GENE-REGULATORY INTERACTIONS AND MECHANISMS OF TARGET GENE SELECTiON OF THE DROSOPHILA HOMEODOMAM PROTEIN FUSHT TARAZU

Andrzej Nasiadka

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Molecular and Medical Genetics University of Toronto

CICopyright by Andrzej Nasiadka (200 1) Naüonal Library Biôiiihépue nationale du Canada Acquisitions and AçquWitians et Bibliograptiic Services services bibliographiques 395 WaKiStreet 395. Ne WeUmgton OnawaON KlAON4 Octa~aONKlAOW Canada Canada

The author has graated a non- L'auteur a accordé une licence non exclusive licence dowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distriiute or seli reproduire, prêter, disûiher ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or elecîronic formats. la forme de microfiche/fh, de reproduction sur papier ou sur format électronique.

The author cetains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thése ni des extraits substantiels may be printed or ohenvise de celle-ci ne doivent être imprimés reproduced wittiout the author's ou autrement reproduits sans son permission. autorisation. This dksertation ii dedicated to my mother, Janina Nasiadka, and fo the memory of my fother, Jozef Nasiadka. Gene-regulatory Interactions and Mechanisms of Target Gene Selection of the Drosophila

Homeodomain Protein Fushi Tarazu

Doctor of PhiIosophy, 7001

Andrzej Nasiadka

Department of Molecular and Medical Genetics, University of Toronto

Abstract

The Fushi tarani protein (FTZ), a member of the homeodomain protein family, is involved in a number of developmental processes during Drosophila embryogenesis. These include subdivision of the embryo into rnetameric units and specification of these units with unique segmenta1 identities.

To execute these bnctions, FTZ acts as a transcription factor. Although FTZ has been the subject of intense snidy, the mechanisms of FTZ-dependent regulation are still, for the most part, unknown.

One of the issues that 1 have addressed is which of the genetically identified FTZ downstream genes are directly and which are indirectly regulated by FTZ. In my studies, 1 have carried out a kinetic analysis of downnream gene responses to an ectopic pulse of FTZ expression (chapter 2).

The results have demonstrated that the responses of putative target genes fall into easily distinguished temporal windows. t argue that these different windows reflect the responses of direct and indirect targets. Based on these studies, 1 propose that FTZ involvement in the formation of even-numbered parasegments is to establish a number of direct gene regdatory interactions. 1 aIso suggest that in the Drosophila embryo. FTZ acts not only as a transcriptional activator but also as a repressor.

Another issue that I have addressed is how cofactors involved in FTZ-dependent regulation affect FTZ fùnction. 1 have approached this by analysing the activity of a chimeric FTZ derivative that contains a potent and autonomous heterologous activation domain (chapter3). The results of this study suggest that FTZ target gene selection is conuolled at the IeveI of DNA binding and that

FTZ binds and regdates a small subset of Drosophila gnes. This implies that FTZ cofactors act ptimarily by recmiting FTZ to the regulated target gene promoters- However, some of FTZ cofactors may also conml FTZ transcriptionai activity as regdation of this activity has been found cmcial for precise tempord and spatial expression of FTZ mget genes. Acknawledgmenis

1 would like to acknowledge those who have contributed to this work. Fim, 1 would like to thank my Ph.D. supervisor, Dr, Henry Krause, for his enomous heip and support over the years. [ remember, as if it were yesterday, when i came to his lab for the first time. 1 jus1 arrived in Canada from Spain, and my English was not very good. Dunng our first meeting, Henry was patiently explaining to me powerful transcriptional properties of the VPI6 activation domain. To my surprise, [ understood most of what he was saying. It was not the last time that his patience was preatly appreciated. 1 also want to acknowledge other members of my advisoy commirtee: Dr. J. Segall and Dr. A. Spence. who replaced Dr. V. Giguere as the third commirtee member. 1 express rny appreciation for their advice and guidance in my research as well as for their close reading of my dissertation, making corrections. and proposing revisions. [ am likewise indebted to Dr. H. Lipshitz and Dr. M. Biggin for their valuable critiques on portions of my dissertation. 1 also wish to thank Dr. K. Cadigan, Dr. J. Greenblart and Dr. C. Smibert for their helpful comments and criticisms. 1 owe thanks to Dr. A. Percival-Smith for his gracious assistance and suppon during the initie1 stage of rny graduate research. He taught me the principles of Drosopkila genetics and techniques for generating transgenic animals. 1 also wish to express my gratitude to Dr. D. Fitzpanick, Dr. S. Treizenberg, Dr. C. Desplan. Dr. S. DDardo, and Dr. W. Gehring for providing me with DNA constructs. antibodies. and fly lines. To a11 the mernbers of the Krause lab. past and present. 1 express my heanfeit appreciation for their help. suppon and encouragement. 1 particularly want to tfiank John CopeIand, Sarah Hughes, Bruce Dietrich. tzhar Livne-Bar, Hong L. Hung, and Sam Scanga for their friendship. t ah thank AlIan Grill who worked with me as a summer student. In addition. 1 am thankful to Andrew Simmonds whose cornputer skills enriched this project. A special debt of gratitude is owed to Megan and Kyle Kohb-Wiebe for their fnendship and help. Over the years, Megan and Kyle have kindly accommodated my numerous requests for fly food and apple juice plates. [n addition. 1 want to thank Heidi Sampson and Gilbert dosSantos for their fiendship and help as well as providing me with cinnarnon buns on weekends, My thanks also go to Zofia Glogowska (pani Zosia) for her support and encouragement. 1 wish to express rny gratitude to my fiends Brea Lenehan and Christine Erbs. Brett's kindness. generosity. encouragement and, above all. friendship are beyond teIling. AI1 of these have been sincereiy appreciated and treasured. especiaily during some very Qing times. C ahthank Chris for her support and Company. Poeuy and foreign films that we have read and watched together might have not directly affected my scientific advances. but they sure. Iifred up my spirits at times when Zt was desperately needed. Most of dl, 1 would like to thank my family for their love. assistance. and unfaltering confidence in me. I owe special _gratitude to my parents for king so faithful in theu support for my graduate work. Unfortunately, chanks to my father corne too late as he suddenly died several months ago. His extraordinary altitude towards life, always a me inspiration to me, conmbuted enonnously to my perseverance in this long projecr. My gratitude also goes to Marysia, Jurek, and Janusz for their assistance and encouragement during rny graduate work. And to anybody 1 may have unintentionaIly ornitted. thank you. Table of Contents

.. Thesis Abstract ...... it ... Acknowledgements ...... iii Table of Contents ...... v .. - List of Figures...... viii List of Abbreviations...... ix .- List of Drosophila Cenetic Symbols ...... ------...... 'rit List of Symbols ...... xiv

Chapter 1: Introduction 1 .1. Developmental transcriptional regulatory hietarchies ...... -1 1-1 -1. History ...... -7 1.1.2. Developmental functions of transcriptional cascades ...... 3 1.1.3. Drosophila segmentation...... 5 1.1 .3.l. Identification of factors required for the formation and specitkation of the Drosophila anterior-posterior body auis ...... 10 1.1 3.2. Materna1 genes ...... 13 1. I.J.J.3 1 Gap genes ...... - ...... 16 1.1.3.4. Pair-mle genes ...... 1 7 1.1.3 .l. Segment polarity genes ...... 25 1.l.3.6. Horneotic genes ...... 28 1.?. Characterization of the fi: gene ...... 30 1.2.1- Identification and isolation of the* gene ...... 30 1.2.7. Spatial and temporal expression offi: ...... 3 1 1.2.3. Transcriptional regulation of the$= gene...... 3 4 1.2.3.1. Analysis of che frz promoter ...,..-...... - .--...... 34 1-22-2. Trans-acting regulators ofjz expression ...... 3 9 1.2.4. The fushi tarm protein (FTZ) ...... - ...~...... - 47 1.2.4.l. The homeodomain ...~...~...... -.-...... 50 12-42Nuclear localization signals ...... 50 1 .2 -4.3. PEST regions,...... *.-*-...-...... -...... 50 1.2.4.4. Transcriptional activity of FTZ in vitro, in cultured Drosophila celIs and in yeast ...... ++...... ---...... *.....--.--.-..---.-5 1 I U.5. FTZ fimction in the developing embryo ...... 53 1.3. Determinants of developmental and regulatory specificities of homeodomain proteins ....~....,,,,..,..,.~~~~~~~.~..~~..~...... ~.~~~~~~....~...... ~....64 1.3.1. Sequence-spcific DNA binding of the homeodomain ...... 64 1.3.1.1, DNA-binding specificity of the homeodomain is not suficient to account for unique developmental hnctions of homeodomain proteins ...... 65 1.3 .2 . Contributions of combinatorial interactions to funcrional specificities of homeodomain proteins ...... 67 1.3.2.1 MATa2, MATal and hKM1 ...... 68 1.3.2.2. HOX and PBC proteins...... 69 1.32.3. OCT- I and OCT-2 ...... 72 1.3.2.4. I-POU and Cfl-a ...... 73 1.3.2.5. Unc-83 and Mec3...... 73 1.33. Othcr DNA-binding domains...... 74 1.3.4. Protein expression levels ...... 75

Chapter 2: Kinetic analysis of segmentation gene interactions in Drosophifa embryos 2.1 Abstract...... 78 2.2 Introduction ...... 78 2.3 Materials and methods ...... 81 2.4 Results ...... , ...... 83 7.5 Discussion ...... 101

Chapter 3: Mechanisms regulating target gene selection by the homeodomain-contaiaiag protein Fushi tarazu 3.1 Abstract...... 112 3.2 Introduction ...... 112

'I J.J.. Materials and methods ...... 1 15 3.4 Results ...... 117 . . 3., Discussion ...... 138

Cbapter 1: Discussion and future work 4.1. Summary ...... 143 4.1.1. Kinetic assessrnent of FTZ gne-regulatory interactions...... 143 4.1.2. The mechanisms by which cofactors affect FTZ function...... 144

4.2. Discussion and future work ...... ,, ...... 144 42.1. Are homeotic genes regulated by FTZ?...... 145 4.2.2. How many FTZ target genes are there in the Drosophila genome.. 1J6 423. Modes of FTZ-dependent regulation ...... 148 4.2.4. fiz autoregulation ...... 149 vi . 4.2.5. Modulators of FTZ activity ...... 15 1 4.2.5.1. Factors affecting the temporal window of FTZ transcriptional activity ...... 152 4.2.5.1. Functional interactions of FTZ with aperiodically distributed factors ...... 154 4.2.5.3. Functional interactions of FTZ with periodically distributed factors ...... 157 4.2.6. Mechanisms controlling regulatory and developmental specificities of Q50 homeodomain proteins ...... 162 4.3. Conclusion ...... 163 Refereaces ...... 166

vii List of Figures

Figure 1.1 Metamerization of the Drosophila body plan ...... 7 Figure 1.2 Embryonic development of Drosophila ...... 9 Figure 1.3 Types of segmenta1 mutant phenotypes ...... 12 Figure 1.4 Spatial expression of materna1 and gap gene proteins ...... -1 5 Figure 1.5 Expression of pair-mle and segment polarity genes ...... 19 Figure 1.6 Initiation of expression patterns of segment polarity ...... 24 Figure 1.7 Evolution of& stripe expression ...... 3 Figure 1.8 The fi- promoter subunits...... 37 Figure 1.9jz protein sequence ...... 49 Figure 2.1 Kinetics of endogenous fk activation...... 86 . * Figure 2.2 The kinetics of en and& activation... are similar...... ,...... A9 Figure 2.3 Direct and indirect responses of wg ...... 91 .. Figure 7.4 prd is required for acttvation of rvg ...... 94 Figure 1.5 Expression of eve is unaffected by ectopic FTZ ...... 97 Figure 5.6 Fi2 activates odd and represses slp ...... 1 00 Figure 1.7 Indirect regulation of gsb by FTZ ...... 103 Figure 7.8 Direct gene-regulatory interactions triggered by FTZ ...... 108 Figure 3-1 Structure, activity and expression of FTZ derivatives ...... 1 19 Figure 3 .2 Cuticular phenotypes caused by ectopic FTZVP 16 ...... 123 Figure 3.3 Effects of ectopic FTZ and F'TZVP16 on the endogenous fi gene...... 126 Figure 3.4 Effects of FTZVP L6 on engraiied (en) ...... 129 Figure 3.5 Effects of FTZVP16 on genes that are repressed by FTZ ...... 132 Figure 3.6 Effects of FTZVPI6 on non-FTZ target genes...... 13.5 Figure 3.7 The FTZ homeodomain is dispensable for most FTZVP 16 activities...... 1 37 Figure 4.1 Enhancement of FïZ and FTZVPl6 activity in the posterior portion of the embryo ...... 156 Figure 4.2 Negative regulation of FTZ activity by SLP ...... 161 List of Abbreviatians

323 ffE 323 bp& proximal Enhancer AI first abdominal segment A2 second abdominal segment A3 third abdominal segment A4 fourth abdominal segment AS fifih abdominal segment A6 sixth abdominal segment A7 seventh abdominal segment A8 eighth abdominal segment A9 ninth abdominal segment A 10 tenth abdominal segment AE -Autoregulatory Element AEL After Egg Laying AF2 -Activation bnction 3 approx. approximately asg -a-specific genes bHLH basic Nelix bop Helix Bm Bombyx mari CAT -Chlorarnphenicol AcetyIIransferase

cDNA -corn plementary DNA C/E B P CCMTIEnhancer-Binding Botein CNS -Central -NOUS %stem DtG digoxigenin Dpe -Downstream gornoter dement DNA deoxyribonucleic acid DNase deo.qribonuclease dUTP deoxy-uracil-triphosphate

EMSA -Electrophoretic Mobility Shift bsay EL -Embryo bngth fAE fi- Activation Eletnent FANT -FTZ-FI. NF-l, protein N, DK €DE fioual Element FTZA 1 12- 1 18 a version of FTZ with a deleted nuclear receptor box FTZAC a version of FTZ with 338-413 amino acids deleted FTZAHD a version of FTZ with most of the homeodomain deleted FTSAHDVP 16 a version of FTZAHD that contains a heterologous activation domain of VP 16 FTZAN a version of FTZ with 101-150 amino acids deleted FTZANVP 16 a version of FTZVPI6 with 1-271 amino acids deleted FTZT763 A a version of FTZ in which Thr263 is substituted with Ala FTZT263 D a version of FTZ in which Th263 is substituted with Asp FTZVP 16 a version of FTZ that contains a heterologous activation domain of VPl6 gai galactosidase HCF -Host CeIl Factor HD homeodomain hnRNP heterogeneous nuclear ribonucleoprotein complex HS -Heat Shock hsg -haploid-specific genes hsp70 heat shock 70 promoter HSV -Herpes Simplex Yirus IE immediate Early In r Initiator JAK Janus protein tyrosine kinase Lb labial segment LBD -Ligand Binding -main LtM -LIN- I1, LSLET-1, and MEC3 MADS -MCM 1, m. DEFA and $RF MAT mating type MBF -Mediator of BmFTZ-F 1 MCM maintenance of minichromosomes Md mandibular segment MD€ mogenic &termination Factor MEF Wocyte-specific Bhancer-binding -ctor MeOH methanol 'L& relative molecular mas mRNA messenger ribonucleic acid Mx maxillary segment NLS -Nuclear bcalization Signal ORF open Reading Frame PAGE polyacrylamide gel electrophoresis PBX pre-B ce11 homeobox PCR periplasm peripheral cytoplasm PEST protein region rich in Pro, Glu, Ser, and ïhr PKA CAMP-dependent protein kinase POU -PITT QCT, WC PPAR -Peroxisome Proliferator-btivated kceptor PS parasegment RNA ribonucleic acid 52 -Schneider line 2 Sm -Sodium Dodecyl Sulfate snRNA gnall ouclear RNA STAT -Signal Jj-ansducer and Activator of lYanscription T 1 first thoracic segment T2 second thoracic segment T3 third thoracic segment TALE nree Amino Acid bop Extension TBP -TA-Binding Protein TFIIA -Transcription Eactor HA TFIlB Transcription Eactor IIB tS gmperature sensitive UV List of Droso~hilaCenetic Svmbols

A34B deficiency Df(2L)slp abdrl abdominal A AbdB Abdominal B .-!df- 1 lllcohol dehydrogenase rranscriprion factor-1 Antp -4nrennapedia ANT-C Antennapedia Complex bcd bicoid bfd bzirronhead b.r bithorar BX-C bithorax Cornplex cad catidal .O .O Curly of Oster. second-chromosome balancer dCTBP Drosophila C-terminal Binding Prorein Ddc Dopa dticarbo.ylase Dfd Deformed Dhr39 Drosophila hormone recepfor in 39 DI1 Disral-less ems emp@ spiricales en engrailed eve even-skipped e-rd e-rrradenricle Fh fork head FM6 first-chromosome balancer fi= furhi tara-= jz5 fushi rarazu mutant allele @Hf 4 fihi tarua mutant allele frRP' &hi tara- mutant aIleIe fr=i*20 *hi tara- mutant aIlele fi-fl fk transcriptionfaclor 1

PO grottcho gsb goosebers, gr gianr h haip hb hunchback hkb huckebein HOM homeotic hlh homothorar kni knirps KT ~PP~I [ab labial nos nanos odd odd-skipped

OP0 odd-paired ord orrhodenticle pb proboscipedia pbx postbirhorax PcG Polycomb group prd paired prd= "5 " puired mutant al lele ptrmilio run runt rom rosy mutant allele Srubble Scr Sex combs reduced sloppy-paired squid third-chromosome balancer railless rrirhorax group trk tramtrack Ual Ultra abdominal-like Ubx Ultrabithoras wingless :erknullr

xiii List of Sybals

number alpha beta delta degree Celsius micro ~-base pair hour kilobase pairs kilodalton molar concentration percentage Chapter 1:

Introduction -7 Transcriptional hierarchies play important roles during the development of bath prokaryotic and eukaryotic organisms. One of the best-investigated transcriptional cascades is the segmentation hierarchy operating during early Drosophila embryogenesis. This hierarchy is required for subdividing the embryo into metarneric units and rendering these units with unique segmental identities. Although the segmentation hierarchy has been a subject of intense study, a number of issues still remain to be addressed. One of these is the distinction of which genetically characterized regulatory interactions within the hierarchy are direct and which indirect. Additional efforts should be directed towards a beaer understanding of mechanisms underlying combinatorial regulation, which is commoniy employed to control transcriptional events at segmentation gene promoters. In this thesis, 1 have attempted to address these issues by focusing on the function of a transcriptional regulator operating within the Drosophila segmentation hierarchy, a homeodomain- containing protein called Fushi tarani (FTZ). In the introduction, 1 first provide an overview of developmental transcriptional cascades, placing an emphasis on the Drosophila segmentation hierarchy. This is followed by a description of thefushi tarau gene (fc). Finally. 1 discuss the strategies used by homeodornain-containing regulators to achieve their regulatory and developmental specificities. One of these strategies involves combinatorial interactions between homeodomain- containing proteins and their specific cofactors.

