nearly identical DNA sequences and the principal difference is their specificity for

the central position in the recognition site. Considering the high degree of sequence

similarity within the homeodomain of the HDZip proteins, it is possible that these

proteins interact with very similar DNA binding sites. Moreover, it is likely that

proteins that belong to the same subclass exhibit identical binding site preferences, such

that all HDZip I and II proteins have the same specificity as ATHB1 and ATHB2,

respectively. To test this hypothesis, the binding specificity for a number HDZip class

I proteins was analyzed, starting with ATHB5. A PCR assisted binding-site selection

procedure was used to select for high-affinity binding sites for ATHB5 from a pool of

DNA-fragments containing a 12 bp random core sequence. Strikingly, the majority of

the clones contained the pseudopalindromic sequences CAAT(A/T)ATTG or

CAAT(G/C)ATTG, confirming that the HDZip I protein ATHB5 binds DNA in a

similar fashion as ATHB1. In contrast to ATHB1, ATHB5 did not discriminate

between binding sites in which the identity of the central positions was different as the

frequency of recovered clones containing either an A/T or a G/C pair in the central

position was essentially the same. In electrophoretic mobility shift assays, ATHB6 and

16 interacted with similar affinity to either A/T or G/C containing sequences. This result

demonstrates that ATHB6 and 16, similar to ATHB5, are independent on the identity

of the central position in the binding site. In contrast, ATHB1 showed a distinct and

ATHB3 and 13 exhibited a slight preference for an A/T pair in the central position

specificity. Thus, it is evident that HDZip proteins interact with very similar binding

sites, but central position specificity appear to vary among HDZip class I proteins.

24

What is the determinant of central position specificity in the HDZip proteins? A

DNA binding model of HDZip proteins presented by Sessa et al., (1993) predicts that

HDZip and bZip proteins interact similarly with DNA and that Arg55 of the HDZip

domain is functionally equivalent to Arg243 of the yeast bZip protein GCN4. When

complexed with the GCN4 recognition site, TGA(C/G)TCA, Arg243 of one GCN4

monomer donates hydrogen bonds to the guanidine of the central pair, whereas Arg243

of the other monomer makes phosphate backbone interactions (Ellenberger et al., 1992).

The importance of Arg55 for central position specificity of HDZip proteins was tested

experimentally by a mutational analysis of the ATHB2 HDZip domain (Sessa et al.,

1997). The results indicated that Arg55 of ATHB2 is important for central position

specificity, but residues outside helix 3 also contribute to binding-specificity. Thus,

specificity for the central position may be attributable to the conformation of the Arg55

side chains in the HDZip domain. This finding is consistent with the fact that Arg55 is

invariant among all HDZip proteins and therefore, this position alone can thus not

explain the apparent diversity of central position preference for HDZip proteins.

ATHB7 and 12 have alternative binding site specificities (I)

In contrast to ATHB1, 3, 5, 6, 13 and 16, neither ATHB7 nor ATHB12 interacted with

the HDZip consensus site CAATNATTG. This was a surprising result considering the

fact that all HDZip I and II proteins, including ATHB7 and 12, are highly similar in

sequence within the homeodomain and it is rather unlikely that these proteins would

differ dramatically in their DNA binding specificities. One possible explanation for the

result is that the ability of the ATHB7 and 12 proteins to bind DNA is dependent on

correct post-translational modifications, e.g. protein phosphorylation. This possibility

is consistent with the presence of a potential phosphorylation site at a position in the

homeodomains of ATHB7 and 12, which is absent in homeodomains of other class 1

HDZip proteins. Furthermore, the transcription of both genes are induced by ABA,

osmotic stress and drought conditions (Söderman et al., 1996; Lee and Chun, 1998), and

in the case of ATHB7 by a signaling mechanism that involves protein phosphorylation/

25

dephosphorylation (Söderman et al., 1996), suggesting that this signaling pathway might

also act on ATHB7 directly on the protein level by a similar mechanism.

Another possibility is that ATHB7 and 12 interact with different binding-sites

than those of other HDZip I proteins. In support of this hypothesis, ATHB12 has

recently been demonstrated to specifically interact with the palindromic sequence

TCAATTAATTGA (Chun and Lee, 1998, US Patent No 5,981,729). A closer

inspection of the ATHB12 recognition site reveals that it consists of two non-

overlapping TCAATT half-sites, as opposed to the HDZip consensus-binding site

CAATNATTG, which consists of two CAATN half-sites overlapping in the central

position. Thus, the principal difference between the two sites is the spacing between the

half-sites, whereas the half-sites as such are essentially identical. This finding suggests

that the interaction of ATHB12 at the protein-DNA interface is the same as for other

HDZip proteins, but that ATHB12 adopts a slightly different structure that affects the

spacing between half-sites. Furthermore, considering the high degree of sequence

similarity between ATHB7 and 12, it is likely that ATHB7 exhibits an identical DNA-

binding specificity as that of ATHB12.

The mechanism by which half-site spacing in HDZip proteins is determined is

not clear. In bZip proteins, it has been shown that the short surface spanning the fork

region, which connect the leucine zipper and the DNA-binding basic region, is the major

determinant of half-site spacing (Kim and Struhl, 1995). However, the region of HDZip

proteins that corresponds to the fork region in bZip proteins is conserved among all

HDZip proteins, including ATHB7 and 12, suggesting that this mechanism of half-site

spacing determination is not applicable to HDZip proteins.

Another possibility is that HDZip proteins, in analogy with e.g. the Drosophila

homeodomain proteins AntP and en, perform DNA contacts in the minor groove of

DNA through arginine residues at positions 3 and 5 in their N-terminal arms of the

homeodomain (reviewed in Gehring et al., 1994). A closer inspection of the region of the

homeodomain preceding helix 1, all HDZip class proteins, except ATHB7 and 12, have

a basic stretch of amino acids, which shows resemblance to the N-terminal extention of

classical homeodomain proteins. Although not tested experimentally, it is possible that

the N-terminal region of HDZip proteins is functionally equivalent to that of AntP and

26

en and determines DNA-binding specificity through minor groove interactions. If that

were the case, the alternative spacing preference of ATHB7 and 12 could be a

consequence of the alternative architecture of the N-terminal region.