1.1. DEVELOPMENTAL TRANSCRPTIONAL REGULATORY HIERARCHlES

1.1.1. HISTORY

Transcriptional cascades employ sequential gene regulation at the levei of manscript initiation, The idea that expression of eukaryotic developmental loci can be organized into hierarchical nenvorks was originally inspired by the prokaryotic operon-repressor mode1 of Jacob and Monod (Jacob and Monod 1961). In 1962, Waddington suggested that -in organisms which are evolutionarily more advanced than bacteria numbers of 'Jacob-Monod systems' might have ken interlinked. If a structural gene controlled by an operator in the first system produced a substance which functioned as a repressor substance for an operator in a second system, we woutd have the possibility of 'cascade repression': and if there were a number of links of this kind, complex systems might be buiIt up which eshibit some of the tendency towards ineversibility which is commonly found in embryological materials but which is hardly accounted by the simple Jacob-Monod scheme3 (Waddington 1962). "A theoty for gene regulation of higher organismst proposed by Britten and Davidson (Britten and Davidson 1969) was one of the first well-elaborated hypotheses on the hierarchical organization of developmental genes. ïhe authon proposed that large genome sizes of higher orpnisrns do not arise as a result of an increased nurnber of structural genes but from expansion of regdatory networh coordinating genetic activities that are required for a &en state of differentiation. Transcriptional cascades were therefore postulated to represent powerful systems integrating numerous batteries of developmental genes. The authors assumed that the genome of higher organisms is typically inactive due to the histone-mediated repression and that gene reguiation requires specific activation at the Ievel of transcript initiation. Interestingly. this activation was postulated to occur upon sequence-specific DNA binding of "an activator RNA" rather than regulatory proteins. This was to assign a function to a large pool of non-translated RNAs found in the nuclei of higher cell types. In Drosophila, the first modeis of transcriptional hierarchies were proposed by Garcia-Bellido to account for regulation and regulatory propenies OFthe homeotic genes (Garcia-Bellido 1977), and by Ashbumer et al. to describe temporal control of polytene chromosome puffing (Ashbumer et al. 1974). Drosophila homeotic genes are required to assign distinct identities to particular body segments (Lewis 1978) (see below), Mutations in these loci cause one part of an organism to develop into the likeness of another part. Since homeotic activities appear to control development by selecting among alternative developmental pathways or prograrns. these genes are referred to as selector genes. In his model of a transcriptional cascade (Garcia-Bellido 1977), Garcia-Bellido proposed that cornpartment-specific expression of selector genes is controlled by differentially distributed activators interacting with ubiquitous repressors. The rnechanism of this activation was suggested to involve DNA modification of the selector locus. Afier induction, homeotic gene products were proposed to act combinatorially to regulate the expression of realisator genes. products of which would be directly involved in the specification of morphogenetic cellular propenies. Thus, the divergent developmentai pathways of Drosophila compartments were postulated to be defined by the sequential action of activators. selectors. and realizators. Ashbumer et al. proposed a transcriptional cascade mode1 for the genetic control of ecdysone-induced polytene chromosome puEng (Ashbumer et al. 1974). Pufis are locally enlarged regions of polytene chromosomes. They represent actively transcribed genes. Temporal patterns with which puffs appear provide information regarding gene networks. In response to an ecdysone pulse, two temporally different secs of puffs can be distinguished, "early" and -late" puffs. "EarIy" puffs are rapidly induced. whereas the appearance of -late" puffs is only observed several hours afier exposure to the hormone, Ashburner and colleagues (Ashbumer et al. 1974) proposed that the eariy puffs are directly induced by receptor-bound ecdysone, whereas the late puffs arise as a consequence of activities of the early puff proteins. In addition to inducing the late puffs. the early puff products were also postulated to repress their own activity, self-attenuating the regulatory response to the hormone. The model fiirther suggested that ecdysone directly teptesses lace puff activity, preventing theu prernature induction by the early puff proteins.

1.13. DEVELOPMENTAL F'UNCTIONS OF TRANSCRIPTIONAL CASCADES 4 Transcriptional hietarchies play diverse developmental hnctions in borh prokaryotes and eukaryotes. In Bacillus subtilis, for example, a regulatory cascade operates during sporulation where it is required for coat biogenesis (Zheng and Losick 1990). Each gene set in this cascade is responsiblr for the production of a transcriptionat activator for the next gene set in the sequence. A predominant pans-acting regulator in this cascade is a subunit of the bacterial RNA polymerase holoenzyme, the sigma factor. which is required for promoter recognition. Another example of a prokaryotic hierarchy is sequential gene expression during flageltum biogenesis in Caulobacrer crescenzus, Escherichia coli and Salmonella yphimurium. Flagellum biosynthesis involves interactions of many gene products operating at several different levels within the transcriptional cascade. In Caulobacrer crescenrus. a hierarchical regulatory nenvork conmls the timing of expression of individual flagellar genes (Ohta et al. 1991). In consequence, the hierarchy dictates the order of flagellum assembly. as this order is correlated with the time of gene expression (Lagenaur and Agabian 1978; Loewy et al. 1987: Minnich and Newton 1987). In Drasophila. one of the most comprehensively investigated transcriptional hiecarchies is the segmentation cascade openting during early embryogenesis (Ingham 1988). The role of this cascade is to subdivide the etnbryo into increasingly smaller metamenc units that are related to segmental primordia, Transcription factors at higher levels of the segmentation hierarchy are disaibuted in the fom of broad gradients, They induce a large number of locally expressed subordinate regulators whose combinatorial action specifies different cell faces and subdivides the embryo into a regular array of rnetameric units. The se_mentation hierarchy is the subject of my study and is considered in more detail below, During the development of higher eukaryotes. transcriptional regulatory cascades play important roles during organogenesis and ce11 differentiation, For example. much of the regdation of adipocyte differentiation is controlled by a regulatory hierarchy whose major components belong to the CEBP and PPAR families of transcriptionaI regulators (Brun et al. 1996). The adipogenic cascade is strongly influenced by a variety of environmental signals that regu1a:e activities of transcription factors through covalent modification. protein-protein or protein-ligand interactions (Fajas et al. 1998). The ernployment of a transcriptional regulatory hierarchy provides multiple control ievels, allowing for eficient integrarion and sirnultaneous coordination of diverse hormonal and nutritional signals- A transcriptional cascade also appears to be important for the myogenic pathway during specification of the skeletal muscle cell lineage. Major components of this hierarchy are basic helk- loop-helix (bHLH) yogenic determination tactors (MDFs) which become activated in early development by upstream repulators (Arnold and Winter 1998). MDFs autorepulate their own expression and activare downstream muscle-specific senes either directly, by bindhg their promoter conml regions (Lassar et a!. 1989: Wentworth et al. 1991): or indirectiy, through activation of intermediary regulatory factors such as MEF-7 fmily pmteins (Gossett et al. 1989; Csejesi and 5 Olson 1991). MEF-2 proteins act independently or in conjunction with MDFs to induce muscle- specific transcription (Edmondson et al. 1992). The generation of mature blood celb dso seems to depend on hierarchical gene-regulatory interactions. In tecent years, many transcriptional reçulators required for hematopoietic stem cell differentiation have been identified (Shivdasani and Orkin 1996). Based on the timing of their expression and hnction, these regulators have been organized into a molecular cascade (Clevers and Grosschedl 1996). Further research is required to confirm this order and to elucidate mechanisms underlying the hematopoietic regulatoty network.

The process of segmentation is initiated during early stages of embryogenesis. lt relies predominantly on the sequential transcriptional regulation of several classes of genes. The immediate outcome of this cascade is the subdivision of the embryo into 14 metameres which as of yet are not segments but parasegments. the hndamental units in the segmentation of the Drosophila embryo (Figure 1.1) (Martinez-Arias and Lawrence 1985). Morphologically, the parasegments are defined by transient grooves that appear on the surface of the embryo afier gasuulation (Figure 1.2). The parasegments are initially simil. but each wiII come under the control of a specific set of genes. eventuûlly acquiring its own identity (Garcia-Bellido 1975: Garcia-Bellido et al. 1979). Clonal analysis via induced mitotic recombination has demonmted that the anterior margin of each parasegment is a boundary of ceII Iineage restriction. i.e. cells and their descendents from one parasegment never niove into adjacent ones (Garcia-Bellido et al. 1973). This finding was funher confirmed by marking single cells with a fluorescent compound during early embryogenesis and observing their descendants during later mges (Vincent and O'Farrell 1992). Such domains of Iineage restriction are known as compartments (see Figure 1.1) (Garcia-Bellido et al. 1973: Garcia-Bellido 1975: Garcia-Bellido et al. 1979). Compartments are most likely established for growth control and the regulation of size and shape (Crick and Lawrence 1975: Lawrence and Morata 1976: Karlsson 1984: Gubb 1985). In addition. compartments. and in panicular some of their features such as compamnental boundaries. conninite a source of morphogens, developmentaliy active substances capable of organizing the pattern in a concentrationdependent manner (Garcia-Bellido 1975: Lawrence and Morata 1976: Schubiger and Sctiubiger 1978; Wilcox and Smith 1980; Karlsson 1981: Meinhardt 1983: Karlsson 1983: Meinhardt 1986b; Lawrence and Smhl 1996). Despite the fact that parasegmental gmves are transient (Martinez-Arias and Lawrence 1985), the Iineage restriction of the anterior margin of the parasegrnent is carried over into the segments of the larva and the adult (Sauhl 1984; Cohen et al. 1991; Couso et al. 1993; Couso and Gonzalez-Gaitan 1993). The parasegment. are out of register with the fmal segmenta1 units by approximately half a se-ment. Thus, each se-ment is made up of the posterior region of one Figure 1.1 Metamerhtion of the DrosopltiZo body plan. The early Drosophila embryo is subdivided into metarneric units called parasegments. During later developrnent, the anterior region of a parasegment becornes the posterior portion O€ a segment. Segments are thus offset hmthe original parasegrnena by about half a segment. Upon its esrablishmen&a parasegment is also a cornpartment. Later, a limage restriction develops within each parasegment subdividing it into an anterior (designated as a) and posterior (designated as p) compamnents. Two compartments can be also distinguished within each segment. The segment specification of the late embryo is canied over into the larva and the adult. In the adult, appendages such as legs and wings develop on specific segments. Md, Mx, Lb are gnathaI segments, which becorne fused to fonn the head region. Tl-T3 are thoracic segments whereas A 1-AI O are abdominaf segments. (Adapted from Lawrence 1992) Early embryo

Late embryo

First instar lama

acmn -

Adult fly 8 Figure 1.2 Embryonic development of Drosophila. After fertilization (A), nuclei divide and begin to migrate to the periphery of the embryo (B). At stage 4, they reach the cortex and form a syncytial blastodem (C). The pole cells, which later give nse to the genn Line of the animal. form at the posterior end of the ernbryo (C). [t is at this time that productive zygotic transcription is first detected. Shortly thereafier, cellularization begins in the cortical cytoplasm, with plasma membranes gradually progressing from the apical to the basai periplasm (stage 5) (D). During this process, the base of each ce11 rernains open to the undedying yolk rnass. Nonetheless, free diffusion of factors benveen nuclei, a distinctive feanire of previous stages, is now greatly Iimited. Cellularization of the blastodem is followed by gastrulation (stage 6). Dunng this stage, first morphogenetic movements take pIace, resutting in the formation of the cephalic and ventral hrtows (E). Almost concurrent with gastrulation is the process of the germ band elongation (F,G). In this process, the posterior end of the embryo (with the pole cells) is dnven round to the dorsal side of the egg. As a result, a horseshoe arrangement of the germ Iayer tissue is fonned (G). Germ band elongation is followed by further ceIl movement, This includes determination events in the nervous system. During stage 10, parasegmental grooves (first morphological markers of segmentation) becorne visible (G). In stage II, genn band extension is at its mauimum. The genn band then reaacts (stage E),and as it does so, the epidennal parasegmental grooves are nanslated to the position of &te definitive segmental -erooves, which coincide with the location of tracheal pits (H). The stages that follow (stages 13-17) include events such as formation of the penpheral nervous system and muscles, fusion of the anterior and posterior midgut, head involution and dorsal closure. They eventually result in creation of the first instar larva. which hatches from the egg case about 32-24 hours afler fertilization. All embryos are oriented with anterior to the left and dorsal to the top. (Adapted from Slack 1991) B) CLEAVAGE AND NUCLEI MIGRATION F) EARLY GERM BAND EXTENSION cephalic amniosenisa pole fuKow\ i

anteriormidgutt venual posterior rnidaur

G) ADVANCED GERM BAND ErnNSION blasmderm pole cclls vitellophagc nuclci

D) CELLULAR BLASM)DERM

H) GERM BAND RETRACTION

dorsal L O parasegment and the anterior region of the next one (Figure 1.1) (Mam'nez-Anas and Lawrence 1985). As a result, each segment, right after its establishment, is made up of two distinct compartments separated by a boundary of lineage restriction (Garcia-BeiIido et al. 1973; Martinez- Arias and Lawrence 1985).

1.1.3.1. ldentiticatioa of factors required for the formation and specification of the Drosophila anterior-posterior body sis

The genetic control of anterior-posterior axis formation involves the complex interplay of maternally provided and zyprically expressed information. Advances in our understanding of early Drosophila development have ken made largely due to the systematic application of genetic approaches. Using cuticular patteming of the larva as an assay, several genetic screens have been conducted (Nusslein-Volhard and Wieschaus 1980; Jurgens et al. 1984; Nusslein-Volhard et al. 1984: Wieschaus et al. 1984: Schupbach and Wieschaus 1986b; Pemmon et al. 1989). These screens resulted in the identification of both maternal and zygotic genes required to instruct cells as to their developmental fare along the anterior-posterior and dorsai-ventral axes. The maternal genes that control anterior-posterior patteming cm be grouped into three classes (se Figure 1.3A): those affecting development of the anterior region consisting of the head and thoracic segments (e-g. bicoid), thosr affecting development of the posterior region consisting of the abdominal segments (e-g. nanos), and those affecting the teminal regions consisting of the non-segmented acron and telson (e.g. rorso) (Figure 1.3A) (Nusslein-Volhard et al. 1987). The zygotic genes that control the anterior-posterior a..is also fall into three general classes: the gap gtmes, the pair-nile genes, and the segment polaricy genes (Figure 1.3B, C, D) (Nusslein-Volhard and Wieschaus 1980). The gap and pair-nile genes predominantiy encode transcription factors. Embryos with mutations in gap loci (e.g. mutations in Knrppel and knirps) are deleted for large areas of the normal cuticular pattern spanning several contiguous segments (Figure 1.36). In contrast pair-rule mutants display periodic aberrations with deletions of dternate segment-wide regions. Deletions in different pair-mle mutants are typically out of phase (Figure 1.3C). The third group, the segment polanantymutants, also displays a pattern of repetitive deietions (Figure 1.3D). However. in contrast to pair-rule mutants, segment polarity deletions occur within every segment. For many of these mutations, the deleted pattern is replaced with a duplication of remaïning regions. In addition to mutations affecting the number and polarity of Drosophila segments, there are also mutations affecting segmental identity. These mutations give rise to homeotic transformations in which the whole segment or structure is transfonned into the likeness of another one. Homeotic transformations are associated with the bnctions of homeotic genes. which are located in two gene clusters, the Anrennapedia complex (ANT-C) and the bithorax cornplex (BX-C).Historically, the homeotic loci were the first regional speçification genes to be discovered Mutations of the BX-C and II Figure 13 Types of segmental mutant phenotypes. The drawings indicate the regions deleted from the normal pattern in mutant larvae. Dotted regions repcesent denticle bands, dotted lines the segmental boundaries. The regions missing in mutant Iarvae are indicated by the hatched bars. The transverse lines connect corresponding regions in mutant and wild type larvae. SeveraI classes of mutations that affect the anterior-posterior patterning have been identified. They include mutations in materna1 genes (A), gap genes (B), pair-rule genes (C), and segment polarity genes (D). (Adapted from Nusslein-Volhard and Wieschaus 1980) A) MATERNAI, GENE MUTATIONS

bicoid mo group mutation) (pristerior ppmutation) (minal group mutation

B) GAP GENE MUTATIONS

K'"epel C) P-4IR-RULE GEiNE MUTATIONS

w-' 13 associated segmental transformations in the central and poserior parts of the embryo were originally described by Ed Lewis (Lewis 1978). Inspired by this pioneering work, Tom Kauthan's laboratory cm-ed out a screen for zygotic lethal mutations affecting the anterior portion of the embryo. This screen resulted in the identification of ANT-C loci (Lewis et al. 1980% Lewis et al. I980b).

1.1.3.2. Materna1 geaes

bicoid (bcd) is a matemal gene required for the development of the anterior portion of the ernbryo (see Figure 1.3A) (Frohnhofer and Nusslein-Volhard 1986). Its mRNA, localized to the anterior pole of the oocyte. is not translated until after fertilization. After translation, the bcd protein (BCD)difises from the anterior end and forms a concentration gradient along the anterior- posterior axis (see Figure 1.4) (Driever and Nusslein-Volhard 1988b; St- Sohnston et al. 1989). The bcd protein is a homeodomain-containing transcription factor. it acts as a morphogen. setting up the positional information in the anterior portion of the ernbryo. in addition to the BCD gradient, gradients of other matemal products affecting anterior- posterior patterning are established in the Drosophila embryo (see Figure 1.4). One of them is a gradient of hunchback (hb) protein (HB). hb is expressed from both the materna1 and zygotic genornes (Taua et al. 1987: Schroder et al. 1988). Matemal transcript is uniformly distributed and, when uniformly translated, the development of the posterior portion of the embryo is inhibited (Hulskamp et al. 1989: Struhl 1989). This is due to HB-mediated negative regulation of the posterior zygotic genes. To prevent this inhibition, the expression of HB in the posterior of the embryo is repressed by nanos (nos). nos mRNA is localized to the posterior pole of the ernbryo (Wang and Lehmann 1991). Translation of the nos mRNA and active transport of the nos protein (NOS) produces a protein gradient analogous to that of BCD, but in the opposite direction (Wang and Lehmann 199 [: Dahanukar and Wharton 1996). NOS acts along with uniformly dismbuted PumiIio (PUM)to block hb rnRNA translation in the posterior half of the embryo (Murata and Wharton 1995), thereby generating a matemal HB gradient overlapping that of BCD (see Figure 1.4). Instructive information for the development of the posterior portion of the embryo is provided by a gradient of the homeodomain-containing transcription factor Caudal (CAD) (Mlodzik et al. 1985). The siope of this gradient ascends towards the posterior pole of the embryo (see Figure I .4) (Mlodzik et al. 1985: Macdonald and Struhl 1986: Mlodzik and Gehring 1987). This gradient anses hmunifonnly distn'buted caudal (cad) maternai transcript as a resulc of a direct inhibition of CAD translation by BCD, which binds regdatory sequences within the 3'-UTR of the cad mRNA (Dubnau and Struhl 1996; Rivera-Pomar et al. 1996). Thus, BCD acts as both a repressor of translation and as a transcriptional reguiator (described in next sections). 14 Figure 1.4 Spatial expression of materna1 and gap gene proteins. In both panels the horizontal lines represent the embryo length (EL), with the left end corresponding to the anterior pole of the embryo (100% EL), and the right end to the posterior pole (0% EL). Coloured lines represent the expression of matemal proteins (top panel) or gap proteins (bottom panel). The BCD, matemal HB, and CAD morphogen gradients (top panel) assign positionai values aIong the anterior- posterior suis of the embryo. ïhese gradients control local expression of zygotic gap genes whose proteins have either bell-shaped or gradient dismbution (bottom panel). See text for full gene narnes. (Adapted fiom Pankratz and Jackle 1993) MATERNAL GENE PRODUCT GRADIENTS

100%EL 0% EL (anterior) (posterior)

GAP GENE PRODUCT GRADIENTS

100% EL O'TD EMS BTD (anterior) 16 Taken together, the activities of materna1 genes of the anterior and postenor systems establish three transcription factor gradients (those of BCD, HB and CAD) along the anterior- posterior avis of the embryo (Figure 1.4). Formation of these gradients is facilitated by the qncj2ial environment of the early embryo. These gradients provide the zygotic genome with the positional information required for further patterning of the Drosophila ernbryo. In addition to the anterior and posterior systerns, pattern formation along the anterior- posterior avis also includes the terminal system, which ovemles the activity of the two former systems at the very anterior and posterior ends of the embryo. The terminal system acts through a signal transduction pathway that originates from the outside of the egg and culminates with the local activation of two gap genes, huckebein (hkb) (Weigel et al. 1990: Bronner and Jackle 1991) and tailless (tll) (Pignoni et al. 1990). Both of these genes are expressed in and required for setting up the terminal regions of the rmbryo (Figure 1.4). Formation of head structures is dependent on the combination of these two genes and the anterior determinant bcd [Bronner and Jackle 1991; Pignoni et al. 1992).

1.1.3.3. Gap genes

Positional cues provided by materna1 products are interpreted by zygotic genes. The first group of zygotic genes chat directly responds CO materna1 transcription factors is the gap gene class. Gap genes have been classified into three groups (Pankratz and Jackle 1993). Antenor gap genes, required for the proper development of the head and the thotau, include hunchback (hb), orrhodenticle (ord), emps, spiricales (ems) and buttonhead (brà). knirps (hi)and giant (gz) belong to posterior gap genes. necessary for development of the abdominal se-ments: whereas huckebein (hkb) and tailless (fil)are considered terminal gap genes, required for the formation of the acron and the telson. gt also affects a region of the embryo that is controlled by bcd and could be considered an anterior gap gene as well (Pankratz and Jackle 1993). In addition to these three groups. there is also centrally acting Kruppel (Kr). Al1 of the gap genes encode transcription factors (Rosenberg et al, 1986: Tautr et al. 1987: Nauber et al. 1988; Dalton et al. 1989: Rothe et al. 1989; Finkelstein and Pemmon 1990; Pignoni et al, 1990; Capovilla et al. 1992; Walldorf and Gehring 1992)- Gap genes are expressed during the syncytial blastoderrn stage (Figure 1.X) in a series of distinct bmad domains along the anterior-posterior avis (Figure 1.4) (hippie et al. 1985; Dalton et al. l989; MohIer et al. 1989; Pignoni et al. 1990; Capovilla et ai. 1992). The positions of these dornains correspond to the location of defects caused by gap gene mutations. Gap gene expression is directly induced by _mdients of matemal transcription factors. In the anterior portion of the embyo, for emple, gap genes such as hkb, 111'. qgotic hb, gr, otd, ems, btd. and Kr are activated by BCD (Tautz 1988; Dalton et al. 1989; Driever and Nusslein-Vohard 1989: Finkelstein and Perrünon 1990; Eldon and Pirrotta 1991: Kraut and Levine 1991b; Pignoni et ai. 1992). An important question is how the continuous information provided by the concentration values of the BCD 17 gradient gets translated into discrete domains of zygotic gene expression. This apparently involves differential gap gene responses to different threshold concentrations of BCD. Analysis of the zygotic hb promoter has provided a clue as to how this might be accomplished in moiecular tenns (Driever and Nusslein-Voihard 1988b; Struhl et al. 1989). In particul- it has been found that the zygotic hb prornoter contains two types of BCD binding sites: high-affinity and low-affinity sites. Using a reporter gene assay, it has been demonstrated that transcriptional activation mediated by Iow-afinity binding sites takes place only in the more anterior portions of the embryo, where BCD concentration is high (Driever and Nusslein-Volhard 1988b; Struhl et al. 1989). In contrast, activation mediated by high-afinity binding sites also occurs in more central regions of the embryo, where BCD concentrations are lower. Thus, affïmity of BCD-binding sites present within the cis-regulatory region of a target gene could determine the threshold concentration of BCD at which target genes become activated. Although different binding-site aanities can generate distinct domains of zygotic gene expression, the BCD concentration gradient is shallow and it has been argued that additional mechanisms employing cooperativity are required to establish sharp bcrders of gap genc expression (Driever 1993). Synergism can be achieved through hornomeric interactions between BCD molecules and cooperative binding to regulated promoters (Ma et al. 1996: Yuan et al. 1996; Burz et al. 1998). [n addition, hetemmeric interactions may also be involved. It has been demonstrated, for example, that BCD synergizes with materna1 Hl3 to precisely position spatial activation of zygotic hb (Simpson-Brose et al. 1994). Further refinement of expression borders of zygotic hb or ocher gap genes is based on cross-regulatory interactions benveen these genes (Jackle et al. 1986: GauI and Jackle 1987: Gaul and Jackle 1989: Hulskamp et al. 1990; Gaul and Jackle 1991; Hoch et al. t991: Kraut and Levine 199 la: Steingrimsson et al. 199 1: Capovilla et al. 1992; Hoch et al. 1992: Struhl et al. 1992). Spatially restricted domains of gap gene expression. their CO-expression in different cornbinations. and levels of their expression (see Figure 1.1) provide positional information for further patterning of the Drosophila embryo. This information, in conjunction with that of materna1 _gadients. conuols the spatial and temporal expression of the next class of zygotic segmentation mes. the pair-rule genes.

1.1.3.4. Pair-rule genes

A characteristic feature of a pair-rule cuticular phenotype is the deletion of altemate segment-wide units (Figure 13C) (Nusslein-Volhard and Wieschaus 1980). This perïodic requirement for pair-mie gne functions is reflected in their spatial expression during ear1y embryogenesis. In most cases, pair-rule uanscripts are distributed in a pattern of regularly spaced stripes (Figure 1.5) (Hafen et al, 1984a; Ingham et al. 1985; Harding et al. 1986; Kilchhetr et al. 1986; Gergen and Butler 1988; Coulter et al. 1990: Grossniklaus et al. 1992). These expression patterns are the f~ 18 Figure 1.5 Expression of pair-rule and segment poiarity genes. A schematic representation of three consecutive parasegmental intervals is shown. Each parasegment is separated by vertical dashed lines and is designated by tevt and solid black lines (at the top of each panel). The row of circles at the top of each panel represents one row of nuclei through the parasegments along the anterior- posterior auis of blastoderm and gastntlating rmbryos. Spatial gene expression is presented in the form of boxes. During the cellular blastodenn stage (top panel) almost al1 pair-rule genes are expressed in seven smpes with a double parasegment periodicity. Spatial patterns of different pair- mle genes. although ofien overlapping, are typically out of phase with one another. Even-numbered parasegments are distinguished by expression offushi rurairi (shown in red), whereas odd-numbered ones express rven-skrjped (dark blue). At the end of cellular blastoderm and at the beginning of gastrulation (boaom panel) spatial expression of some pair-mle genes undergoes transition from 7 to 14 stripes (cg. run, odd, prd. and slp). Also during this time, expression of segment polarity genes is initiated. Spatial patterns of three segment polarity genes, rngruiled, wingless and gooseberry are shown. CELLOLAR BLASTODERM EXPRESSION OF PAIR-RULE GENES

I u I hairy (h) even-sicippeâ (ew) I I I I fushi tarazu (fb) I I I I I I odd-skipped (Ad)

2% iI I M. ..I piired (prd) 1 1 I I I sloppy prired (slp) odd-paired (opa)

EXPRESSION OF PAIR-RULE AND SEGMENT POLARITY GENES DURING GASTRULATION

runt (run) haiw (h) even-skipped (eve) fushi tarazu (ftz)

odd-skipped (odd)

poired (PW sloppy paired (slp)

odd-paired (opa)

engrailcd (en)

wingless (wg)

goosberry (gsb) 2 O sign of metameriration in the embryo. Since pair-mle transcription initiates at the end of the syncytial blastoderrn stage, it is accompanied by the process of cellularization (see Figure 1-33). Cell membranes progressing dom from the apical to the basal surface gradually inhibit €ree diffusion of pair-rule products. In some cases, free diffusion of pair-rule products is further prevented by subceltular locakation of their transcripts to the apical compartrnent (Davis and Ish-Horowicz 199 1: Francis-Lang et al. 1996). initiation of pair-rule transcription is directly regulated by the activities of materna! and gap genes. The analysis of pair-rule expression in matemal and gap mutant embryos suggested that the position of each pair-rule stripe is controlled by different combinations of matemal and gap regulators, which act through distinct smpe-specific cis-acting elements (Carroli and Scott 1986: Carroll et al. 1986b; Ingham et al. 1986: Mlodzik et al. 1987: Carroll and Vavra 1989: Hoopr et al. 1989). [ndeed. evidence for independent stripe-specific elements has been found. For example, it has been dernonstrated that different mutations in the regulatory region of the hairy (h) gene promoter result in the elimination of specific sets of individual stripes (Howard et al. 1988; Hooper et al. 1989). Additional evidence for stripe-specific regulatory elements has corne from transgenic experirnents showing that different cis-acting elemeiits from the h promoter as well as the rvrn- skipped (eve) promoter direct the expression of reporter genes in positions corresponding to individual h and eve stripe domains (Goto et al. 1989: Harding et al. 1989; Howard and Stmhf 1990: Pankratz et a!. 1990: Riddihough and Ish-Horowicz 1991: Stanojevic et al. 1991: Small et al. 1992; Small et al. 1993; Hartmann et al. 1994: Langeland et al. 1994; Small et al. 1996; La Rosee et al. 1997: Hader et al. 1998). At the genetic and molecular levels, the mechanisms çcnerating expression of h and rve stripes are sirniiar to thosc controlling gap gene expression. In both cases, a stripe is generated by shallow transcription factor gradients of matemal origin and comparatively sharp gradients formed by the zygotic gap proteins. To date, the regulation of rve stripe 3 is the best-characterized example of how pair-mle stripe expression is controlled (Frasch et al, 1987; SrnaIl et al. 1991: Smafl et al. 1992). Genetic studies revealed that activation of eve stripe 2 requires the synergistic action of BCD and HB (Small et al. 1991: Small et al. 1992; Simpson-Brose et al. 1994: Arnosti et al. 1996). The precise borders of this stripe are established through repression imposed by the bordering gap genes g and Kr (Small et al. 1991: Stanojevic et aI. 1991; SmaII et al, 1992). Analysis of the eve promoter in transgenic studies led to the isolation of a small regulatory element chat is suficient to mediate eve stripe 2 espression. Within this element, binding sites for BCD, HB,GT and KR have ken identified (Stanojevic et al. 1991). These binding sites partially overlap. These findings led to a relatively simple mode1 of how a single pair-mle stnpe is activated. According to this mode], direct binding of Hi3 and BCD to the eve snipe 2 element causes activation. This activation is prevented by repressing factors whose binding either excludes binding of the activators or blocks their activation potential (Srnall et al. 1991; Stanojevic et al. 1991; Small et al. 1992; Amosti et al. 1996). Other stipes of eve as well as those of h are thought to be regulated in a similar way (Pankratz et al. 1990 Langeland and 2 1 Carroll 1993; Lardelli and Ish-Horowicz 1993; Hamnann et al. 1994; Langeland et al. 1994; La Rosee et al. 1997; Hader et al. 1998). In addition to activation by localized activities of matemal or gap genrs, initiation of some pair-rule stripes also depends on the function of a ubiquitous activation system. For emple, the expression of nmf (mn) stripe 5 and eve suipes 3 and 7 has been shown to require the JAK-STAT signaling synem (Binari and Pemmon 1994; Hou et al. 1996; Yan et al. 1996: Harrison et al. 1998). This suggests that enhancers controlling expression of these saipes may be directly recognized by one or more Drosophila STAT proteins (Darnell et al. 1993). Despite the stripe-specific effect of the JAK pathway on run expression. deletional analysis of the mn promoter has failed to identify smaller autonomous modules (Klingler et al. 1996) that woutd independently control its expression in individual smpes the same way as they control expression of h and eve (Howard et al. 1988; Goto et al. 1989: Pankm et al, 1990: SmalI et al. 1991: Small et al. 1992). A failure to dissect the run promoter has ken attributed to the cornpiexicy of this promoter. It has been suggested that response elements required for expression of individual mn stripes could be widely dispersed and therefore impossible to be identified by deletional analysis (Kiingler et al. 1996). ïhe pair-rule genes eve, h and run were originally designated as "primary" pair-rule genes. This was based on the premise that initiation of their transcription is in a direct response to matemal and gap gene cues. and not to uther pair-de sene products (Carroll and Scott 1986; Howard and Ingharn 1986; Ingham and Gergen i988). It was also argued that. after the establishment of eve, h and mn expression. the products of these genes function as direct and principal sources of regulatory information required for genenting periodicity in the remaining pair-rule gene expression patterns (Ingham and Maninez-Arias 1986: Carroll 1990: Dearolf et al. L990). These other genes, exemplified by fushi tarcu v:) and odd-sRipped(0dd). were designated secondary pair-nile genes. This hierarchical subdivision within the pair-de cIass was supponed by initial genetic nidies. In particular, it was demonstrated that early expression of): is affected by mutations in the primary pair-n.de mes and that no altention in the early expression of h. eve or run was detected injiz mutant embryos (Carroll and Scott 1986; Howard and Ingharn 1986: Ingham and Gergen 1988). More recent genetic and molecuIar studies have led to the revision of the primarylsecond~ pair-rule gene paradihm (Gutjahr et al. 1993; Klingler and Gergen 1993: Yu and Pick 1995: Saulier-Le Drean et al. 1998). In particular. detaifed analysis offi- transcript expression demonstrated that individuai stripes do not initiate simultaneously. as might be expected ifJi= periodicity were established by the pre-existing periodic cues. Instead,fi,T smpes come up in a regional and sequential manner. implying that aperiodic cues play a major role in their establishment (Yu and Pick 1995). In addition. re-examination of fe transcription in embryos mutant for the primary pair-de genes has revealed that activities of the primary pair-rule genes do not affect initiation but rather later stages offi= expression (Yu and Pick 1995). Similar conclusions as to the role of the pkary pair-rule gens in the establishment of strïpe expression of the secondary ones have been reached for odd. In particular, odd transcription initiates 23 in a single stripe suggesting that odd responds directly to aperiodic eues set up by gap and matemal pmducts (Coulter et al. 1990). odd also appears to be requued for the initiation of the primary pair- mle genes (Saulier-Le Drean et al. 1998). Taken together, these results dernonstrate that the pair-mle gene hierarchy, as described, is incorrect and that initial regdation of al1 pair-nile genes may directly depend on the activities of the matemal and gap loci. These initial patterns are &en refined by cornplex cross-regulatory interactions among the pair-mle genes themselves (see Figure 1.6). Unlike the transient expression of materna1 and gap genes, expression of the pair-rule genes persists for an extendeci period of tirne. This prolonged expression, in some cases, depends on autoregulation (Hirorni et al. 1985; Hitomi and Gehring 1987; Harding et ai. 1989). At the end of cellularization, the patterns of sorne pair-rule genes undergo a transition from 7 to 14 stripes (Figure 1.5). This transition is predominantly based on cross-regulatory interactions among the pair-rule çenes (Baurngartner and Nol1 1990; Klingler and Gergen 1993). In the case of run. eve. odd and sloppy pnired (slp) the new stripes are generated between the previous ones, whereas in the case of paired @rd) the existing stripes are spIit in the middle. The prolonged and rapidly evolving spatial expression of the pair-rule genes accounts for the substantially larger and more complex promoters that control these genes as compared to those controlling gap gene expression. One of the roles of the pair-rule genes is to divide the embryo into 14 parasegments (Lawrence 1992). Two pair-rule genes. eve andfi. play a particularl important role in detemining the size and boundaries of these parasegments (Martinez-Anas and Lawrence 1985; DiNardo and O'Farrell 1987: Lawrence et al. 1987: Lawrence and Johnston 1989a; Kellerman et al. 1990). During cellularization of the blastoderm. the expression of these genes falls preciseiy within the parasegmental boundaries such that eve is expressed in odd-numbered parasegments and& is expressed in the even-numbered ones (Hafen et al. 1984a; Martinez-Arias and Lawrence 1985: Carroll and Scon 1986; Frasch and Levine 1987; Krause et al. 1988: Karr and Kornberg 1989: Lawtence and Johnston 1989b). The relative levels of eve andfi activity appear to determine the sizes of their respective parasegments by setting up the location of parasegmental borders (Lawrence and Johnston 1989a: KelIerrnan et al. 1990: Manoukian and Krause 1992: Hughes and Krause 2001). Once each parasegment is delimited. it behaves as an independent developmental unit, rernaining under the contro1 of a particular set of genes (Lawrence 1992). Pair-rule proteins, most of which encode transcription factors (Laughon and Scon 1984: Frïgeno et al. 1986: Harding et al. 1986: Macdonald et al. 1986: Gerpn and Butler 1988: Couher et al- 1990: Grossniklaus et al. 1992: Benedyk et al. 1994). activate nvo classes of genes that are required for Werdevelopment of the parasegments the segment polarity and homeotic genes. 2 3 Figure 1.6 Initiation of segment polarity geae expressiun. Expression of segment polar@ genes initiates at the end of cellular blastoderrn (stage 5) as a result of cornplex, predominantly negative gene-regulatory interactions mediated by pair-rule gene pmducts. A schematic representation of three consecutive parasegments is shown. The row of circIes represents a row of cells aiong the anterior-posterior mis. Spatial gene expression is ptesented in the fom af boxes. Positive gene-regulatory interactions are marked in green and negative interactions in red. Expression of the segment polariry genes tinpiled [en) and wingiess (wg) initiates in single ceIl rows on either side of the parasegmental border. ODD # PS EVEN # PS ODD # PS .w 2 5 1.1.3.5. Segment polarity genes