In summary, HDZip proteins are similar in sequence within their HDZip

domain, which suggest that they should interact with very similar DNA sequences.

Nevertheless, the diversity in central position specificity among HDZip proteins,

manifested either as preference for a certain pair in this position or as preference for an

alternative spacing between half-sites, appear to be larger than initially expected. Thus,

the functional differences between different HDZip proteins may in part be due to this

diversity in their DNA-binding.

Heterodimer formation between HDZip class I proteins (I).

The fact that some of the HDZip proteins have very similar DNA-binding specificities

and yet appear to have distinct biological functions raise the hypothesis that the

targeting of HDZip proteins in vivo to the correct promoters is dependent on

interactions with other proteins. A special case of such protein-protein interaction is the

formation of heterodimers between different members of HDZip proteins.

The interaction between a recombinant bacterial ATHB5 protein and in vitro

translated 35S-labeled HDZip proteins was analyzed and the result showed that ATHB5

interacted with ATHB6, 7, 12 and 16 with different apparent affinity, but not with

ATHB1, suggesting that HDZip proteins show selectivity in dimerformation, specified

by the dimerization domain.

In bZip proteins, such as the protein products of the proto-oncogenes c-fos and

c-jun, the leucine zipper domain itself is capable of dictating specificity (O´Shea et al.,

1992). Specificity depends on electrostatic interactions between amino acids of e and g

positions of the heptad repeats (abcdefg)n that form the hydrophobic interface in the

two monomers, which is consistent with the known crystal structures of GCN4

(Ellenberger et al., 1992) and a Fos/Jun heterodimer (Glover and Harrison, 1995). An

interhelical salt bridge hypothesis (Vinson et al., 1993), based on functional analyses of

leucine zippers suggests that the identity of amino acid in the g and e position of each

27

unit determine which proteins that have the potential to dimerize. Applied to the class 1

HDZip proteins, this hypothesis predicts the proteins to have the capacity to form

both homo- and heterodimers. The in vitro data on ATHB5 indicate that the protein in

fact does show selectivity in dimer formation, suggesting that HDZip proteins rely on a

mechanism different from that described for bZip proteins to dictate dimerization

specificity.

The selective heterodimer formation between HDZip class I proteins might

provide the plant with transcription factors that, even when interacting with the same

target sequence as the original homodimer, have distinct activation characteristics. Thus,

it is possible that HDZip proteins constitute a functional complex of homo- and

heterodimeric transcription factors in the plant, providing a mechanism by which the

plant can increase the level of complexity in transcriptional regulation on the protein

level.

Expression analysis of ATHB5 (II)

A first step to understand the function of HDZip proteins is to determine the tissue and

timing specificity of HDZip gene expression. In the case of ATHB5, this gene was

previously shown to be expressed at moderate levels in adult leaf, stem and root tissues

(Söderman et al., 1994); however, the expression of ATHB5 at other developmental

stages was not analyzed. Analysis of the expression of ATHB5 during early

development showed that ATHB5 gene transcription was turned on early after the onset

of germination and then gradually declined with increasing age of the seedlings.

Additional experiments revealed that ATHB5 mRNA was detectable as early as 12 h

post-imbibition, but not in the dry seed (data not shown). Histochemical staining of

transgenic Arabidopsis seedlings harboring an ATHB5::GUS fusion revealed strong GUS

activity specifically in the hypocotyl of germinating seedlings, concentrated around the

transition zone, which marks the boundary between the hypocotyl and the root.

Staining was retained at the transition zone as the seedlings grew older, but was also

visible with low intensity in the petioles of cotyledons and developing true leaves. Weak

GUS staining associated to vascular tissue was found in root, stem and cauline leaf

28

tissues and no visible staining was found in flower organs. Thus, the expression analysis

of ATHB5 showed that the activity of this gene appears to be under strict

developmental control and activated by the initiation of germination.

The staining intensity at the transition zone of the ATHB5::GUS transgene

decreased markedly when seedlings were treated with 10 µM ABA for 12 hrs. This

effect was specific for ABA as the staining intensity was unaffected by the treatment

with auxin, gibberellic acid, cytokinin or the ethylene precursor ACC. These data

indicate that ATHB5 gene expression is dependent on ABA signaling. This notion was

supported by results from experiments assessing whether ATHB5 expression during

postgerminative growth was dependent on the activity of known ABA response loci.

The expression level of ATHB5 was markedly lower in abi3-1 and abi5-1 mutant

background as compared to the wild type, demonstrating that ATHB5 represents a target

gene for transcriptional regulation by ABI3 and ABI5 during postgerminative growth,

either directly or indirectly.

In summary, the expression analysis allows the distinction of cell types and

developmental stages where the gene primarily exerts its function. In the case of ATHB5,

we can conclude that this protein is functional primarily in cells of the hypocotyl during

seed germination. Moreover, the fact that the ATHB5 expression level is affected by

ABA and lower in the ABA response mutants abi3 and abi5 indicate that the function

of ATHB5 is related to ABA signaling.

A reverse genetics approach to understand the function of HDZip proteins

Approaches such as gene expression analysis to understand gene function are only

correlative and do not necessarily provide a direct link between a gene and its function.

In contrast, by the analysis of null mutations in the genes of interest it is possible to

establish their function simply by assessing the phenotypic effects resulting from the

mutation. This approach has had an enormous impact on the understanding of

developmental processes in a range of different organisms (e.g. Nüsslein-Volhard, 1994;

Burns et al., 1994; Spradling et al., 1995; Brandon et al., 1995).