Segment polarity genes are required to position and maintain the parasegmental boundaries as well as to establish the final segmental boundaries of the larval epidermis (Lawrence and Sampedro 1993). These genes also pmvide Iarval segments with a well-defined antenor-posterior pattern in which the anterior region of each segment carries denticles, and the posterior region is naked (Ingham 1991; Klingensmith and Pemmon 1991; Hooper and Scott 1992; D~Nardoet al. 1994). This pattem is easily seen on the ventral epidennis of the larval abdomen (see Figure 1.1). Mutations in the segment polarity genes atTect the anterior-posterior polarity of the segments, producing mirror-image or tandem duplications of either anterior or posterior parts (Figure 1.3D) (Nusslein- Volhard and Wieschaus 1980: Pemmon and Mahowald 1987; Martizez Anas et al. 1988: Perrimon and Srnouse 1989). Each segment polarity gene is expressed in a pattem that is reiterated in every parasegment (see Figures 1.5 and 1.6). The exact spatial and temporal expression of each gene is unique. In some cases, this expression is restricted CO a narrow stripe of cells (DiNard0 et al. 1985; Komberg et ai. 1985: Baurngartner et al. 1987; Baker 1988; Hooper and Scott 1989; Nakano et ai. 1989; Mohler and Vani 1992), while in others it covers much or all of the parasegment (Riggleman et al. 1989; Bourouis et al. 1990: Eaton and Komberg 1990; Orenic et al. 1990: Preat et al. 1990). Transcription of engrailed (en). for example. is found in the anterior-most row of cells of each parasegment (Fjose et al. I985: Kornberg et al. 1985), and transcription of wingless (wg) is detected in a single cell-wide stripe at the posterior edge of each parasegment (Figure 1.5) (Baker 1988: van den Heuvel et al. 1989). The interface between wg stripes in one parasegment and en stripes in the following parasegment corresponds to the parasegmenta1 border (Figures 1.5 and 1.6). The selective activation of en and wg at the extremes of parasegments establishes polarity within these units. which persists throughout embryogenesis. Expression of both en and wg is induced by pair-rule proteins (Figure 1.6). Initial genetic studies exarnining en expression in various pair-rule mutant backgrounds demonstrated that aiternate en stripes are regulated by different combinations of pairde genes (DiNardo and O'Farrell 1987; Ingham et al. 1988). In particular.li,- and odd-paired (opa)activities are required for expression of even-numbered en stripes. while prd and eve establish the odd-nurnbered en stripes. These responses to two distinct sets of regdatory factors sugest that there are two distinct cis-acting promoter regions thac independently contd each set of en stripes. Although genetic studies such as these identified rrans-acting factors required for en regulation, it was not clear whether these factors play a direct or indirect role in en regulation. For example, eve acts positive[y on en, but in cuitured cells and in vitro, the eve protein (EVE)does not act as an activator but as a repressor (Biggin and Tjian 1989: Han et al. 1989; Johnson and Krasnow 1992: Han and Manley 1993b). This suggests that en regulation by rve could be indirect, occumng through repression of an en repressor. Such an indirect effect is supported by a kinetic assesment of en regulation by EVE (Manoukian and Krause 1992) as weIL as by genetic studies demonstrating that 2 6 eve represses at leas two negative regulators of en. These are encoded by the run and slp genes (Frasch and Levine 1987; Manoukian and Krause 1992; Fujioka et al. 1995). The spatial expression of run and slp abuts the antenor border of odd-numbered en suipes (see Figure 1.6) (Gmssniklaus et ai. 1992; Klingler and Gergen 1993; Manoukian and Krause 1993; Cadigan et al. 1994a; Cadigan et al. 1994b; Fujioka et al. 1995). mn and slp are negativeiy regulated by low concentrations of EVE during early stages of eve expression (~Manoukianand Krause 1992; Fujioka et al. 1995). At this tirne, eve protein has a characteristic bell-shaped distribution within each stipe (Frasch et al. 1987) and acts as a marphogen, capable of concentration-dependent gene regulation (Manoukian and Krause 1992; Fujioka et al. 1995). In contrast to slp and run, prd is repressed by only high EVE concentrations [Manoukian and Krause 1992: Fujioka et al. 1995). This allows for the partial overlap benveen eve and prd expression (Figure 1.6). Differential repression ofprd, slp and run by EVE subdivides the evr domain such that a spatially lirnited narrow prd domain which expresses neither run norsr'p is hrmed (Fujioka et al. 1995). In this domain, odd-numbered en stipes are activated. (DiNardo and O'FarreII 1987; lngham et al. 1988; Momssey et al. 1991). Apart from ngulating odd-numbered en stripes, rve is also required for the establishment OF even-numbered rn stripes (sec Figure 1.6) (Harding et al. 1986; Macdonaid et al. 1986). These stripes are also regulated by 3: (Howard and Ingham 1986; DiNard0 and OfFarrell 1987: Ingham et al. 1988) and odd (DiNardo and O'Farrell 1987; MuILen and DiNardo 1995; Saulier-Le Drean et al. 1998).jiz is a positive regulator of en, whereas odd acts as a negative regulator. During syncytial blastoderm and early cellularization, odd stripes coincide withjk expression (Figure 1.5) (Coulter et al. 1990; Manoukian and Krause 1991: Benedyk et al. 1994; Fujioka et al. 1995; Mannervik and Levine 1999). By gastnrlation, however, odd expression is repressed in the anterior most cells of each FTZ stripç (Figure 1.5) (Manoukian and Krause 1992; Mannervik and Levine 1999). It is these anterior- most FTZ cells, devoid of ODD, where en is activated (Lawrence et al. 1987; Manoukian and Krause 1993). rvr hzs been shown to genetically interact with odd such that, despite the complete lack of activation or even-numbered en stripes in embryos mutant for eve (Harding et al. 1986; Macdonald et al. 19861, these stripes are partiaily restored in eve, odd double mutant background (DiNardo and O'Farrell 1987). Based on these findings, a model bas been proposed according to which eve is required for the 'clearing' of odd from the anterior-most cells of eachjz stripe (Manoukian and Krause 1992), It has been suggested that repression of odd is mediated by minor or secondary suipes oleve. These sa'pes smt king expressed in even-numbered parasegments at the end of cellularization (see Figure 1.6) (Harding et al. 1986; Macdonald et al. 1986). The level ofeve in these stripes is relatively low. This causes differential repression of odd without affectingjlz expression, which is only repressed by higher EVE concentrations (Manoukian and buse 1993). Altematively, it has been argued that it is the early bel[-shaped dpes of EVE that are responsible for odd repression (Fujioka et al. 1995). Despite thcse differences, the model was confirmed by genetic studies demonstrating that in eve mutant embryos odd is not repressed in the anterior-most cells of each FTZ stript (Fujioka et ai. 1995). Recently: it has been reportai that repression of odd in the 2 7 even-nurnbered parasegments depends on an interaction between eve and a Drosophila homologue of RPD3 histone deacetylase (Mannervik and Levine 1999). This repression of odd also depends on the function of a pair-rule gene opa (Benedyk et al. 1994). In embryos mutant for opa. odd coexpression with j: persists, and a significant delay in activation of en in even-numbered parasegrnents is observed. ofia is also required for the timely activation of odd-numbered en stripes (Benedyk et al, 1994). As with induction of en. the precise activation of wg depends on combinatorial regulation by pair-rule gene pmducts. Initial genetic studies demonstrated that even-numbered wg stripes (located in even-numbered parasegrnents) are controlIed by fi and prd, whereas odd-numbered stripes are regulated by eve and opa (Ingham et al. 1988; Mullen and Daardo 1995). Bothjiz and eve act negatively. while prd and opa act positively (Ingham et al. 1988; Ish-Horowicz et al. 1989; Manoukian and Krause 1992). wg expression initiates in the posterior-most cells of each parasegment aflerfi: and eve stripes resolve towards the anterior edges (Figure 1.5). Sincejk and eve domains continue to narrow without wg spatial expression becoming progressively broader, there musc be additional regulators delimiting the domain of wg transcription. The antenor border of wg expression in even-numbered parasegments is established through negative regulation imposed by odd (Figure l .6) (Mullen and DiNardo 1995; Saulier-Le Drean et al. 1998). prd, genetically identified as a positive regulator of rvg (Ingham et al. 1988; CopeIand et al. 1996), negatively regulates odd. preventing its posterior expansion into the wg domain (Figure 1.6) (Mullen and DiNardo 1995). prd also positively regulates odd-numbered wg &pes. as their intensity decreases in embryos mutant for prd (Figure 1.6) (Copeland et al. 1996). Establishment of wg expression also depends on the function of the pair-rule gene opa (Ingham et al. 1988; Benedyk et al. 1994). In ernbryos homozygous for opa mutations, odd-numbered ivg stripes never initiate, whereas expression of some even-numbered stripes is severely delayed (Benedyk et al. i994). After their establishment, the immediate maintenance of rvg stripes depends on the pair-rule gene slp (see Figure 1.6) (Cadigan et aï. I994a: Cadigm et al. 1994b). slp was proposed to control wg expression directly, by activating wg, as well as indirectiy, by negatively regulating some wg repressors such as>:. eve and en (sec Figure t.6) (Grossniklaus et al. 1992; Cadigan et al. l994a: Cadigan et al. 1994b). As a result, sip function defines regions competent to express ivg (Cadigan et al. 1994b). The establishment of competence groups within each parasegment, capable of expressing either en or wg, appears critical for subsequent maintenance of parasegmental polarity and cell-cell signaling (Ingham et al. 199 1; Ingham and Martinez Arias 1992). Although expression patterns of wg and en initially depend upon the activities of the pair- rule genes. both genes continue to be expressed long atler the pair-mle products have disappeared fiom epidermal cells. Genetic studies demonstrated that the maintenance of en expression requires a funcuonai wg gene (DiNado et al. 1988: Martizez Arias et ai. 1988; Bejsovec and Martinez Arias 199 1; Heernskerk et al. 199 1); and, sirnilarly, expression of wg is not maintained in the absence of en activity (Martizez Anas et al. 1988; Bejsovec and Martinet Arias 1991). Since wg and en are 2 8 expressed in mutually exclusive domains (Figure IS), neighboring cells have to send signais to one another to reinforce their positional identities. It is important to emphasize that, ünlike the gap and pair-rule products, the segment polarity proteins act in a cellular rather than syncytial environment and their activities participate in intercellular communication. Segment polarity genes therefore code not only for transcription factors but also for proteins with diverse functions, capable of generating or transmitting a signal (Desplan et al. 1985; Baumgartner et al, L987; Rijsewijk et al. 1987; Hooper and Scott 1989; Nakano et al. 1989; Bourouis et al. 1990; Orenic et al. 1990; Preat et al. 1990; Riggleman et al. 1990; SiegFried et al. 1990). Accordingly, some of the segment polarity products act as secreted extracellular ligands, while others function as transmembrane receptors or cytoplasmic proteins relaying the signal from the plasma membrane to the nucleus. The intercellular communication between wg- and en-expressing cells is thought to involve most of the other segment polarity genes (Martizez Arias et al. 1988; lngharn 1991; Limbourg-Bouchon et al. I99I; Peifer et ai. 199 1; Siegfried et al. 1992). Intercellular signaling set up between wg- and en- expressing cells maintains the parasegmental borders. It was proposed by Meinhardt (Meinhardt 1986a) that such mutual interactions between the cells that fonn the boundaries contribute to their straightness and make them extrernely stable structures. After its establishment, a boundaq becornes a pattern element itseif, a source of positional information and a point of reference for funher pattern formation and growîh (Blair 1993; Diaz-Benjumea and Cohen 1993). Indeed, a failure in establishment of parasegmental borders leads to severe dismption in specification of positional identity within each parasegrnent (Nusslein-Volhard et al. 1984: Baker 1987). Prior to ceil differentiation, the transition fiom parasegmental to segmental units occurs. Hatfway through embryonic development, parasegmental boundaries becorne shallow: and, posterior to thern, vety deep grooves arise that demarcate the incipient segmental bounduies. This transition is reflected in the morphogenetic movements and patterns of gene expression associated with segmental structures (Martinez-Arias et al. 1987) suggesting that the distinguished character of segmental units is related to their emergence as the definitive anatomical and hnctional uni& of the mature animal.

1.1.3.6. Homeotic genes

Each segment has a unique identity that is already clearly visible in the characteristic pattern of denticles on the ventral surtàce of the lacva (see Figure 1.1). Specification of segmental identity is carried out by the homeotic genes. Mutations in these loci result ÜI homeosis; the transformation of individual or groups of segments into the likeness of others (Bateson 1894). For example, flies with the bithorm mutation have part of the haltere, a balancing organ on the thüd thoracic segment, transformed into part of a wing (Bridges and Morgan 19î3: Lewis I978), whereas flies with the 29 dominant Anrennapedia mutation have their antennae transforrned into legs (Kaufinan et al. 1980; Struhl 198 1b).

The Drosophila homeotic genes, collectively referred CO as HOM genes, are organized into hvo complexes: the bithorar and Anrennapedia complexes (BX-C and ANT-C, respectively). 60th complexes are located on the right am of the third chromosome (Lewis 1978; Kaufman et al. 1980). The BX-C comprises three genes, LrItrabithoraic (Ub-r),abdominal-A (abd-A), and Abdominal-B (Abd-B). These genes function to diversi@ parasegments 5-14 (Bender et al. 1983: Karch et al. 1985; Sanchez-Herrero et al. 1985; Tiong et al. 1985). In Iarvae lacking the whole of the BX-C, every parasegment fiom 5 to 13 deveiops in the sarne way and resembles parasegment 4. The identity of this parasegment cm therefore be considered as a default state upon which new identities are superimposed (Lewis 1978). The homeotic loci in the ANT-C include labial (lab), proboscipedia @b). Deformed (Dfn), Sa contbs reduced (Scr) and dnrennapedia (Anrp) (Lewis et al. 1980a; Lewis et al. 1980b: Wakimoto and Kaufman 1981; Merrill et al. 1987). The function of these genes is to control parasegmenral identity anterior to parasegment 5, Interestingly, the order in which the HOM genes are required along the anterior-posterior avis in the embryo is the same as the order of their location on the chromosome. This order is reflected in the spatial and temporal expression patterns of the HOM genes along the anterior- posterior avis during development. Ubx, for example, is on the proximal side of abd-A on the chromosome and is more anterior in its pattern of embryonic expression (Akam 1983; Akam and Martinez Arias 1985: Harding et al. 1985). Expression patterns of the HOM genes are usually restricted to the areas of the body where their function is required. The initial transcription domains of most HOM genes are parasegmentai. with different parasegments expressing distinct combinations of these genes. (Akam 1983: Levine et al. 1983: Akarn and Martinez Arias 1985; Beachy et al. 1985; Kuroiwa et al. 1985: Martinez Arias 1986; Casanova and White L987: Chadwick and McGinnis 1987: Martinez-Arias et al. 1987: Peifer et al. 1987). Ubx, for example. is expressed in al1 parasegments from 5 to 14 (Akam 1983: Akam and Martinez Arias 1985); abd-.4 is expressed more posteriorly. fiom parasegment 7 to 14 (Harding et ai. 1985); and Abd-B still more posteriorly, from parasegment IO to 14 (Beachy et al. 1985: Casanova and White 1987; Peifer et al. 1987). The spatial expression of the HOM genes is IargeIy detennined by the products of the gap and pair-rule genes (Ingham et al. 1986; lngharn and Martinez-Anas 1986: White and Lehmann 1986; Harding and Levine 1988: Jack et al. 1988; Irish et ai. 1989). For example, spatial expression of Ubx in a broad central zone dunng the Laie syncytial blastaderm (Akam and Martinez Anas 1985) is defined by repressive action of HB (Le. ù3x is expressed where HB is absent) (Irish et al. 1989; Qian et al. 199 1; Zhang et al. 199 1). Later, during cellular blastoderm, the transient pair-rule character of the L16x pattern suggem reylation by pair-nile genes. ïhese pair-rule features of Ubx expression disappear infi mutant embryos ([ngham and Martinez-Arias 1986), indicating thatfi is required for positive tegulation of übx (see below). This regulation by the pair-nile genes helps 3 0 narrow the domain of homeotic gene expression patterns and aligns them with the parasegmental borders (Muller and Bienz 1992). Control of HOM gene expression by the gap and pair-rule proteins is transient. AFter these proteins disappear, continued expression of the homeotic genes depends on autoregulation (Bienz and Trernml 1988; Kuziora and McGinnis 1988; Chouinard and Kaufman 199 1; Lamka et al. 1992; Tremml and Bienz 1992) and cross-regulatory interactions among the homeotic genes (Hafen et al. 1984b; Struhl and White 1985). In addition, the maintenance of their expression also depends on the activities of two additional groups of genes - the Polycomb group (PCC)(Lewis 1978; Struhl 1981a; Jurgens 1985; Paro 1990; Muller and Bienz 1991) and frithorar group (nxG) (Ingham 1985; Kennison and Tamkun 1988; Tamkun et al. 1992; Breen and Harte 1993). The proteins of the PcG rnaintain transcriptional repression of homeotic genes by converting their early transient repression into permanent silencing. There are many genes in the PcG. Their products probably form a complex that brings about a heterochromatic-like state, which is propagated through subsequent ceIl divisions. The acrivity of the PcG is opposed by the genes belonging to rmG, some memben of which were identified in genetic screens for PcG suppressors (Kennison and Tamkun 1988). The function of the rmG is to rnaintain expression in those cells where the HOM genes have been initially turned on. Since there are more parasegments than homeotic loci, differential spatial expression of the HOM genes and their combinatorial function are required to account for all the segmental morphological diversity. Al1 homeotic genes encode homeodornain-containing transcription factors (Levine and Hoey 1988) indicating that they speci@ segmentai identity by reguiating the expression of other genes. Although the task of identi-ing the target genes of HOM regulators has proved diftïcult, a number of their targets have been identified (GouId et al. 1990; Wagner-Bemholz et al. 1991; Gould and White 1992; Graba et al. 1992; Botas 1993). Some of these genes are themselves transcription factors (Vachon et al. 1992; Jones and McGinnis 1993) suggesting that the developmental effects of the HOM genes are propagated in some cases by an enended transcription factor cascade.