29

Mutations created by insertions of either T-DNA or transposable elements are

particularly advantageous as they provide direct means by which the mutated gene can

be isolated. Individual lines bearing insertions in the gene of interest are easily identified

by amplification of flanking regions of the inserted element using polymerase chain

reaction (PCR) with a combination of gene-specific primers and primers complementary

to border sequences of the inserted element. This method was originally developed for

Drosophila where PCR was used to screen for P-element insertions (Ballinger and

Benzer, 1989; Kaiser and Goodwin, 1990) and has also been utilized in other organisms

such as Caenorhabditis elegans (Zwaal et al., 1993). In Arabidopsis, this approach has

been used to isolate plants carrying insertions in a range of different genes (e.g.

McKinney et al., 1995; Krysan et al., 1996; Winkler et al., 1998). Considering the

relatively small genome size of Arabidopsis, it has been estimated that it is sufficient to

screen around 100 000 lines in order to find a plant carrying an insert in any gene of

interest (Azpiroz-Leehan and Feldmann, 1997).

A limitation to screens for loss-of-function mutations is that the recovered

mutants often lack discernable phenotypes, due to genetic redundancy. Redundancy is

evident among members of the ethylene receptor gene family (Hua and Meyerowitz,

1998) as well as among homeotic genes that control flower development in Arabidopsis

thaliana (reviewed in Martienssen and Irish, 1999). An alternative strategy to reveal

gene function is to assess the phenotypic effect of enhanced or reduced levels of gene

activity by expressing the gene in either sense- or antisense orientation from the strong

constitutive promoter of the cauliflower mosaic virus (CaMV) 35S gene. The anti-sense

approach will suffer from the same problem as that of insertion lines, namely gene

redundancy, unless the expression of the anti-sense construct has the capacity to

inactivate several genes simultaneously that are related in sequence and act in a

redundant fashion. In contrast, overexpression of the gene can provide functional

information of a gene, which will be manifested through a visible phenotype. One

complication of high-level expression from the 35S-promoter is that it results in gene

activity in tissues or developmental stages where the gene is not normally expressed and

hence, resulting in phenotypic deviations from wild type that are not related to the

30

function of the gene. However, a careful expression analysis of the gene should facilitate

the distinction of phenotypes that reflect the actual function of the gene.

Overexpression of ATHB5 in transgenic Arabidopsis (III)

Phenotypic analysis of ATHB5 overexpressor lines (35S::ATHB5) revealed that these

plants did not differ from wild type in their overall growth and development. However,

the 35S::ATHB5 plants differed from wild type in their seedling morphology.

Hypocotyls of young seedlings were shorter and radially expanded at the base. In

addition, the primary roots as well as petioles of cotyledons of 35S::ATHB5 seedlings

were shorter than those of the wild type. Thus, the phenotypic deviations from wild

type seen in the 35S::ATHB5 plants are consistent with the results from the expression

analysis of ATHB5, which suggest that it is likely that the ATHB5 gene exerts its

function primarily in hypocotyls of germinating seedlings and not in the adult plant. The

effect of the plant hormone ABA on the intensity of GUS staining in the ATHB5

promoter GUS-fusion plants, as well as the downregulation of ATHB5 in the ABA

response mutants abi3 and abi5, suggested that overexpression of ATHB5 in transgenic

plants might affect the response of the seedlings to growth inhibition by ABA. This

notion was confirmed by experiments assessing the sensitivity of wild type and

35S::ATHB5 plants to the inhibitory effects of exogenously applied ABA on seed

germination and seedling root growth. The results from these assays showed that the

35S::ATHB5 line was more sensitive to ABA at all concentrations tested as compared

with the wild type. The ABA concentration needed to achieve 50% inhibition of either

seed germination or root growth was two-fold lower for the 35S::ATHB5 line than for

the wild type. These results suggest that overexpression of ATHB5 interferes with ABA

signaling, resulting in seedlings hypersensitive to ABA. The finding that the

35S::ATHB5 seedlings were hypersensitive to ABA was consistent with the results

from experiments assessing ABA regulated gene expression, demonstrating that the

ABA responsive gene RAB18 was hyperinduced by ABA in 35S::ATHB5 seedlings as

compared to wild-type.

31

During seed development the level of endogenous ABA peaks during mid to late

embryogenesis before returning to low levels in the dry seed (Rock and Quatrano, 1995).

ABA has been demonstrated to be important for the establishment of seed dormancy

since Arabidopsis mutants either defective in ABA biosynthesis or in the ability to

sense ABA fail to become dormant (Koorneef et al., 1982; 1984). Furthermore,

mutations that enhance the sensitivity to ABA confer hyperdormancy (Cutler et al.,

1996; Gosti et al., 1999). Consequently, seeds from 35S::ATHB5 plants that are

hypersensitive to exogenous ABA should therefore also be hyperdormant. However,

overexpression of ATHB5 does not affect seed dormancy, indicating that ATHB5 does

not regulate ABA sensitivity during seed development and consequently does not take

part in the regulation of seed dormancy. This hypothesis is consistent with the fact that

ATHB5 is not expressed in wild-type plants during seed maturation when dormancy is

established.

In summary, the functional analysis of the HDZip gene ATHB5 provides

evidence for that this gene is active in the regulation of germination and postgerminative

growth, possibly as an ABA dependent negative regulator of cell expansion. Clearly,

germination and postgerminative growth is highly dependent on ABA signaling and it

has been suggested that the inhibitory effect of ABA on these processes is related to the

inhibition of storage reserve mobilization, since this hormone alters the pools of

available nitrogen and energy as well as prevents the degradation of seed storage proteins

(Garciarrubio et al., 1997). It is possible that the mechanism by which ATHB5 regulate

germination and postgerminative growth is through restriction of energy and metabolites

as a component of an ABA response pathway in Arabidopsis thaliana.