1.2. CHARACTERIZATION OF THE fh CENE

In my studies of the Drosophila segmentation hierarchy, 1 have focused on the regulatory circuitry controlled by the product of the pair-rule genejkshi tmew Cfi=). What folIows is an overview of thefi: gene.

12.1.lDENTIFICATIûN AND ISOLATION OF TBE* Cm

Thefi gene was identified ïndependently in two genetic screens. The first screen was carried out in the Kauhan laboratory (Kautinan et ai. 1980; Lewis et al- 1980a; Lewis et al. 1980b). Tt was designed to identiQ new horneotic loci in the ANT-C, The second screen, cartied out by Nusslein- Volhard and Wieschaus, was designed to uncover loci required for metarnerization of the Drosophila body plan (Nusslein-Volhard and Wieschaus 1980; Jurgens et al. 1984). Embryos homozygous for nul1 alleles ofjk die pnor to hatching (Wakimoto and Kauhan 1981). These embryos develop only half the number of segmentai units (Wakimoto et al. 1984). Based on this phenotype, the gene was narned '?khi tarazu" which means "not enough segments" in Japanese (Kuroiwa et al. 1984; Wakimoto et al. 1984). Deletions in#= mutants affect altemate parasegments, with even-numbered parasegments lost and odd-numbered parasegrnents expanded (see Figure 1.3C) (Wakimoto et al. 1984). Studies with a temperature sensitive (ts) allele oCftz dernonstrated thatjlz segmental activities are primarily required during the blastoderm stage (Wakimoto et al. 1984; Weiner et al. 1984). Thefiz gene has been analyzed molecuiarIy. The gene has a 1.9 kb transcription unit (Kuroiwa et al. 1984; Weiner et al. 1984). Thefiz transcript is spliced together from two exons that are separated by an intron of approximately 150 bp. The cloning of the gene was conhed by the finding thatfiz transcription unit is disrupted in two& mutations,/r~~~~at~dft?~' (Weiner et al. 1983). Isolation of theh gene revealed that a portion of the& transcript encodes a highly conserved DNA-binding homeodomain (Garber et ai. 1983; Scott et al. 1983; Kuroiwa et al. 1984: Laughon and Scott 1984; McGinnis et al. 1984b; Scott and Weiner 1984). This suggested thatjk encodes a transcription factor, capable of DNA binding in a sequence-specific manner (Laughon and Scott 1984).

1.2.2. SPATIAL AND TEMPORAL EXPRESSION OF fn

Nonhern blot analysis revealed that fiz expression peaks at 2-4 hours AEL @fier Egg. Laying) and persists weakly untiI 10-12 hours AEL (Kuroiwa et al. 1984; Weiner et al. 1984). This correlates with the functionai requirement detemined by use of the rs allele (Wakimoto et al. 1984; Weiner et al. 1984). The spatial distribution of fe transcripts was first estabiished by in situ hybridization to fked embryonic sections (Hafen et al. I984a; Weir and Kornberg 1985) and then to whole-mount embryos (Yu and Pick 1995),$z expression in differenr stages of embryogenesis is shown in Figure 1.7. The rapid evolution offi: stripes is due to the very dynamic transcriptional regulation offi as well as to the high instability of thefrz mRNA (Edgar et al. 1986; Edgar et al. 1987; Riedl and Jacobs- Lorena 1996). The stability of thefi= mRNA decreases as a function of time &er fectilization. The estimated half-life varies fiom 14 minutes whenjz is first expressed to 6 minutes when the srripes have completely formed (Edgar et a[. I986). Duringfi,- expression,fiz uanscnpts accumulate exclusiveIy in the apical periplasm 3 2 Figure 1.7 Evolution offtt stripe expression. Jz transcription, visualized by in situ hybridization, is show at different stages of embryogenesis.fi transcription initiates pnor to the beginning of cellularizatian in a broad region extendhg throughout rnost of the posterior two-thirds of the embryo (A). Expression then increases non-unifomly resulting in the establishment of two diffuse bands (B). At the onset of cellularkation, the levels of transcripts in future stripe 1 and 2 increase further (C). Stripe 5 also becomes more defined. and the beginnings of stripe 3 appear (C). At this stage, the low levers of transcnpt that were previously detectable throughout most of the ernbryo have faded to undetectable levels. Thereafter, stripe 3 sharpens and splits from stripe 2 (D). In addition, a new difise band emerges near the posterior end of the embryo. This is fused stripe 6 and 7, which progressively spliis (D). ïhe last stripe to emerge is stnpe 4 (E,F). As cellularization is close to cornpletion, transcript levels in each stipe increase further and the stripes sharpen considerably covering 3-4 cells in width, rnarking the position of even-numbered parasegments (G). At gasnulation, there is a rapid decrease in the width of stnpes 1-6 (H). Sharp anterior borders remain constant as the posterior edges resolve, eventually forming stnpes that are approximately I cell wide. The width and intensity of stripe 7 appears unchanged. Dunng germ band extension, levels of RNA in stripes 1-6 decrease (1) and are no longer detectable by the end of this stage (J). Stripe 7 persists a Iittie longer. (The pictures ofjz vanscript expression come from my unpublished data)

3 4 (penpheral cytoplasm) (Hafen et al. 1984a; Weir and Kornberg 1985; Davis and Ish-Horowicz 1991: Francis-Lang et al. 1996; Lall et al. 1999). This localization is pnrnarily under matemal genetic control (Francis-Lang et al. 1996). One of the factors required is the hnRNP-binding protein encoded by the squid gene (sqd) (Lall et al. 1999). Apical localization offi transcripts is microtubule- dependent and is probably based on cytopiasmic transport rather than vectorial nuclear export (LalI et al. 1999). The apical localization offi mRNA dun'ng cellularintion of the blastoderm (see Figure 12D) reduces mRNA and protein difision, and was proposed to ensure that the pmtein acts locally in the precise spatial domains defined by fe transcription, Likejk mRNA, thejk protein (FTZ) has a very short half-life (Edgar et al. 1987; Kellerman et al. 1990). Consequently. FTZ expression closeiy follows that of thefi transcript (Carroll and Scott 1985: Krause et al. 1988). Immunohistochemical labeling demonstrated thatfr= stripes are 4-5 nuclei wide during first half of cellularization (Carroll and Scott 1985; Krause and Gehring 1988: Krause et al. 1988). Each smipe has a bel[-shaped dismbution of protein with the borders being poorly defined, However, with the progression of cellularization. Ievels of the protein in the nuclei at the borders of the stripes become equivalent ta those in the middle, yielding sharp on-off borders. By the end of cellularization, stripes one thmugh six becorne narmwer. These changes involve clearing of the protein from posterior nuclei of FTZ stripes (Krause and Gehring 1988). In addition to the seven stripes of blastoderm expression. anti-fi immunostaininç also detected fi: protein in a subset of cells in every segment of the developing nervous sysrem (6-10 hours ML)(Carroll and Scott 1985: Krause and Gehring 1988; Krause et al. 1988) and non-penodic expression in the developing hindgut (12-14 hours EL)(Krause and Gehring 1988: Krause et al. 1988). In the nervous system.ft= is transiently expressed in sorne neural precursors and glial cells. and subsequently also in MPl. dMP2. aCC and RP2 neumns and more weakiy in their sister celIs vME, PCC and RPI (Doe et al. 1988).ft-- expression in the nervous system was also detected through the analysis of expression of thefi promoter/reporter gene fusion constructs (Hiromi et al. 1985: Doe et al. 1988).

1.2.3.1. Analysis of the frz promoter

The temporal and spatial pattern of& expression is controlled primarily at the level of transcript initiation. This is based on the finding that. when the& coding region was replaced by the lacZ ORF in promoter fusion experiments, the pattern of fLgaIactosidase activity obsemed in gem- Iine transfonned embryos was simihr to the naturat pattern ofjk expression (Hiromi et al. 1985). Thefi,- locus served as a mode1 for studying transcriptional contro1 of spatial gene expression. Thefi promoter is relatively compact (Hiromi et al. 1985). It has been demonstrated that 10 kb of the genomic sequence including 6 kb of the 5' sequence, 3 kb of codùig region and 2 kb of 3' sequence is able to rescuefi; mutants (Hiromi et ai. 1985; Hiirni and Gehring 1987). The 5' regdatory region 3 5 of 6 kb is apparently sufficient to reproducejk expression (Hiromi et al. 1985; Hiromi and Gehring 1987). Three independent cis-acting elements can be distinguished in the 5' region (see Figure 1.8) (Hiromi et al. 1985; Hiromi and Gehnng 1987). These elements are the zebra element (-0.67 kb to O kb), the neurogenic element (-2.5 kb to - 0.67 kb) and the upstream enhancer element (-6.1 kb to -3.4 kb) (Hiromi et al. 1985: Hiromi and Gehring 1987; Doe et al. 1988). The zebra element is sufficient to establish spariatly restricted fc-like stripe expression. It directs the expression of the lacZ reporter gene in seven stripes localized primarily in the mesoderm (Hiromi et al. 1985; Hiromi and Gehring 1987). Deletional studies of the zebra element demonstrated that this promoter unit is composed of multiple mialler ejements [Topol et al. 1987; Dearolf et al. 1989a; Dearolf et al. 1989b; Topol et al. 1991). Each of these elements confers either activation, repression or both activation and repression (Dearolf et al. 1989b; Topol et al. 1991). DNA sites mediating activation include elements responding to several types of general activators. Alone. these elements direct reporter gene expression in a continuous band of cells throughout the germ band. One such regulatory element confers activation in a posterior ta anterior concentration gradient in the posterior half of the embryo (Dearolf et al. 1989a: Dearolf et al. 1989b). DNA-response elements mediating repression include several different pair-nile repressor elements. These elements inhibit* transcription in the nuclei located in the interstripe regions offi= expression- Based on these findings, a rnodel of howfi smpe expression is generated has been proposed (Dearolf et al. 1989b: Dearolf et al. 1990). According to this model. the ft pattern is established as a result of global activation and subsequent localized repression. This model is supported by ubiquitous dismbution off: transcnprs and protein prior to cellularization (Hafen et a[. I984a: Weir and Kornberg 1985: Karr and Kornberg 1989). In addition, this model is also supported by the observation that injection of protein synthesis inhibiton into early embryos results in uniform expression offi RNA (Edgar et al. 1986). This suggests that the action of short-lived repressors pIays a major role in generatingjk stripes. The neurogenic element directs@ expression in the nervous system (Hiromi et al. 1985: Hiromi and Gehring 1987). Flies carrying thefi: gene without the neurogenic element do not survive to adulthood (Doe et al. 1988). suggesting thatji: function in the nervous system is potentially required for full viability. There is some concern. however, that deIetion of the neurogenic element removes uncharacterized stripe regulatol elements required for segmenta] expression that may have been missed in the initial promoter analysis. The upstream element enhances fi= stnped expression established by the zebra element (Hiromi et al. 1985: Hiromi and Gehring 1987: Pick et ai. 1990). [t may also play a rote infi stripe establishment (Hiromi and Gehring 1987). The upstrearn eiement is active in both orientations relative to and at vanous distances hmthe zebra element (Hiromi et al. 1985: Hiromi and Gehnng 1987). Moreover. it generates sevenjiz-like stnpes when linked to a heterologous promoter 3 6 Figure 1.8 Theh promoter subunits. The zebra element directs expression in a seven stnpe pattern: the neurogenic element directs expression in the nervous system; and the upstream enhancer contains elements required for&-dependent autoregulation. The regdatory elements distinguished within the upstream enhancer inchde the distal and proximal enhancers (shown as open boxes). The proximal enhancer was fimher subdivided into Prox A and Prox B elernents. In addition, 323 jPE and the Autoregulatory Element (A€) were also distinguished within the proximal enhancer. FTZ horneodomain-binding sites within the A€ elernent are show as triangIes. The kilobase pairs (kb) given are from the start of transcription. Thefi coding region is presented as a black box, the intron is shown in white.

3 8 (Hiromi and Gehring 1987). All octhese properties are common to regulatory units called enhancers (Sertling et al. 1985). The upstream element has no effect onfrz expression in the nervous system (Hiromi and Gehring 1987; Pick et al. 1990). Two independently acting enhancers have ken distinguished within the upstream ekrnent: the distal enhancer and the proximal enhancer (see Figure 1.8) (Pick et al. 1990). The distal enhancer directs expression of seven stripes in the mesoderm, whereas the proximal enhancer directs expression in snipes that span both ectodermal and mcsodermal primordia. Two units: ProxA and Pro* were distinguished within the proximal enhancer (sec Figure 1.8) (Pick et al. 1990). Pro.x.4 directs expression in seven mesodemal stn'pes, whereas fusion genes containing only ProxB were not expresscd during embryogenesis (Pick et al. 1990). However, Proxi3 is required to act in conjunction with Profi to direct expression in the ectodenn (Pick et al. 1990). Thus. the proximal enhancer consists of at lest two interacting repulatory regions. Biochemicai studies with staged D~osophilaembryo nuclear extracts identitied multiple protein binding sites in the proximal enhancer (Harison and Travers 1988; Han et al. 1993; Han et al. 1998). Many of them were identified in a regulatory fragment of 323 bp, referred ro as 323 PE Vi,-Proximal Enhancer) {Figure 1.8) (Han et al. 1993). 323 fPE directs embryonic expression of a lac2 reponer gene in a pattern that is identical tu tfiat of the Full-length proximaI enhancer (Han et al. 1993). Site-specific mutagenesis of protein binding sites within 323fPE revealed a high degree of redundancy among these sites (Han et al. 1998)- In addition. many of the binding sites interact with several different proteins, suggesting rhat there may also be redundancy among rrans-acting positive regdators OF theh gene (Han et al. 1993) (çee below). Such redundancy demonstrates thatji:

regulatory system is highly buffered at multiple Ievels. possibly CO ensure proper expression of thefi gene (Han et al. 1998; Yu et al. 1999). The expression of upstream elernentllacZ fusion genes in severt fi=-like smpes decreases to almost undetectable levels in embryos mutant forfi (Hiromi and Gehring 1987; Pick et al. 1990). This indicates that the action of the upstream element is dependent upon the presence of its own protein product (Hiromi and Gehring 1987). BiochemicaI studies revealed that the upstrearn element contains multiple binding sites for the purified FTZ DNA-binding homeodornain (Pick et al. 1990)- Several of these sites are located within the proximal enhancer and are part of the 430-bp autoregdatory eIement (A€) (se Figure 1.8). The AE was on'ginally identified in functional transgenic studies through deletional analysis of thefi pmmoter (Pick et al. 1990; Schier and Gehring 1992; Schier and Gehring I993a). The AE directs lac2 reporter gene expression in sevenfrz- like stripes in transgenic embryos (Schier and Gehnng 1992; Schier and Gehring 1993a). The AE maps to the same region within the proximal enhancer as the biochemically defined 323 fPE (Figure 1.8) (Han et al. 1993: Schier and Gehring 1993a). The AE is dightly larger by additionai 23-bp at the S'-terminus and 83-bp at the 3--teminus. There is also FTZ-independent striped expression of the upstream element suggesting that this elernent may play a role not only in strÏpe maintenance but also in &pe establishment (Hiromi 3 9 and Gehring 1987). For the latter role, the upstrearn elernent would have to directly respond to stripe-determining signals. It has been suggested that products of the gap genes might regulatefiz gene expression via interactions with the upstrem element (Hiromi and Gehring 1987). Striped expression of this element also suggests regulation by pair-rule genes. A detaited expression analysis of reporter genes under the control of individual fi: regulatory elements demonstrated that fusion genes containing both the zebra and upstream elements fldiacC) are expressed in a pattern distinct îiom those containing either element alone or a superimpasition of the çwo individual patterns (Yu and Pick 1995). The differences are quantitative as well as qualitative (Dearolf et al. 1989b: Yu and Pick 1995). Thus, fc stripes do not arise as a simple additive effect of the zebra and upstream elements but their establishment requires a synergistic interaction between these nvo elements. The molecular mechanisms underlying this interaction are presently unknown. Althoughfi-like stripes can be generated by a fusion gene containing both the zebra imd upstream elements. reporter gene expression directed by these elements does not mimic precisely endogenous fi= stripes (Yu and Pick 1995). For cxample. the order in which the smpes of the reporter gene are established differs fiom that of the endogenous fi= gene. These ditferences in stripe formation suggest that the precise expression of* is controlled by elements Iocated outside the 5-kb upstream promoter. lndeed. it has been reponed that rescue of thefi: phenotype by transfomant Iines decreases signitkantly upon deletion of 3' DNA (Hiromi et al. L983').