High-level expression of ATHB13 in transgenic Arabidopsis (II)

Increased expression levels of ATHB13 in transgenic plants (35S::ATHB13) did not

result in any major phenotypic deviations in overall growth and development as

compared to the wild type. However, when grown in vitro on standard medium, the

35S::ATHB13 seedlings differed from wild type by decreased cotyledon width. It was

evident that this effect was mediated by sucrose as the phenotypic deviation

32

disappeared when metabolizable sugars were omitted from the growth medium. The

effect of sucrose on the cotyledon width of the 35S::ATHB13 seedlings was dose-

dependent, since cotyledon width decreased with increasing sucrose concentrations.

Measurements of the size of individual epidermal cells demonstrated that the sucrose-

dependent decrease in cotyledon width of 35S::ATHB13 seedlings was a consequence of

inhibited epidermal cell expansion. The effect of ATHB13 overexpression on cotyledon

cell size and shape is most likely dependent on sucrose signaling, since neither non-

metabolizable sugars nor genetic crossings with transgenic plants with altered hexokinase

activities affected the morphology of 35S::ATHB13 seedlings. This hypothesis is

consistent with results from experiments analyzing sugar regulated gene expression,

which demonstrated that the expression of a subset of known sugar-responsive genes

were induced by sucrose to a higher level in 35S::ATHB13 seedlings than in the wild-

type. Taken together, these data show that ATHB13 affect both morphology and gene

regulation in response to sucrose, thus representing the first transcription factor shown

to directly take part in a sucrose signaling pathway.

Interpretation of phenotypes resulting from constitutive expression (II, III)

As stated above, high-level expression from the 35S-promoter can result in ectopic gene

activity, i.e. the gene is active in tissues or stages where it is not normally active. In the

case of the 35S::ATHB5 plants however, the expression analysis show this gene to be

active in the hypocotyl early after the onset of germination and dependent on the ABA

response genes ABI3 and ABI5, consistent with the ABA hypersensitivity phenotype

during seed germination conferred by the gene when expressed at high levels. In addition,

the fact that the phenotypic deviations seen in the ATHB5 overexpressors are specific

for the germination stage suggests that the target genes, which ATHB5 activates, are

only accessible during this developmental stage. Thus, it is possible that ATHB5 needs

to interact with other factors that are active during postgerminative growth in order to be

functional. In contrast, the ATHB13 gene is not expressed in cotyledons where it exerts

its effect when overexpressed (J. Hanson, in preparation). Thus, the effect of ATHB13

on cotyledon cell size and shape is most likely a consequence of ectopic expression and

33

does not directly reflect the function of the gene in the wild-type plant. This hypothesis

is supported by the observation that ATHB23, a close relative to ATHB13, is expressed

in cotyledons and has a similar effect as ATHB13 on cotyledon cell shape and size when

overexpressed (J. Hanson, in preparation).

Isolation of T-DNA insertion lines in HDZip class I genes (II, III, IV)

In order to obtain loss- or reduction-of-function plants of HDZip genes, the anti-sense

approach has been used for a number of HDZip genes. However, this method has

proved to be inefficient for a majority of the genes tested. Instead, PCR based screens

for insertion mutants in HDZip I genes were performed. A population of 6000 T-DNA

insertion lines with an average of 1,5 insertion events per line has been deposited at the

Arabidopsis Biological Resource Center (ABRC) as DNA "superpools" of 1000 lines

each. A primary screen of this population identified insertions in five different HDZip

genes, ATHB5, 6, 7, 40 and 51. The insertions in ATHB5, 7, 40 and 51 most likely result

in null mutations (Azpiroz-Leehan and Feldmann, 1997) as these mutation are either

located within intron sequences or, in the case of the ATHB7 insertion, in the

5´untranslated leader sequence. In contrast, a T-DNA inserted upstream of the

transcription start site of the ATHB6 gene might affect gene expression of ATHB6,

resulting in a "knock-down" mutation (Krysan et al., 1999) of the ATHB6 gene.

In the case of ATHB13, a putative null mutant line was isolated using the same

strategy as described above but from a collection of lines harboring En-1 transposons

(Baumann et al., 1998). This insertion was located in the 5´ splice site of the first intron

of ATHB13 and most likely results in a null-mutation of the ATHB13 gene.

The mRNA and protein levels of the gene have to be assessed in order to confirm

that the insertion has created a null-mutation. This was done for the ATHB5 insertion

line and the result showed that there was no detectable ATHB5 mRNA or proteins in

extracts of this line, confirming that this insertion in the first intron of the ATHB5 gene

results in a null-mutation. To what extent the insertions in ATHB6, 7, 40 and 51 disrupt

gene transcription and translation awaits further testing.

34

When analyzing the phenotype of the insertion lines it became evident that none

of these lines exhibited any visible phenotypic deviations from wild type, neither during

seedling growth nor as adult plants. The insertion lines of ATHB5 and 13 were subjected

to more thorough phenotypic analyses with respect to ABA and sucrose sensitivity,

respectively. In neither case did the loss-of-function mutations confer any phenotypic

deviations as compared to the wild type. One explanation to the lack of distinguishable

phenotypes in the insertion lines might be that HDZip genes are only functional under

specific physiological conditions, i.e. unless the mutant plant is subjected to conditions

in which the gene is essential, no phenotypic deviations from wild-type will be visible.

Another explanation is gene redundancy, i.e. the loss of gene activity in these insertion

lines might be compensated by other HDZip genes. Considering the analysis of DNA-

binding preferences of HDZip proteins it is likely that some of these proteins interact

with identical binding sites and activate the same target genes in vivo.