1.2.3.2. Trans-acting regulators offtz expression

Matemal-effect nenes and nan aenes As discussed earlier in this chapter. materna1 coordinate genes initiate segmental subdivision of the embryo. Loss of Function of these genes results in substantial developmental defects (Macdonald and S&ruhl 1986: Schupbach and Wieschaus 1986a; Driever and Nussfein-Volhard 1988~ Schupbach and Wieschaus 1989: Schupbach and Wieschaus 1991). Sincefi: Lies downstream of these genes in the segmentation hierarchy. fi= expression is affected by maternai-effect mutations (Carroll et ai. 1986b; Macdonald and Stnihl 1986; Frohnhofer and Nusslein-Volhard 198f; Mlodzik et al. 1987: Lehmann 1988). [t is not clear, however, which maternai genes regdatefi directly and which indirectly via the gap and other pair-rule genes. It has been suggested that the product of the caudal (cad) gene (CAD) is a direct posterior- specific activator offi transcription (Dearolf et al. I989a). The cad protein is a horneodomain- containing transcription factor (Mlodzik et al. 1985) expressed in a posterior-to-anterior gradient in Drosophila embryos (Macdonald and Srnihl 1986; Mlodzik and Gehring 1987).fi,- expression in cad mutant embryos is eliminated or greatly reduced in the posterior three stripes (Macdonald and Struhl 4 0 1986). Conversely, when cad is expressed ectopically under the control of a heat-shock promoter. the posteriorfi stripes are broadened (Mlodzik et al. 1990). Cis-acting elements that mediate CAD- dependent regulation have been identified in the zebra element (Dearolf et al. 1989a; Dearolf et al. 1989b). Consistent with the distribution of CAD, these elements direct the expression of reporter genes in a continuous posterior-to-anterior concentration gradient (Dearolf et al. 1989a; Dearolf et al. 1989b). CAD binds in vitro to multiple copies of a 5'-TTiATG-3' sequence located within these elements and activates transcription of fusion constructs containing copies of this sequence in transient CO-transfectionassays in Drosopiiila Schneider-2 cells (Dearolf et al. 1989a). Point mutations which disrupt the in vitro binding of CAD to its recognition sites also eliminate the posterior-specific expression in transfomed embryos and reduce the level of activation in co- transfection assays (Dearolf et al. 1989a). These findings suggest that the product of the cad gene functions as a direct regulator offi transcription. Mutations in the gap gene Kr. which is normally expressed in a domain spanning parasegments 3-10 (Knipple et al, 1985). reducefi expression to 4 or 5 stripes shified towards the middle of the embryo with a yap where jk stripe 4 normally occurs (Carroll and Scott 1986: hgham et al. 1986: Howard 1988). This demonstntes a positive role of Kr injz expression. In contrast. the remaining gap genes appear to have a negative effect onfi (Carroll and Scott 1986; Frasch and Levine 1987: Mahoney and Lengyel 1987; Kosman and Small 1997). For example, embryos homozygous for strong kni alleles e.xhibit it wide band of ectopicfi staining extending across the area where the third through sixth stripes offi normaily fonn (Carroll and Scott 1986: Frasch and Levine 1987). Consistent with this negative regulation. IocaI ectopic expression of the kni product in the anterior ponion of the embryo results in repression ofjz stripes 7 and 3 (Kosrnan and Smnll 1997). It is not yet known which of these effects are direct and which are indirect.

Rrrrulation of fi= bv eair-rule penes A characteristic feature of the Drosophila segmentation hierarchy are mutual gene-regulatory interactions among loci within the same class of segmentation genes. These interactions also occur arnong the pair-rule class. It has been demonstrated that mutations in haiv ih), runt (nm),even- skipped fet'e), slopp-v paired (slp), odd-skipped(odd) and odd-paired (opa) specificaIly affect the smped pattern offi (Carroll and Scott 1986; Harding et al. 1986; Howard and lngham 1986; Frasch et al. 1988: lngham and Gergen 1988; Carroll and Vavra 1989: Lawrence and Johnston 1989a; Benedyk et al. 1994; Cadigan et al. 1994b; Mullen and DiNardo 1995; Tsai and Gergn 1995; Yu and Pick 1995: Saulier-Le Drean et al- 1998). The h gene encodes a protein (H) belonging to the family of basic heliu-loophelix (bHLH) transcriptional regulators (Rushlow et al, 1989). in h mutant embryosfiz stripes broaden, demonstrating that h acts negativefy onje (Carroll and Scott 1986: Howard and Ingham 1986; Ingham and Gergen 1988; Carroi1 and Vavra 1989; Lawrence and Johnston 1989a; Tsai and Gergen 1995; Yu and Pick 1995). Consistent with negative reguiation offi, h is expressed in cells that are 4 1 out of phase with those expressingfl:, although some cells do express both h andfi at certain times (see Figure 1.5) ([ngham et al. 1985; Carroll et al. 1988). Negative regulation offi,. by h has also been demonstrated using ubiquitous expression of H, which leads to a great reduction injk expression (Ish-Horowicz and Pinchin 1987). This regulatory effect is detected soon afler ectopic H expression, suggesting that H repressesj= directly. Direct regulation offi by H is also suggested by the tinding that ectopic expression of a chirneric protein, in which H is hsed to the potent activation domain of VP16, leads to activation offi throughout the embryo (Jimenez et ai. 1996). It was initially proposed that the stripedjz panem is established by generaHy activated transcription followed by interstripe repression mediated by h (Carroll 1990; Dearolf et al. 1990). The order of appearance of fr,- stripes, however, is not inversely correlated with the order of appearance of haiv (h) stripes, as would be expected ifjz stripes were established through h repression (Yu and Pick 1995). Funhermore, the seven fi= stripes are correctly initiated in embryos carrying mutations in h (Yu and Pick 1995),frz expression in these embryos is not affected until the end of cellularization. This indicates that h is not invotved in the establishment offl: periodiciry but is required for precise spatial& expression during later stages (Howard and lngham 1986; ingham and Gergen 1988; Yu and Pick 1995). The exact mechanism by which h repressesfi transcription is not known. However. it appears that h needs to interact with corepressors to negatively regulate gene expression (Jirnenez et al. 1997: Nibu et al. 1998: Poortinga et al. 1998). Two candidate corepressors of H. Groucho (GRO) (Paroush et al. 1994: Fisher et al. 1996: Jimenez et al. 1997) and dCtBP (Nibu et al. 1998; Poortinga et al. 1998) have been identified. 50th factors bind the h protein. Repression offi appears to be GRO-dependent (Jimenez et al. 1997). Functionid studies of the fi prornoter elements revealed that the stripes generated byjk zebra eIementllacZ fusion genes are broadened in h mutant embryos, suggesting that the h protein acts through the zebra element (Hiromi and Gehring 1987). Further analysis revealed that this negative effect is mediated by a 32-bp sequence referred to as tDEl @z -Dual Element 1) (Topo1 et al. 1991; Tsai and Gergen 1995). However, fDEl does not contain matches to the H consensus binding sequence, and H does not bind this element in vitro (Tsai and Gergen 1995). It has been proposed that H affects the activity of the fDEl element through interactions with other proreins (Tsai and Gergen 1995). Candidates for these proteins are FTZ-FI and DHR39lFTZ-F2, both of which are nuclear receptors capable of binding to €ûE1 and regulating its expression in transient transfection assays (Ueda et al. 1990; Ayer et al. 1993; Ohno et al. 1994) (se below). Since FTZ-FI behaves as an activator and DHR39 acts as a repressor that interferes with FTZ-FI-dependent activation, it has been proposed that H regulates fDEl by differentially affecting the relative activities of FTZ-FI and DHR39 (Tsai and Gergen 1995). The runr (run) gene encodes a protein that contains a DNA-binding and protein dimerization motif referred to as the RUNT-domain (Kagoshima et al. 1993). At the blastoderrn stage, each ft: stripe overlaps with the posterior half of each mn stripe (see Figure 1.5) (Kania et al. 1990). ïhis phasing suggests that run may act as a positive regulator offi:. Indeed, embryos lacking nm tùnction 4 2 show reduced expression and premature elimination offi stripes (Cano11 and Scott 1986; Howard and Ingfiarn 1986; Ingharn and Gergen 1988; Lawrence and Johnston 1989a; Lawrence and Johnston 1989b; Tsai and Gergen 1995; Yu and Pick 1992).fiz stripes decay in a variable order in run embryos indicating that the requirement for run varies qualitatively for differentflz stripes. As with h, the effects of run onfk are not detected until the end of celiularization (Yu and Pick 1995), indicating that run is not involved in establishment offi,. stripes but rather in their refinement. Consistent with positive regulation ofjz by run, ubiquitous expression of run throughout the embryo inducesfi ectopicarly within the germband (Manoukian and Krause 1993; Tsai and Gergen 1994; Tsai and Gergen 1995). However, fi,f does not respond immediately to ectopic expression of the Runt protein (RUN), suggesting thatjz activation by run may be indirect (Manoukian and Krause 1993). Expression offi zebra element/lacZ fusion genes is also compromised in run mutant embryos, suggesting that run acts through the zebra element (Hiromi and Gehring 1987). Further analysis revealed that run, like h, acts through DE1 (Tsai and Gergen 1995). Since the DNA sequence of fDEl does not contain matches to RUN consensus binding sites, and RUN does not bind to this element in vitro (Tsai and Gergen 1995); it has been suggested that, as with H, this regulation is sornehow rnediated by FTZ-F1 and DHR39 (Tsai and Gergen 1995). rven-skipped (eve) encodes a homeodomain-containing transcription factor (EVE) (Harding et al. 1986; Macdonald et al. 1986). eve expression during the blastoderm stage occuis in odd- numbered parasegments and is therefore complementary to that offi (see Figure 1.5) (Frasch and Levine 1987). This suggests that EVE acts as a negative regulator offi. However, ectopic expression and genetic studies have revealed that regulation offi 5y EVE is complex. tndeed, ubiquitous expression of EVE activates Jz ectopically (Manoukian and Krause 1992)- This positive effect, however, is temporally limited to early embryogenesis. Consistent with this observation, embryos canying nul1 mutations of eve show reduced expression ofjz (Yu and Pick 1995).fr,. stripe I is the most sensitive to the loss of evr Function and is almost completely absent in embryos mutant for evr. This suggests that eve has a positive effect onfi expression. In later stages, ectopic expression of EVE repressesjii (Manoukian and Krause 1992). Both repression and activation ofjiz by EVE occur with rapid kinetics suggesting that both regulatory events are direct (Manoukian and Krause 1992). The negative effects of eve on& are also observed in embryos carrying ts alleles of eve (Frasch et a[. t988).fi stripes in these embryos are shifled anteriorly into the odd-numbered parasegments. The extent of the shift depends on the strength of the eve mutation such chat sbonger mutant alleles of me cause increased anterior progression offi,- strips. The rnechanism by which EVE regulates jz expression is not known. In vitra and tissue culture assays suggest that EVE rnay repressjz transcription by interfering with either TBP tùnction or TBP binding to the TATA element (Han and Manly 1993b: TenHarmsel et al. 1993; Austin and Biggin 1995; Um et aI. 1995; Li and Manley 1998). The odd-skpped (odd) gene encodes a zinc-fmger-containing transcription factor (ODD) (Coulter et al. [990). As with the regdation ofjz by eve, the regulation offi by odd is complex, 4 3 During early cellula~ization,odd am positively on f~.This is based on the observation that the intensity offi blastodem expression in odd mutant embryos is significantly decreased (Saulier-Le Drean et al. 1998). Also, ubiquitous ODD expression at this stage rapidly induces ecropicfi expression throughout the germband (Saulier-Le Drean et al. 1998). Consistent with positive regulation ofjz by ODD, the blastoderm expression pattern of odd precisely overlaps that offi (sec Figure 1.5) (Manoukian and Krause 1992; Fujioka et al, 1995). The regulatory effect of odd onfi= changes during the course of embryogenesis such that. from gaçtruIation on, ODD no longer activates but repressesjz (Mullen and DiNardo 1995; Saulier-Le Drean et ai. 1998). This negative regulation contributes IO thefi stripe resolution towards the anterior border of even-nurnbered parasegments (Mullen and DiNardo I995). At the same time, odd expression is cleaced out hmthe most-anterior cells of the even-numbered parasegments (Manoukian and buse 1992) and, in consequence, the expression domains of odd andJi= no longer overlap (see Figure 1.5). [t has been suggested that the ability of odd to act as both a positive and negntive regulator reflects its interactions with different cohctors (Sauliec-Le Drean sr al. I998). The sloppy paircd (slp) locus consists of two related genes, slpl and slpl (Grossniklaus et al. 1992). Both encode products beionging to the forkhead family of transcription factors (Weigei and Jackle 1990; Hacker et al. 1997; Kaufmann and Knochel 1996). In slp mutant (A34B) embtyos, the expression offi is ectopicaliy activated such chat an extra set of sevenji: suipes arises in the posteriar regions of even-numbered parasegments during germband extension stage (Cadigan et al. 1994b). Ectopic expression of slp results in repression ofjz expression (Cadigan et al. 1994a). These findings suggest that slp acts as a negative regulator offi expression. The product of the slpl gene (SLPl) has been show to bind to DNA in a sequence-specific fashion and CO inceract with thefi,' proximal enhmcer in a modified yeast No hybrid screen (Yu et al. 1999). One high afinity and several lower afinity SLPl binding sites were identified in thej: proximal enhancer (Yu et al. 1999). SLPI has also been shown to repress transcription via the hi& afinity binding site in yeast cells [Yu et al. 1999). These observations suggest that SLPI acts as a direct repressor off:. fi= expression is also affected by mutations in the pair-nile gene odd-paired (opa), which negatively regdates fi- expression (Benedyk et al. 1994). This regulation is stipe-specific, predominantly affecting fi= stripes 2 and 4. It is not clear whether these effects represent direct or indirect regulation (Benedy k et al. 1994). Finally, the pniduct of thefi: gene itseif is invdved in its own regulation. In embryos mutant for&, expression directed by thefrz promoter initiates but fails to reach fuli expression levels and then fades prematurely (Hiromi et al. 1985: Hiromi and Gehring 1987). Conversely, ubiquitous expression of lTZ from a heat shock inducible transgene (HSFTZ) activates the endogenous gene, expanding all& stripes by one or two ceils anteriorly (Ish-Horowicz et al. 1989; Fiapatrick et al. I992).& autoreguiation has been the subject of intense srudies and is discussed in more detail below, together with the regulatory properties of the)= protein. GAGA factor GAGA factor, a protein with proposed functions in chromatin remodeling, is encoded by the Triihurux-iike gene (Farkas et al. 1994; Tsukiyama et al, 1994). Mutations in this gene abolish expression offi (Bhat et al. L996), indicating that GAGA positiveiy regulatesfi expression. Analysis of thejz promoter has revealed several binding sites for GAGA (Tsukiyama et al. 1994; Tsukiyama et al. 1995). One of them is fAE3 Uiz btivation Element 1) in the zebra element (Topol et al. t 991). Deletion of the fAE3 site results in a notable loss of reporter gene expression in transforrned embryos (Topol et al. 1991). GAGA has also ban shown to act as a transcriptionai activator of the fi= gene in cell-fiee systems (Biggin and Tjian 1988; Soeller et al. 1988). This activation is detected on chromatin-assembIed remplaces but not on naked DNA, indicating that the role of GAGA injz regulation is to remodel chromatin structure in a positive way (Tsukiyama et al. r 994).

1.2.3.2.2. fb promoter binding proteins

FTZ-F I FTZ-FI is a member of the nucleac receptor superfamily that was first identified as a factor that binds to thejz promoter (Ueda et al. 1990: Lavorgna et al. 1991; Thummel 1995). Two FTZ- FI binding sites. referred to as fDE ICfc Duai Clement 1) and DE2 (1i- Dual Elernent 3,were originaIly identified in the zebra elernent (Ueda et al. 1990). Specific mutations (2- or ébase substitutions) in one of these sites (fDEI) eliminate FTZ-FI binding in vitro and lead to Ioss of zebra elemenr-directed stnped expression of reporter genes in deveioping ernbryos, particularIy in stripes 1, 2. 3, and 6 (Ueda et al. 1990). These observations suggest that FTZ-FI is a transcriptional activator of& that may have stripe-specific effecrs (Ueda et al. 1490). Loss of expression of the zebra element-lac2 construct in stripes 3 and 6 has aiso ken observed for weak na alleles (Hirorni and Gehring 1987). ïhis supports the idea that the effect of run onfk is mediated by FTï-Fi (Tsai and Gergen 1995) (see above). Three FTZ-FI binding sites, referred to as a repeated FANT module, were also identified within the 323 fPE portion of the proximai enhancer (Han et al. 1993; Han et al. 1998). FTZ-F1 binding ta 323 ff E was detected usinp in vitro assays (Han et al. i993) as well as a yeast one-hybrid screen (YU et al- 1999)- Two of these FTZ-F1 binding sites are evolutionarily conserved in the Drosophila hydei version of thefi* proximal enhancer (Schier and GeMng 1993a). Mutations in single FANT modules have no detectable effect on 323 PE activity in vivo (Han et al- 1998). However, mutations of al1 FANT repeats have a smng effect on lac2 fusion gene expression, demonsaating that these sites are functionally redundant (Han et al. 1998). These findings sugpest that FTZ-FI acts as a direct transcriptional activator of ftz expression. Consistent with this is the 45 expression profile of FTZ-F 1. In particular, FTZ-F I mRNA is expressed matemally and is detectable in a time window fiom O to 4 hr (Lavorgna et al. 1991; Ayer et al. 1993; Yu et al. 1997), which overlaps withfiz expression. FTZ-Fl bas been shown to activate transcription in a number of systerns (Ayer et al. 1993; Ohno et al. 1994). This inchdes activation of transcription via the FANT module or the full-length 323 fPE in yeast cells (Yu et al. 1997; Han et al. 1998). The FANT binding sites interact not only with FTZ-F1 but also with at Ieast three other nuclear factors (Han et al. 1998) (see below). Based on this finding it has been proposed that there is redundancy at the level of rrans-acting factors that regulatejk expression (Han et al, 1998). This suggests the possibility thatfi= enhancer-directed gene expression may survive the absence of one of these factors so long as the other activators are present. Analysis of mutations in the fi=-# gene is consistent with this postulated redundancy as fir strîpes can still form in the absence of functional jiz-fl pmtein (Guichet et al. 1997; Yu et al. 1997). However, it has beert observcd that levels offii expression may be reduced infrr-fl mutant embryos (Han et al. 1998). To elucidate the mechanism by which FTZ-€1 regulatesjz, in vitro transcription systems derived from Bombyx mori posterior silk gland cells and human HeLa cells were used (Li et al. 1994). Silkworrn Bombp mori has a FTZ-F 1 homologue tenned BmFTZ-F 1, which recognizes the sarne DNA sequences as FTZ-F1 (Ueda and Hirose 1990). The silkwonn system has an advantage in that it can provide large quantities of material for biochemical Fractionation studies (Ueda and Hirose 1990). The HeLa ceIl extract does not contain FTZ-F1-like activity and can therefore serve as a recipient in iùnctional complementation assays for active components derived from the Bombyx mori silk -gland extract. These studies led to the isofation of two polypeptides MBF l (Mediator of BmnZ-F 1 type 0 and MBF2 (Mediator o€Bmf.TZ-FI type 2) that fom a heterodimer and rnediate transcriptional activation of thejz promoter by BmFTZ-FI or FTZ-FI (Li et al. 1994). Neither MBFI. MBFZ, nor a combination of these pmteins binds DNA on its own. MBF-1 interacts with BmFTZ-FI (or FTZ-FI) and stabilizes complexes of these proteins on DNA (Li et al. 1994; Takemam et al. 1997). MBFl also makes direct contact with TATA-binding protein (TBP) (Li et al. 1994: Takemaru et al. 1997). MBF2, in tum, directly interacts with the p-subunit of TFtIA (Li et al. 1997). On the basis of these data, it has ben proposed chat MBFl and MBF2 form a heterodimer which serves as a bridge (a mediator) connecting BmFTZ-FI (or FTZ-FI) with the general transcription machinery (Li et ai. 1994; Li et al. 1997 Ilakernani et al. 1997). This would stabiiize the protein-DNA interactions, promote the formation of the pre-initiation complex at the core promoter, and lead to an induced Ievel of transcription by RNA potymerase II (Li et al. 1994; Li et al. 1997). Since both h and run have been irnplied to act through the sarne regulatory element (fDE1) as FTZ-FI (Tsai and Gergen 1995) (see above), it has been suggested that the opposite regulatory effects of h and run on& could involve dimption or stabilization, respectively. of the bridging structure between FTZ-FI and gnerai transcriptional machinery (Li et al. 1997). 4 6 DHR39lFTZ-FI 6 DHR39 (also called FTZ-FIP) encodes a member of the nuclear hormone receptor superfarnily that is closely related to FTZ-FI (Ayer et al. 1993; Ohno and Petkovich 1993; Ohno et al. 1994). DM9is expressed continuously throughout embryonic development (Ayer et al. 1993; Ohno and Petkovich 1993). ln vitro DNA-binding studies demonstrated that the DNA-binding specificity of DHR39 is very sirnifar to that of FTZ-FI and that both nuclear receptors recognize fDEl with almost identical afinities (Ayer et al. 1993; Ohno and Petkovich 1993; Ohno et al. 1994). ln addition, DHR39, like FTZ-FI, binds to the proximalfiz enhancer (Yu et al. 1999). This suggests that FTZ-FI and DHR39 could coregulatefrr transcription through the same regulatory sequences. Using in vitro and comsfection assays, it was investigated whether these nuclear receptors act alone or in combination, and whether they have synergistic or antagonistic effects on fDE1 (Ohno et al. 1994). The results octhese studies demonstrated that FTZ-FI and DHR39 bind as monomers to fDE1 and that. unlike FTZ-FI, which positively regulates transcription, DHR39 does not activate reporter gene eupression, presumabiy due to the formation of a transcriptionally inen complex (Ayer et al. 1993; Ohno et al. 1994; Yu et al. 1999). tt has also been demonstrated that, when these two nuclear receptors are coexpressed in tissue culture cells, transcriptional activation attributed to FTZ-FI alone is suppressed. These results are consistent with FTZ-FI and DHR39 having antagonistic effects on fi: regulation by competing for common regulatory elements (Ohno et al. 1994). Genetic support for the role of DHR39 infi,- regulation is presently lacking.