In summary, the use of a PCR-based systematic approach for mutant isolation in

Arabidopsis HDZip I genes proved to be successful; from a population of 6000

insertion lines, T-DNA insertions in five different genes were identified. The fact that

these lines lack distinguishable phenotypes emphasizes the importance of establishing

careful screens for growth conditions in which the activity of the target genes are

required. Alternatively, for those genes that act redundantly, genetic crosses between

plants that bear mutations in these genes have to be performed

35

SUMMARY

HDZip proteins constitute a large family of transcription factors in plants and with the

close completion of the Arabidopsis genome sequencing, we estimate that the total

number of HDZip I and II genes will not exceed 26. HDZip proteins interact

specifically with pseudopalindromic binding sites, in which each half-site is occupied by

one HDZip monomer. The difference in binding specificity between HDZip proteins

resides primarily in their preference for the central position of the binding site, whereas

the interaction with each half-site appears to be conserved. The finding that HDZip I

proteins are capable of forming both homo- and heterodimers in vitro suggests that

HDZip proteins might be involved in a complex network of interacting transcription

factors in the plant.

The HDZip I protein ATHB5 was shown to be active in the regulation of seed

germination and postgerminative growth, possibly as an ABA-dependent regulator of

cell expansion. ATHB13 affects leaf morphology as well as gene regulation in response

to sucrose, suggesting that ATHB13 takes part in a sucrose signaling pathway. The

functional analysis of these two proteins was primarily based on the interpretation of

overexpression phenotypes. Establishment of HDZip I gene function based on

phenotypic analyses of mutant plants identified by reverse genetics screens from

collections of T-DNA insertions and transposable elements has as yet not been

successful. The reason for this is not clear; however, it is possible that HDZip genes are

redundant or that the conditions in which HDZip genes are required are not known.

36

ACKNOWLEDGEMENTS

I wish express to express my sincere thanks to all those who have contributed to this

thesis:

My supervisor, Professor Peter Engström, for accepting me as PhD-student and

teaching me scientific thinking and writing.

Johannes, for interesting discussions about the deeper secrets of HDZip science;

invaluable help with computer-problems, and for giving me the opportunity to eat fast

food every now and then.

Thanks are due to Eva Söderman for being the perfect room-mate.

Eva Söderman, Eva Sundberg and Johannes Hanson, for critical reviewing parts of this

thesis.

The members of the Prof lab, Yan, Mats and Jens, for providing interesting discussions

and good friendship during labwork; Mats and Jens for making the Prof lab the most

funky lab in the building. Thanks to Peter, again, for letting Prof lab stay funky!

I wish to thank all people that have contributed to this thesis by technical assistance:

Marie Lindersson, Marie Englund, Eva Burén, Kajsa Pöntinen, and Agneta Ottoson.

Birgitta, thank you for helping me with all imaginable administrational problems.

Thank you Ingela for nostalgic conversations about good old times! Thank you Anneli,

Sandra, Katarina, and the rest of cellskapet for making fysbot a nice place to work in!

Last, but not least, I would like to thank:

Hanna for love and support, but also for invaluable comments on all manuscripts of this

thesis.

Svante, for being the sunshine of my life!

My parents and Hannas parents for continuos support and for taking care of Svante

during the work with this thesis.

37

REFERENCES

Aoyama, T., Dong, C.H., Wu, Y., Carabelli, M., Sessa, G., Ruberti, I., Morelli, G. andChua, N.H. 1995. Ectopic expression of the Arabidopsis transcriptional activator Athb-1 alters leaf cell fate in tobacco. Plant Cell 7: 1773-1785.

Aso, K., Kato, M., Banks, J.A. and Hasebe, M. 1999. Characterization ofhomeodomain-leucine zipper genes in the fern Ceratopteris richardii and the evolutionof the homeodomain-leucine zipper gene family in vascular plants. Mol Biol Evol.16(4):544-52.

Azpiroz-Leehan, R. and Feldmann, K.A. 1997. T-DNA insertion mutagenesis inArabidopsis: going back and forth. Trends Genet 13: 152-156.

Baima, S., Nobili, F., Sessa, G., Lucchetti, S., Ruberti, I., and Morelli, G. 1995. Theexpression of the Athb-8 homeobox gene is restricted to provascular cells in Arabidopsisthaliana. Development 121: 4171-4182.

Ballinger, D.G., and Benzer, S. 1989. Targeted gene mutations in Drosophila. Proc.Natl. Acad. Sci. USA 86: 9402-9406

Barton, M.K. and Poethig, R.S. 1993. Formation of the shoot apical meristem inArabidopsis thaliana: an analysis of developmen in the wild type and and in the shootmeristemless mutant. Development 119: 823-31.

Baumann E., Lewald J., Saedler H., Schulz B. and Wisman E. 1998. Successful PCR-based reverse genetic screens using a En-1-mutagenised Arabidopsis thalianapopulation generated via single seed descent. Theor Appl Genet 97:729-734.

Bechtold, N., Ellis, J. and Pelletier, G. 1993. In planta Agrobacterium mediatedtransfer by infiltration of adult Arabidopsis thaliana plants. C. R. Acad Sci Paris316:1194-1199.

Bellman, R. and Werr, W. 1992. Zmhox1a, the product of a novel maize homeobox gene,interacts with the Shrunken 26 bp feedback control element. EMBO J. 11: 3367-74

Billeter, M., Qian, Y., Otting, G., Muller, M., Gehring, W.J. and Wuthrich, K. 1990.Determination of the three-dimensional structure of the Antennapedia homeodomainfrom Drosophila in solution by 1H nuclear magnetic resonance spectroscopy. J MolBiol 214:183-97.

Brandon, E.P., Idzerda, R.L. and McKnight, G.S. 1995. Targeting the mouse genome: acompendium of knockouts. Curr Biol. 8:873-81.

Burns, N., Grimwade, B., Ross-Macdonald, P.B., Choi, E.Y., Finberg, K., Roeder, G.S.and Snyder, M. 1994. Large-scale analysis of gene expression, protein localization, andgene disruption in Saccharomyces cerevisiae. Genes Dev. 9:1087-105.

38

Bürglin,T.R., Finney, M., Coulson, A. and Ruvkun, G. 1989. Caenorhabditis eleganshas scores of homoeobox-containing genes. Nature 341:239-43.

Bürglin, T.R. 1994. A comprehensive classification of homeobox genes. In: Duboule, D.(Ed.) Guidebook to the homeobox genes, Oxford University Press, Oxford, England, pp.27-71.

Carabelli, M., Sessa, G., Baima, S., Morelli, G. and Ruberti, I. 1993. The ArabidopsisAthb-2 and -4 genes are strongly induced by far-red-rich light. Plant J 4: 469-479.

Carabelli, M., Morelli, G., Whitelam, G. and Ruberti, I. 1996. Twilight-zone andcanopy shade induction of the Athb-2 homeobox gene in green plants. Proc Natl AcadSci 93: 3530-3535.

Chan, R.L. and Gonzalez, D.H. 1994. A cDNA encoding an HD-zip protein fromsunflower. Plant Physiol. 106:1687-8.

Chang, C., Kwok, S.F., Bleecker, A.B. and Meyerowitz, E.M. 1993. Arabidopsisethylene-response gene ETR1: similarity of product to two-component regulators.Science 262:539-44.

Cutler, S., Ghassemian, M., Bonetta, D., Cooney, S. and McCourt, P. 1996. A proteinfarnesyl transferase involved in abscisic acid signal transduction in Arabidopsis. Science273:1239-1241.

Di Cristina, M., Sessa, G., Dolan, L., Linstead, P., Baima, S., Ruberti, I. and Morelli, G.1996. The Arabidopsis Athb-10 (GLABRA2) is an HD-Zip protein required forregulation of root hair development. Plant J 10: 393-402.

Ellenberger, T.E., Brandl, C.J., Struhl, K. and Harrison, S.C. 1992. The GCN4 basicregion leucine zipper binds DNA as a dimer of uninterrupted alpha helices: crystalstructure of the protein-DNA complex. Cell 71: 1223-1237.

Feldmann, K. and Marks, M.D. 1987. Agrobacterium-mediated transformation ofgerminating seeds of Arabidopsis thaliana: A non-tissue culture approach. Mol. Gen.Genet. 208: 1-9

Frank, W., Phillips, J., Salamini, F. and Bartels, D. 1998. Two dehydration-inducibletranscripts from the resurrection plant Craterostigma plantagineum encode interactinghomeodomain-leucine zipper proteins. Plant J 15: 413-421.

Garciarrubio, A., Legaria, J.P. and Covarrubias, A.A. 1997. Abscisic acid inhibitsgermination of mature Arabidopsis seeds by limiting theavailability of energy andnutrients. Planta. 203:182-7.

39

Gehring WJ. 1992. The homeobox in perspective. Trends Biochem Sci. 8:277-80.

Gehring, W.J., Qian, Y.Q., Billeter, M., Furukubo-Tokunaga, K., Schier, A.F.,Resendez-Perez, D., Affolter, M., Otting, G., and Wuthrich, K. 1994. Homeodomain-DNA recognition. Cell. 78:211-23.

Glover, J.N. and Harrison, S.C. 1995. Crystal structure of the heterodimeric bZIPtranscription factor c-Fos-c-Jun bound to DNA. Nature 373: 257-261.

Gonzalez, D.H., Valle, E.M. and Chan, G.G. 1997. Interaction between proteinscontaining homeodomains associated to leucine zippers from sunflower. BiochimBiophys Acta 1351:137-49.

Gosti, F., Beaudoin, N., Serizet, C., Webb, A.A.R., Vartanian, N. and Giraudat, J. 1999.ABI1 protein phosphatase 2C is a negative regulator of abscisic acid signaling. Plant Cell11:1897-1909

Hall, M.N. and Johnson, A.D. 1987. Homeo domain of the yeast repressor alpha 2 is asequence-specific DNA-binding domain but is not sufficient for repression. Science 237:1007-12.

Hua, J., and Meyerowitz, E.M. 1998. Ethylene responses are negatively regulated by areceptor gene family in Arabidopsis thaliana. Cell 94:261-271

Kaiser, K., and Goodwin, S.F. 1990. "Site-selected" transposon mutagenisis ofDrosophila.. Proc. Natl. Acad. Sci. USA 87:1686-1690.

Kawahara, R., Komamine, A. and Fukuda, H. 1995. Isolation and characterization ofhomeobox-containing genes of carrot. Plant Mol Biol. 27:155-64.

Kim, J. and Struhl, K. 1995. Determinants of half-site preferences that distinguish AP-1and ATF/CREB bZIP domains. Nucleic Acids Research 23: 2531-37.

Klee, H. and Estelle, M. 1991. Molecular genetic approaches to plant hormone biology.Annu. Rev. Plant Physiol. Plant Mol. Biol. 42:529-51.

Koornneef, M., Jorna, M.L., Brinkhorst-van der Swan, D.L.C. and Karssen, C.M. 1982.The isolation of abscisic acid (ABA) deficient mutants by selection of inducedrevertants in non-germinating gibberellin sensitive lines of Arabidopsis thaliana (L.)Heynh. Theor. Appl. Genet. 61:385-393.

Koornneef, M., Reuling, G. and Karssen, C. 1984. The isolation and characterization ofabscisic acid–insensitive mutants of Arabidopsis thaliana. Physiol. Plant. 61:377-383.

40

Korfhage, U., Trezzini, G.F., Meier, I., Hahlbrock, K. and Somssich, I.E. 1994. Planthomeodomain protein involved in transcriptional regulation of a pathogen defense-related gene. Plant Cell 6:695-708.

Krysan PJ, Young JC, Tax F, Sussman MR. 1996. Identification of transferred DNAinsertions within Arabidopsis genes involved in signal transduction and ion transport.Proc Natl Acad Sci U S A 93:8145-50.