T- T- The rramtrack (tkk) gene (also referred to asfi:--) encodes a DNA-binding protein (TKK) belonging to the zinc finger family of transcription factors (Harrison and Travers 1990: Brown et al. 1991; Read and Manley 1992). This pmtein was originally identified on the basis of its interaction with two sites in thefi: zebra element (Harrison and Travers 1990: Brown et al. 1991). In total. eleven TTK binding sites have been identified in thefi regulatory region: five in the zebra element (Harrison and Travers 1990: Brown et al. 1991), one in the region between the proximal and distal enhancers (Hanison and Travers I990), and flve in the proximal enhancer (Harrison and Travers 1990; Han et al. 1993). Mutation of nvo TKK sites in the zebra element disrupts binding of this factor to DNA in vitro and in vivo gives rise to premature and ectopic embryonic expression of reporter genes under the control of the zebra dernent (Brown et al. 1991)- This suggests thatfiz requires active repression during initial nudeat division cycles and that TTK may function as a repressor offi= transcription dunng this tirne. Consistent with this suggestiont ttk expression is cornplementary to that of fk at this tirne and is detected both before and afierfi is actively transcribed (Harrison and Travers 1990; Brown et al. 1991; Read et al. 1992; Read and Manley 1992; Brown and Wu 1993). In addition. ectopic expression of TK within the temporal window offiz expression results in repression of ft~(Read et al, L992; Brown and Wu 1993), whereas a reduction in the dose of the endogenous TTK protein results in the prernature activation oFJ= expression (Pritchard and Schubiger 1996). One of the TTK binding sites Found in the zebra element is fDE2 @z oual EIement 2). This site has a dual character, mediating both activation and repression (Topo1 et al. 1991). Indeed, it has been demonsnated that, in addition to mediating the effects of TTK, this site may also mediate the effects of FTZ-F1 (Ueda et al. 1990). Three other TTK binding sites overlap FTZ-F1 binding sites. These sites are FANT modules within the proximal enhancer of thefi promoter (Han et al. 1993; Han et al. 1998; Yu et al. 1999). Thus, TTK and FTZ-F1 may compete for the same set of DNA- binding sites in vivo. Since 'lTK binding sites appear to mediate the effects of regdators with opposing transcriptional functions, mutation of these sites rnay have mixed resuIts. Indeed, mutation of al1 five TTK sites in 323 fPE abolishes in vivo activity of this eIement (Han et al. 199%).

ADF-1 ADF-1 was tira identified as a sequence-specific transcription factor that interacts with the distal promoter of the alcohol dehydrogenase gene (Adh) (Heberlein et al. 1985; Heberlein and Tjian 1988: Engfand et al. 1990; EngIand et al. 1992). ADF-I protein is not a member of any weil- characteiized class of transcription factors. Some homoiogy was found ro the DNA binding motif of MYB (England ct al. 1992). The Drosophila sronrivaIl gene may also be another potential member of this novel group of DNA binding proteins (Clark and McKearin 1996). ADF-I was shown to activate transcription in vitro and was postulateci to be an activator of a larger number of genes during development (Engiand et al. 1990). Lt has been demonsnared that AD€-1 also interacts with the FANT binding site in j23 fPE (Han et al. 1998). ADF-1 activates transcription in yeast cells via this site as well as via the full- length 323 fPE unit (Han et al. 1998). Consistent with a positive role in Ji- regulation. adf-2 is ubiquitously transcnbed during cellular blastoderm (England et al. 1992). As mention4 above. FTZ- F 1 has also been proposed to activatej: expupression by acting through the FlWT module (Yu et al. 1997). Thus. ADF-I and FTZ-FI may be redundant activators off!: expression. Altematively, FTZ- FI and ADF-1 may interact cooperatively to regdate fi= (Han et al. 1998). Since a mutant phenoype of the adf-1 gene has yet to be reported, the mie of ADF-1 in)z regulation is still speculative.

1.2.4. TEE FUSHI TARW PROTEiiï (FTZ)

The JE gene encodes a protein (FTZ) of 413 amino acids with a predicted molecular mass of 43 kDa and an apparent rnolecular mass of 63 kDa on SDS-PAGE gels [Figure 1.9) (Krause and Gehring 1988; Krause et al. 1988). FTZ contains a large number of senne, threonine, tyrosine and pmhe residues as weil as a large number ofacidic aad basic residues (Figure 19) 4 8 Figure 1.9& protein sequeace. Amino acids are presented as single letters. Letters for serine (S) and threonine (T) residues are red except for TL63 which is shown in purple. The green box indicates the LXXLL motif (the nuclear hormone receptor box). The blue box indicates the PEST sequence and the gray box the homeodomain. The nuclea. localization sequences are show in yellow. MATTNSQSHYSYADNMNMYNMYHPH 25 SLPPTYYDNSGSNAYYQNTSNYHSY 50 50 (Laughon and Scott 1984; Krause and Gehring 1988; Krause et al. 1988). The high content of proline residues is the most likely cause of the anomalous migration on SDS-PAGEgels. FTZ contains a nurnber of interesting domains and motifs (see Figure 1.9). One of them is the evolutionarily conserved DNA-binding homeodomain (Laughon and Scott 1984). Others are nuclear localization signals, PEST regions and N- and C-terminal transcriptional activation dornains (Krause et al. 1988; Fitzpatnck and Ingles 1989; Kellennan et al. 1990; Fitzpatrick et al. 1992).

1.2.1.1. The homeodomain

The homeodomain is a 6 1 arnino acid DNA-binding domain found in a large number of developmental regulatory proteins (McGinnis et al, 1984a; Scott and Weiner 1984). The horneodomain of FTZ is closely related to the horneodornains encoded by neighboring genes in rhe Antennaprdia complex. The in vitro binding propenies of the FTZ homeodomain have been systematically examined (Percival-Smith et al. 1990; Florence et al. 1991)- These studies found that the FTZ homeodomain exists as a monomer in solution, as determined by gel filtration, and binds as a monomer to multimerized FTZ binding sites arranged with a variety of spacings and orientations (Florence et al. 1991). 5'-TAATTGCT-3' is an optimal DNA binding site for the FTZ homeodomain. This was determined based on random selection of oligonuc1eotides (Florence et al. 1991). Systematic mutagenesis of single bases in this sequence did not give any higher affinity binding sites (Florence et al. 1991). The monomer equilibrium dissociation constant for binding of the FTZ horneodomain to its optimal binding site in the context of a 59-bp fragment is t.jxI0'" M (Florence et al. 199 1). Interestingly, the equilibrium dissociation constant for binding of the FTZ homeodomain to non-specific binding sites is 3x1W9M (Florence et al. 1991). This 100-fold difference between specific and non-specific affinity is relatively small and çiven the vast excess of non-specific over spccific binding sites in the penome, this difference may not be suficient to ensure specific in vivo DNA binding.

1.2.1.2. Nuclear localization signals

Nuclear localization sequences (NLSs) are required to target proteins to the nucleus following translation in the cytoplasm. TypicalIy, these motifs are rich in basic residues (Lacasse and Lefebvre 1993). Based on sequence homology, two putative nuclear Iocalization signals cm be distinguished in the N-terminal domain of FTZ and a third one within the N-terminal arm of the homeodomain (se Figure 1.9) (Copetand 1997).

1.2 .13. PEST regions 5 1 PEST sequences target proteins for degradation (Rechsteiner 1988; Rechsteiner and Rogers 1996)- The amino acid residues defining these sequences include proline (P), glutamate (E), senne (S), and threonine (T). As mentioned above, FTZ is rich in these residues. One of the consensus PEST sequences comprises residues 707-221 (see Figure 1.9); (Duncan 1986; Kellerman et al. I990). Several fi missense mutations have been identified in this region (Kellerman et al. 1990). Their phenotypes range fkom mild homeotic transformations of Al to A3 in adults to a more severe anti- fiz phenotype in larvae (Duncan 1986). An anti-fi-- phenotype, characterized by deletion of odd- numbered Cfz-&dependent) parasegrnents, is reciprocal to thefi,- phenotype (Struhl 1985; Ish- Horowicz and Gyurkovics 1988). Since the transformation of AI to A3 is similar to the Ultra- abdominal mutants of Ubx (Lewis 1978), the alleles carrying missense f~ mutations were named -Ultra-gbdominal-like or Ual (Duncan 1986). Analysis of theh protein in embryos carrying Ual alleles reveaied that stripes of the mutant protein fail to retract, remaining overly broad untir the end of stage 9 (KeIlerman et al. 1990). This indicates that the mutant protein has an abnormally long half-life. The persistence of thefi: protein in ectopic locations is most likely what causes the UaUanti-fi= phenotype. Indeed, the anti-fc phenotype was first described in embryos in which FTZ had been ectopically expressed by means of a heterologous promoter (Stnihl 1985).

1.2.4.4 Transcriptional activity of FTZ in vitro, in cultured Drosophila cells and in yeast

The homeodomain-containing FTZ protein performs its functions by regulating the expression of other genes. The transcriptional regulatory functions of FTZ have been investigated in various artificial systems (Jaynes and O'Farrell 1988; Fiupatrick and Ingles 1989; Han et al. 1989; Winslow et al. 1989: Fiapatnck et al. 1992; Peterson and Henkowitz 1992: Ananthan et al. 1993; Colgan et al. 1993). For example. experiments using a cell-free in vitro transcription assay dernonstrated that purifiedfc protein activates transcription by binding to homeodornain binding sites insened upstrearn of the TATA box of a heterologous promoter (Ohkuma et al. 1990). The extent of this activation has been correlated with the amount of thefi,- protein added to the transcription reaction. The transcriptional activity of FTZ was also analyzed in culntred Drosophila SZ cells (Jaynes and O'Farrell 1988; Han et al. 1989; Winslow et al. 1989; Fitzpamck et al. 1992; Ananthan et al- 1993; Coigan et al. 1993). [t has been demonstrated that FTZ acts as a transcriptional activator in these cells. activating al1 reporter constructs tested thus far. [nteresthgly, it has been found that cotransfection together with other homeodomain activators leads to synergistic activation of transcription (Han et al. 1989; Ananthan et al. 1993). For exampie, reporter gne activation by FTZ or Zerknullt (ZEN) is about 7- and 10-fold, respectively, whereas activation by both proteins together is over 900-fold (Han et al. 1989). PRD was also to hnction synergistically with FTZ and ZEN, and cotransfection of ail three genes gives about a 2500-foid increase in activation (Han et al. 1989). ïhese results suggest that synergistic fimtional interactions among FTZ and other factors 52 coutd play an important role during FTZ-dependent replation in the developing embryo. Using FTZ deletion consmcts, it was demonstrated that tbe N-terminal region of FTZ is crucial for the finctional interaction with PRD (Fitzpamck et al, 1992; Ananihan et al. 1993). Interestingly, the FTZ homeodomain was shown to be dispensable for this synergistic activation. Since FTZ with the deleted homeodomain (FTZAHD)cannot bind DNA (Ananthan et al, 1993), this finding suggested that protein-protein interactions could be crucial for targeting FTZ to regulated promoters. The cultured cell system was used to further characterize transcriptionai activation mediated by FTZ. First, deletional analysis was carried out to map FTZ transcriptional activation domains (Fitzpatrick et al. 1997; Ananthan et al. 1993). These studies found that. when bsed to a heterologous DNA-binding domain, the N-terminal 17 1 amino acid residues could activate transcription to simiiar levels as the full-length protein (Fitzpatnck et al. 1992). Transcriptional activation can aiso be induced by the C-terminal96 amino acid residues of FTZ (Fitzpamck et al. 1992). This C-terminal domain contains a potential gluramine-n'ch activation domain. lt has been demonstrated that in cultured cells the activity of this domain cm be blocked by overexpressing tntncated derivatives of TFIIB (Colgan et al. 1993). Since such overexpression does not apparently decrease the level of basal promoter activitv, it was concluded that mutant variants of TFIIB used in this experiment do not interfere with the assembty of the basal transcription pre-initiation cornplex but with the hnction of the activator. It was funher suggested that TFIIB directly interacts with the FTZ C-terminal activation domain. This was proposed based on the observation that an increase in the concentration of this domain can partially reverse the negative effects of murant TFIIB overexpression (Colgan et al, 1993). If TFIIB derivatives do not interact directly with the activation dornain of FTZ but instead titrate a required CO-activator.then an increase in the concentration of the FTZ C-terminal domain should nor overcome the inhibitory effect of these derivatives. FTZ transcriptional activity has also been investigated in yeast cells. Here. FTZ also acts as a transcriptions! activator (Fitzpatrick and ingres 1989). This activity depends on the presence of SWI 1 (Peterson and Herskowitz 1992). However, unlike in Drosophih cuItured cells. the deletional analysis of FTZ in yeast failed to define distinct activation domains. suggesting that the activation domains of FTZ may not have general but rather system-specific propenies (Finpaaick and Ingles 1989). The expression systems described above have been usehi as a convenient alternative to the living embryo for characterizing feanires of FTZ-dependent mscriptional replation. These findings, however. should be ueated with caution as these artifid systems are limited by a number of drawbacks. For e.uample, they Iack the hetemgenity which characterizes Drosaphila embryonic nudei, in which a varie5 of different truiscription factors as weII as chromatin or odier regdatory proteins work in concert to reguiate transcription. Ln addition, most studies using these artificial assays have been done witb regdatory regions that are not acmal FR-response elements in embryos. For example, one promoter that has been show to be responsive to FTZ in culture cells does not produce a FTZ-dependent pattern of expression in îhe ernixyo (Vincent et al. 1990). 1.2.4.5. FTZ function in the developing embryo

1.2.1.5.1. Developmental roles of FTZ

In the develop ing Drosophila embryo, FTZ is required to delimit parasegmentai boundaries which form between FTZ-expressing and FTZ-non-expressing cells. The formation of these boundaries is necessary for the establishment of developmental fields and gradients of positional information (Lawrence 1992; Lawrence and Struhl 1996). FTZ function is also required to render the parasegments with unique identities. Thus, during early Drosophila embryogenesis, FTZ coordinates two different and unrelated processes: the allocation of cells to parasegments and the cornmitment of cells to segment-specific cell fates. In addition, FTZ is also required in the CNS for specification of neuronal identities (Doe et al, 1988). For exampie. in the absence offi activity, the avon morphology of RP2 neurons is transformed toward that of their siblings. the RPl neurons (Doe et al. 1988). These developmental fiinctions of FTZ are performed through the regulation of specific classes of genes. In particular. FTZ-dependent regulation of segment polarity genes is crucial for the establishment of parasegmental borders, while regulation of the homeotic genes leads to specification of parasegmental identities. The following sections contain a description of genes regulated by FTZ.

1.2.1.5.2. ftz downstream genes

Regulation of uair-mle genes bv FTZ Genetic studies demonstrated that some of FTZ downstrearn genes belong to the pair-nile class of segmentation genes. Pair-rule genes regulated by FtZ include run and prd. The effect of FTZ an mn is not detected until gastrularion (Frasch and Levine 1987: Ingham and Gergen 1988; Kiingler and Gergen 1993). During this time, nm expression undergoes a transition from 7 to 14 snipes (Figure 1.5), and negative replation of nm by FTZ is apparently required for this transition (Klingler and Gergen 1993). It is not clear whether bis regulatory effect is direct or indirect. The deIay with which this effect occurs supports the notion that the regulation is indirect. As with nrn. a relatively normal seven-smpe prd pattern is established in blastodenn embryos mutant forfi,- (Baumgartner md Noll 1990). The blastoderm mipes of prd are six cells wide, spanning the even-numbered parasegments and therefore overlapping stripes offi= (see Figure 1.5) (Bopp et al. 1986: Frigeno et al. 1986; Kilchherr et al, 1986: Baumgartner and Nol1 1990). At the end of cellularization. expression in each prd smpe retracts hmthe middle two cells (Bopp et al. 1986; Frigerio et al. 1986; Kilchherr et al, 1986; Baumgaroier and Nol1 1990; Gutjahr et al. 1994). As a resul~14 two-ceil wide stripes ofprd are generated (see Figure 1.5). Each of these stripes overlaps a parasegmentai border. Regulation of prd by FI'Z is necessary for the transition fiom the 54 7- to 14- stripe pattern. In embryos mutant for fi=, each blastoderm stripe of prd resolves asyrnmetrically establishing a three-cell wide anterior stripe and a single-cell wide posterior smpe (Baumgartner and Nol1 1990). Thus,& appears to have two effects onprd in wild-type embryos:): negatively regulates prd in al1 anterior celis of the resuiting two-cell wide gaps and positively regulates prd in al1 anterior cells of the two-cell wide posterior stripes. It is not clear which regulatory effects are direct and which indirect.