Krysan, P.J., Young, J.C. and Sussman, M.R. 1999. T-DNA as an insertional mutagenin Arabidopsis. Plant Cell11:2283-2290.

Kubo, H., Peeters, A.J., Aarts, M.G., Pereira, A.. and Koornneef, M. 1999.ANTHOCYANINLESS2, a homeobox gene affecting anthocyanin distribution and rootdevelopment in Arabidopsis. Plant Cell 11: 1217-1226.

Landschulz, W.H., Johnson, P.F. and McKnight, S.L. 1988. The leucine zipper: ahypothetical structure common to a new class of DNA binding proteins. Science240:1759-64.

Latchman, D.S. 1998. Eukaryotic transcription factors, 3rd ed., Academic Press, SanDiego, USA

Lawrence, P.A. and Morata, G. 1994. Homeobox genes: their function in Drosophilasegmentation and pattern formation. Cell 78:181-189.

Lee, Y.H. and Chun, J.Y. 1998. A new homeodomain-leucine zipper gene fromArabidopsis thaliana induced by water stress and abscisic acid treatment. Plant MolBiol 37: 377-384.

Liu, Y.G., Mitsukawa, N., Lister, C., Dean, C. and Whittier, R.F. 1996. Isolation andmapping of a new set of 129 RFLP markers in Arabidopsis thaliana using recombinantinbred lines. Plant J 10:733-6.

Lloyd, A.M., Barnason, A., Rogers, S.G., Byrne, S.C., Fraley, R.T. and Horsch, R.B.1986. Transformation of Arabidopsis thaliana with Agrobacterium tumefaciens.Science 234, 464-66

Long, J.A., Moan, E.I., Medford, J.I. and Barton, M.K. 1996. A member of theKNOTTED class of homeodomain proteins encoded by the STM gene of Arabidopsis.Nature 379:66-9.

Lu, P., Porat, R., Nadeau, J.A. and O'Neill, S.D. 1996. Identification of a meristem L1layer-specific gene in Arabidopsis that is expressed during embryonic pattern formationand defines a new class of homeobox genes. Plant Cell 8:2155-68.

Martienssen, R. and Irish, V. 1999. Copying out our ABCs: the role of gene redundancy

41

in interpreting genetic hierarchies. Trends Genet 15:435-437.

Masucci, J.D., Rerie, W.G., Foreman, D.R., Zhang, M., Galway, M.E., Marks, M.D.and Schiefelbein, J.W. 1996. The homeobox gene GLABRA2 is required for position-dependent cell differentiation in the root epidermis of Arabidopsis thaliana.Development 122:1253-60.

Mattsson, J., Söderman, E., Svenson, M., Borkird, C., and Engström, P. 1992. A newhomeobox-leucine zipper gene from Arabidopsis thaliana. Plant Mol Biol 18: 1019-1022.

Mattsson J. 1995. Homeobox genes in plants. Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology, 140 Uppsala, ActaUniversitas Upsalaiensis.

Mayda, E., Tornero, P., Conejero, V. and Vera, P. 1999. A tomato homeobox gene (HD-zip) is involved in limiting the spread of programmed cell death. Plant J. 20:591-600.

McCourt, P. 1999. Genetic analysis of hormone signaling. Annu. Rev. Plant Physiol.Plant Mol. Biol. 50: 219-43.

McKinney, E.C., Ali, N., Traut, A., Feldmann, K.A., Belostotsky, D.A., McDowell,J.M. and Meagher, R.B. 1995. Sequence-based identification of T-DNA insertionmutations in Arabidopsis: actin mutants act2-1 and act4-1. Plant J 8:613-22.

Meijer, A.H., Scarpella, E., van Dijk, E.L., Qin, L., Taal, A.J., Rueb, S., Harrington,S.E., McCouch, S.R., Schilperoort, R.A. and Hoge, J.H. 1997. Transcriptionalrepression by Oshox1, a novel homeodomain leucine zipper protein from rice. Plant J11: 263-276.

Meijer, A.H., de Kam, R.J., d'Erfurth, I., Shen, W. and Hoge, J.H. 2000. HD-Zipproteins of families I and II from rice: interactions and functional properties. Mol GenGenet. 263:12-21.

Meinke., D.W., Cherry, J.M., Dean, C., Rounsley, S.D. and Koornneef, M. 1998.Arabidopsis thaliana: a model plant for genome analysis. Science 282: 679-82.

Meissner, R. and Theres, K. 1995. Isolation and characterization of the tomatohomeobox gene THOM1. Planta. 195:541-547.

Meyerowitz, E.M. and Pruitt, R.E. 1985. Arabidopsis thaliana and plant moleculargenetics. Science 229: 1214-18.

Moon, Y.H., Choi, D., Kim, J.C., Han, T.J., Cho, S.H., Kim, W.T. and Lee, K.W. 1996.Mol. Cells 6: 366-373.

Moon, Y.H., Choi, S.B., Kim, J.I., Kim, J.C., Han, T.J., Cho, S.H. and Lee, K.W. 1996.

42

Mol. Cells 6: 697-703.

Nusslein-Volhard, C. 1994. Of flies and fishes. Science 266:572-4.

O'Shea, E.K., Rutkowski, R. and Kim, P.S. 1992. Mechanism of specificity in the Fos-Jun oncoprotein heterodimer. Cell 68: 699-708.

Pabo, C.O. and Sauer, R.T. 1984. Protein-DNA recognition.Annu Rev Biochem.53:293-321.

Qian, Y.Q., Billeter, M., Otting, G., Muller, M., Gehring, W.J. and Wuthrich, K. 1989.The structure of the Antennapedia homeodomain determined by NMR spectroscopyinsolution: comparison with prokaryotic repressors. Cell 59:573-80.

Redei. G.P. 1992. A heuristic glance at the past of Arabidopsis genetics. In: Koncz, C.,Chua, N.H. and Schell, J. Eds., Methods in Arabidopsis research, World ScientificPublishing, pp. 1-15.

Rerie, W.G., Feldmann, K.A. and Marks, M.D. 1994. The GLABRA2 gene encodes ahomeo domain protein required for normal trichome development in Arabidopsis. GenesDev 8: 1388-1399.

Rock, C. and Quatrano, R. 1995. The role of hormones during seed development. InDavies P.J., (Ed.), Plant Hormones: Physiology, Biochemistry and Molecular Biology.2nd ed Norwell, MA, Kluwer Academic Publishers. 671–697.pp.

Ruberti, I., Sessa, G., Lucchetti, S. and Morelli, G. 1991. A novel class of plant proteinscontaining a homeodomain with a closely linked leucine zipper motif. EMBO J 10:1787-1791.

Schena, M. and Davis, R.W. 1992. HD-Zip proteins: members of an Arabidopsishomeodomain protein superfamily. Proc Natl Acad Sci U S A 89: 3894-3898.

Schena, M. and Davis, R.W. 1994. Structure of homeobox-leucine zipper genes suggestsa model for the evolution of gene families. Proc Natl Acad Sci U S A 91: 8393-8397.

Schindler, U., Beckmann, H. and Cashmore, A.R. 1993. HAT3.1, a novel Arabidopsishomeodomain protein containing a conserved cysteine-rich region. Plant J 4: 137-50.

Sessa, G., Morelli, G. and Ruberti, I. 1993. The Athb-1 and -2 HD-Zip domainshomodimerize forming complexes of different DNA binding specificities. EMBO J 12:3507-3517.

Sessa, G., Carabelli, M., Ruberti, I., Lucchetti, S., Baima, S. and Morelli, G.1994. Identification of distinct families of HD-Zip proteins in Arabidopsis thaliana. In:Puigdomene´ch, P. and Coruzzi, G. Eds., Molecular-Genetic Analysis of Plant

43

Development and Metabolism, Springer Verlag, Berlin, Germany, pp. 411-426.

Sessa, G., Morelli, G. and Ruberti, I. 1997. DNA-binding specificity of thehomeodomain-leucine zipper domain. J Mol Biol 274: 303-309.

Sessa, G., Steindler, C., Morelli, G. and Ruberti, I. 1998. The Arabidopsis Athb-8, -9and –14 genes are members of a small gene family coding for highly relatedHD-ZIP proteins. Plant Mol Biol 38: 609-622.

Smith, L.G., Greene, B., Veit, B. and Hake, S. 1992. A dominant mutation in the maizehomeobox gene, Knotted-1, causes its ectopicexpression in leaf cells with altered fates.Development 116: 21-30.

Spradling, A.C., Stern, D.M., Kiss, I., Roote, J., Laverty, T. and Rubin, G.M. 1995.Gene disruptions using P transposable elements: an integral component of theDrosophila genome project. Proc Natl Acad Sci U S A 24:10824-30.

Steindler, C., Matteucci, A., Sessa, G., Weimar, T., Ohgishi, M., Aoyama, T., Morelli,G. and Ruberti, I. 1999. Shade avoidance responses are mediated by the ATHB-2 HD-Zip protein, a negative regulator of gene expression. Development 126: 4235-4245.

Swofford, D. L. 2000. PAUP. Phylogenetic Analysis Using Parsimony. Version 4.Sinauer Associates, Sunderland, Massachusetts

Söderman, E., Mattsson, J., Svenson, M., Borkird, C. and Engström, P. 1994.Expression patterns of novel genes encoding homeodomain leucine-zipper proteins inArabidopsis thaliana. Plant Mol Biol 26: 145-154.

Söderman, E., Mattsson, J., and Engstrom, P. 1996. The Arabidopsis homeobox geneATHB-7 is induced by water deficit and by abscisic acid. Plant J 10:375-381.

Söderman E., Hjellstrom, M., Fahleson, J. and Engstrom, P. 1999. The HD-Zip geneATHB6 in Arabidopsis is expressed in developing leaves, roots and and up-regulated bywater deficit conditions. Plant Mol Biol 40:1073-83.

Tornero, P., Conejero, V. and Vera, P. 1996. Phloem-specific expression of a planthomeobox gene during secondary phases of vascular development. Plant J 9:639-48.

Treisman, J., Harris, E., Wilson,D. and Desplan, C. 1992. The homeodomain: a new facefor the helix-turn-helix? Bioessays 14:145-50.

Vinson, C.R., Hai, T. and Boyd, S.M. 1993. Dimerization specificity of the leucinezipper-containing bZIP motif on DNA binding: prediction and rational design. GenesDev 7: 1047-1058.

44

Vollbrecht, E., Veit, B., Sinha, N. and Hake, S. 1991. The developmental gene Knotted-1is a member of a maize homeobox gene family.Nature 350:241-3.

Winkler, R.G., Frank, M.R., Galbraith, D.W., Feyereisen, R. and Feldmann, K.A. 1998.Systematic reverse genetics of transfer-DNA-tagged lines of Arabidopsis. Isolation ofmutations in the cytochrome p450 gene superfamily. Plant Physiol 118:743-50.

Wolberger, C., Vershon, A.K., Liu, B., Johnson, A.D. and Pabo, C.O. 1991. Crystalstructure of a MAT alpha 2 homeodomain-operator complex suggests a general modelfor homeodomain-DNA interactions. Cell 67:517-28.

Zhong, R., and Ye, Z.-H. 1999. IFL1, a gene regulating interfascicular fiberdifferentiation in Arabidopsis encodes a homeodomain-leucine zipper protein. Plant Cell11:2139-2152.

Zwaal, R.R., Broeks, A., van Meurs, J., Groenen, J.T.M., and Plasterk, R.H.A. 1993.Target-selected gene inactivation in Caenorhabditis elegans by using a frozen transposoninsertion mutant bank. Proc. Natl. Acad. Sci. USA 90: 7431-7435