Regulation of segment ~olaritvgenes bv FTZ jk controls the initiation of transcription of the segment polarity genes engrailed (en) and wingless (wg).Expression of rhese genes is first detected at the onset of gastrulation in single-ceIl wide stripes (see Figures 1.3 and 1.6). en initiates in the anterior-most cells of each parasegment, whereas wg in the posterior-most (DiNardo et al. 1985; Fjose et al. 1985; Komberg et al. 1985; Baker 1987). The correct expression of en and wg is crucial for the establishment of parasegmental bordersfi,. positively regulates even-numbered en stripes (see Figure 1.6) (Howard and Ingham 1986; DiNardo and O'Farrell 1987: lngham et al. 1988; Lawrence and Johnston 1989a). These stripes do not initiate infi,- mutant embryos and are ectopically expanded by I- cells anteriorly in embryos where FTZ is expressed ubiquitously by means of a heat-inducible hsp70 promoter (HSFTZ embryos) (Ish-Horowicz et al. 1989; Fitzpamck et al. 1992). Enhancers that mediate FTZ-dependent expression of en have been mapped to the upstream regions of the en promoter (DiNardo et al. 1988) as well as to the tirst en intron (Kassis 1990). Several lines of evidence suggest that FTZ may regulate en directly. In particular. in vitro binding studies found a number of FTZ binding sites in the FTZ-responsive regions of the m promoter (Desplan et al. 1988: Hoey and Levine 1988: Han et al. 1989: Florence et al. 1997). Some of these sites, either alone or within the context of larger Fragments of the en promoter. were capable of mediating fTZ-dependent activation in S2 cells (Jaynes and O'Farrell 1988; Han et al. 1989; Ananthan et al. 1993). In addition. FTZ binding sites in the tint en inuon were also capable of conferring en-Iike reporter gene expression in the developing embryos (Kassis 1990; Florence et al. 1997). Unlike en. wg is negatively regulated by FTZ. Initiation of wg expression in the posterior- most cells of even-numbered parasegments is detected ody afierfi expression has retracted fiom this cell (Ingham et al. 1988). In embryos mutant forfi=, wg expression initiates prematurely filling in the entire even-numbered parasegments (Ingham et al. 1988). This negative regulation of wg by FTZ has also been detected in HSFTZ embryos, where odd-numbered wg stripes are almost completely repressed and the intensity of even-numbered snipes is significantly reduced (Ish- Horowicz et al. 1989; Fitzpatrick et ai. 1992; Copeland et al. 1996).

Reeulation of homeotic genes bv FTZ Expression of the homeotic genes. nomaiiy required for specification of parasegmentai identity, is in part regulated by& For example,fiz modulates the expression of Sa Combs Reduced 5 5 (Scr). In late blastoderm embryos Sm is mngly expressed in parasegment 2 as well as more weakly in other even-numbered parasegments (Ingham and Martinez-Arias 1986; Martinez-Arias et al. 1987). It has been demonstrated that this initial Scr pattern is completely abolished in fc mutant embryos (Ingham and Martinez-Arias 1986; RiIey et ai. 1987), indicating thatfrz acts as a positive regulator of Sm. fi also acts as a positive replator of Aniennapedia (Anip). Transcription of .hp is driven from two independently regulated promoters, Pl and P2 (Laughon et al. 1986; Stroeher et al. 1986: Jorgensen and Garber 1987; Boulet and Scott 1988). Evidence has been presented that FTZ regulates P2 but not Pl (Ingham and Martinez-Arias 1986; Boulet and Scon 1988; Riley et al. 1991). Expression of Anrp from the P2 promoter is first detectable in late blastoderm embryos as a four- cell-wide peak in parasegrnent 4 (Boulet and Scon 1988; Riley et al. 1991). In the absence of):, there is no accumulation of dnrp transcripts from P3 in this parasegrnent (Ingham and Maninez- Arias 1986). The Uhbiihorax gene (Ub-r)is also regulated by FTZ. The expression of Ubx in blastodenn embryos is in the form of transient stipes in parasegments 6, 8, 10 and 12 (Akam and Martinez Arias 1985: Akam et al. 1985: Ingham and Martinez-Anas 1986). This pair-tule pattern is superimposed on a weaker. more ubiquitous pattern of expression. Analysis of Lrbx transcription in embryos mutant forfi revealed that Ubs cnnscripts do accumulate in these embryos but only show the ubiquitous pattern (Ingham and Martinez-Anas 1986: Martinet Arias and White 1988). This demonstrates thatjk is a positive regulator of Lrbx. Molecular analysis of the Ubx promoter revealed that Ubx expression is controlled by multiple cis-control elements (Muller and Bienz 199 1: Qian et al. 1991; Muller and Bienz 1992; Zhang and Bienz 1992: Qian et al. 1993: Chan et al. 1994a). Two elernents BRE and PBX. corresponding to the genetically defined b.r and pbx regulatory regions respectively, confer expression of reporter genes in the even-numbered parasegments 6. 8. 10 and 17 (Muller and Bienz 1991: Qian et al. 1991). The enhancer activities of these elernents have been mapped to short, autonomousIy actin~sequences, 500-1000 bp in length (Muller and Bienz 1991: Qian et al. 1991: Qian et al. 1993). Transgenic and genetic studies demonstrated that FTZ is the primary activator responsible for the mecameric pattern directed by the BRE enhancer. and that Hl3 and TLL are negative regularors restricting FTZ-dependent activation anteriorly and posterioriy, respectively (Qian et al. 1991). Subsequent in vitro studies identified several binding sites for FïZ (4 sites), HB (3 sites) and TLL (3 sites) within the BRE module (Qian et al, 1991: Qian et al. 1993). Deletion or mutation of FTZ binding sites signiticantly decreased the BRE expression, whereas deietion or mutation of HB and TLL binding sites abolished repression in the anterior half of the ernbryo and relaved the posterior boundary of BRE expression, respectively (Qian et aL 1991; Qian et al. 1993). It has been found that binding sites for FTï. Hi3 and TLL are closely ciustered, often overlapping extensively with one another (Qian et al. 1993). Based on this observation, it has been proposed that the repressors may act by preventing binding of FTZ to the Ubx promoter. In vitro snidies demonstnted that HB can indeed block the binding of FTZ as welI as displace FTZ protein 56 pre-bound to an overlapping site (Qian et al. 1993). In addition to competitive binding, repressors of Ubx could antagonize Fm-dependent activation by interfering with the function of FTZ bound to the Ubx promoter as some FTZ binding sites within the BRE enhancer do not overlap appreciably with the repressor binding sites. FTZ can also act through the PBX enhancer of Ubx (Muller and Bienz 1991: Muller and Bienz 1992). As with BRE element, activity of PBX outside the I/bx domain is suppressed by HB (Muller and Bienz 1991; Zhang and Bienz 1992). Also. some FTZ binding sites within PBX are adjacent or overlap binding sites for HB (Muller and Bienz 1992). This suggests that, as in the case of BRE, the activity of PBX may depend on the competitive binding of FTZ and HB to this element (Muller and Bienz 1992). Molecular analysis of the BRE and PBX enhancers suggests that FTZ is a direct transcriptional activator of Ubx (Muller and Bienz 1992). This is further supported by couansfection experiments demonstrating that FTZ can specifically activate Ubx expression in co- transfected S2 cells (Winslow et al. 1989). In contrast to the positive regulation of Scr, rlntp and Ubx, the regulatory effects of FTZ on Deformrd (Dfd) are both positive and negative. The expression of Dfd is first detectable at cellular blastoderm in a single circumferential stripe that is about six cells wide (Chadwick and McGinnis 1987; Martinez-Arias et al. 1987; Jack et al. 1988). Since the posterior boundary of this stripe lies at the posterior boundary of parasegment 1. the expression of Dfd and& at this stage is mutually exclusive. The posterior boundary of blastoderm Dfd expression expands postenorly by one or nvo cells in embryos mutant for fc (Jack et al. 1988). This indicates that FTZ acts as a negative regulator of Dfd. Later. dunng post-gastrulation stages, wild-type expression of Dfd expands into parasegrnent 2 and is found Iaterally in a region constituting the primordiurn of the posterior portion of the mauilliary segment. This expression is not detected in#: mutant embryos indicating that during germ-band extension, fr= acts as a positive reguIator of Dfd (Jack et al. 1988). It has been suggested that this late regulation could be mediated by en and thereby occur indirectly (Jack et al. 1988). The teushirr (tsh) gene is another homeotic gene regulated by FTZ. This gene differs from the conventional homeotic gene in that it is Iocated outside the HOX cluster (Fasano et al. 1991). In addition. rsh does not encode a horneodomain-containing regulator but a zinc finger protein (Fasano et al. 1991). The tsh sene is first expressed at rhe blastodenn stage in the form of a broad 16-cell-wide stripe. Durinp gastnilation, spatial expression of tsh widens and is detected throughout PS 3-13 (Fasano et al. 1991). It has been demonsuated that the postenor portion of the blastodenn tsh dornain is not detected injk mutant embryos (Core et al, 1997) indicating that FTZ positively regdates tsh. The positive regulation of tsh by FTZ was also detected in HSFTZ embryos, where ubiquitous expression of FTZ causes ectopic induction of tsh (Core et al. 1997). It has been suggested that the regulation of tsh by FTZ is mediated by ch-acting eIements located 3' to the tsh gene (Core et al. 1997). This comes hmthe observation that gasariation expression of a mutant tsh gene lacking a large 3' ck-reglatory region is on[y detected in odd-numbered parasegments as compared to the ubiquitous wild-type tsh expression throughout the gemband at this stage (Core et al. 1997). 57 Analysis of the 3'-regulatory region of tsh identitied a FiZ-dependent enhancer which is predominantly active in even-numbered parasegments (Core et al. 1997). This enhancer contains a nurnber of FTZ binding sites thal together wirh flanking srquences, are evolutionarily conserved. These FTZ bindinz sites are required for the function of the 2'-enhancer as their mutation abolishes activity of this element (Core et al. 1997). These studies suggest that FTZ directIy regulates the rsh gene via the 3'-enhancer.

FTZ autoregulation As mentioned in the previous sections, FTZ positively regulates its own expression (Hiromi and Gehrinç 1987). This autoregulation is mediated, in part, by an autoregulatory element (AE) located in the proximal enhancer of thefi= upstream promoter (see Figure 1.8) (Hiromi and Gehring 1987; Pick et al. 1990; Schier and Gehring 1992; Schier and Gehring 1993a). AE has been shown to conferfc-like smped expression on a lac2 reporter gene in transgenic embryos (Schier and Gehring 1997). Multiple binding sites for purified FTZ homeodomain have been identified within AE (Figure 1.8) (Pick et al. 1990). Transgenic experiments demonstrated that these sites are required for AE activity in vivo (Schier and Gehring 1992). Al1 of the sites seem to contribute to enhancer activity. although to different ewents. and no single site is smctly required. A second-site suppressor experiment was carried out to determine whether thefi protein directly activates transcription through its binding sites within AE (Schier and Gehring 1992). In this experiment. three out of the FTZ binding sites in the modified AE were changed to BCD homeodomain binding sites. This decreased the afinities of these sites for FTZ by 40-fold and significantly reduced FTZ-dependent stripe expression of AE. It was subsequently demonstrated that activity of the mutant AE couid be restored by engineering a compensating mutation within FTZ that changes residue 50 in helix 3 of the homeodomain from glutamine to lysine. This switches the DNA-binding specificity of FTZ into that of BCD (Schier and Gehrins 1997). The rpsults of this experiment suggest that AE is a bona fide target of FTZ. AE has been the subject of a sistematic deletional analysis (Schier and Gchring 1993a). This analysis demonstrated that AE does not contain separable subelements that act independently to direct expression in seven smpes, indicating that AE functions as a single autonomous unit. ïhis analysis also revealed that sequences encompassing FTZ in vitro binding sites are not sufficient for enhancer activity. and thar sites other than FTZ binding sites are required for activity of AE. ïhus, aIthough necessary, FTZ is not sufficient for autoregulation, and combinatorial interactions of FTZ with other transcription factors is required for full AE activiy. Several sequence motifs that are candidate etements conmbuting to the regulatory specificity of FTZ have ken identified (Schier and Gehring 1993a). Some of these motifs are consewed in homologous AE eIements fiom other Drosophila species. These motifs have also ken identified in other developmental genes. some of which are regulated by FTZ. For example. good marches to some of the evolutionady conserved AE sequences, which are likety to be binding sites for non-homeodomain cofactors of FTZ, are found in 5 8 the en intron that itself seems to be a target for FTZ binding and FTZ-dependent regulation (Kassis 1990; Florence et al. 1997) (see above). As with FTZ-binding sites, non-FTZ binding sites also appear to be redundantly involved in AE enhancer activity (Han et al. 1998). Based on sequence cornparison, it has been suggested that sorne of these evolutionarily conserved sequences rnight represent binding sites for FTZ-FI or Tramtrack (Schier and Gehring 1993a). As discussed above. both factors were identified as& promoter binding proteins in biochemical studies.

Reeulation of FTZ downstream eenes in the newous -stem The reguiatory propenies of FTZ in the developing nervaus system differ from those characterized during early embryogenesis. This in part manifests itseif in the differential regulation of FTZ downstream genes. For cxarnple, during early embryogenesis FTZ regdates en but nor eve (DiNardo and O'Farrell 1987; Frasch and Levine 1987: Ingham and Gergen 1988; ingham et al. 1988: Lawrence and Johnston 1989b), whereas later during neurogenesis FTZ is required for normal expression of rve but not expression of en (Doe et al. 1988). Sirnilarly. experiments with/r=/acZ fusion genes demonstrated that high levels offi expression in the blastoderm requires interaction of FTZ with the upstrearn enhancer element and that such an interaction does not take place injz- expressing cells in the CNS (Hiromi and Gehring 1987). Regulation of): downstream genes in the nervous system appears to be neuron-specific, This is based on the tinding that the Ubx and eve genes respond to loss of fc expression in different ways in different neurons, For exarnple. fk is required for the expression of Ubx and eve in RP2 neurons and Ubx in RPI neurons. whereasfiz is not required for the expression oieither gene in aCC and PCC neurons (Doe et al. 1988).

1.2.4.53. Horneodomuin-dependent und homeodomuin-independeni acrivities of FTZ

Studies of FTZ function in Drosophila embryos have defined rwo activities, homeodomain- dependent (HD-dependent) and horneodornain-independent (HD-independent) activities. Several lines of evidence demonstrated that the FTZ homeodomain is required for FTZ function in vivo. In particular. two hypomorphicfi alleles are mutations located within the horneodomain. A temperature-sensitive ailele,#?. is a single amino acid change at position 35 of the homeodomain. andfr?p' is a C-terminal deletion starting within the homeodornain (Laughon and Scott 1984). In addition. FTZ with most of the homeodornain deleted, when expressed hm& regulatory sequences, is incapable of rescuing thefi-- cuticular phenotype (Fumkubo-Tokunaga et al. 1992). Analysis of a series of point mutations within the FTZ homeodomain reveded that there is a mong correlation benveen rescue of the ft- phenotype and DNA binding affinity in vitro (Furukubo- Tokunaga et a[. 1992: Schier and Gehring L993b). Tbis suggests that the sequence-specific DNA- binding of FTZ is important for target gene recognition and regulation. fndeed. as described above. the DNA-binding specificity mutant FTZQSOK, in which GlnSO in the FTZ homeodomain was 5 9 substituted by Lys (Lys is found in a BCD-type homeodomain) only weakly activates an enhancer containing FTZ binding sites (Schier and Gehring 1992) and confers only partial FTZ activity to embryos mutant for the endogenousfi gene (Schier and Gehring l993b). In addition to HD-dependent activity, FTZ also possesses HD-independent activity. For exarnple, it has been demonstrated that the FTZ specificity mutant FTZQSOK, when expressed at suficiently high levels, can confer wild type FTZ activity tofiz mutant embryos (Schier and Gehring 1993b). This demonstrates that, under certain circumstances, DNA-binding mediated by the FTZ homeodomain is dispensable (Schier and Gehring 1991; Schier and Gehring 1993b). Similar conclusions came fiom the studies with a hybrid FTZ protein in which FTZ homeodomain was swapped with that of PRD (Morrissey et al, 1991). Transgenic embryos in which this chimeric protein was ectopicaily expressed developed segmenta1 defects similar to those caused by ubiquitous expression of wild-type fTZ protein. Analysis of gene expression in these embryos demonmated that the hybrid protein behaved as if its target specificity were both FTZ-like and PRD-like. For exampie, the chimeric protein retained the ability to activatejfz transcription but was also capable of regulating odd-nurnbered en stripes that normally are only reguiated by PRD. These findings indicate that the specific activity of the FTZ homeodomain is not absolutely required for some regulatory functions of FTZ. Indeed, embryos in which a homeodomain-deleted derivative of FTZ (FTZAHD) was ectopically expressed produced the sarne anti-jz phenotype that was originally detected in embryos throua which the full-length protein was expressed (Fitzpatrick et al. 1992). It is possible that FTZAHD may not regulate al1 FTZ target genes but only endogenousfi. If this is the case. then the anti-frz phenotype could be induced by FTZAHD indirectly, via endogenousfi. This was tested by examining the activity of ectopic ETZAHD infi mutant embryos (Copeland et al. 1996; Hyduk and Percival-Smith 1996). The results showed that, even in the absence of the endogenousfi gene, FTZAHD was capable of ioducing the anti-jz phenotype, This indicates that FTZAHD has the ability to regulate not only the endogenousfiz gene but also other genes whose ectopic regulation conmbutes to the anti-ftz phenotype. It is not clear why ETZAiiû, when expressed fiom theje regulatory sequences, cannot rescue thefi mutant phenotype. [t is possible that FTZAHD is not capable of carrying out al1 the regulatory hnctions norrnally executed by FTZ and that induction offiz-dependent cuticle by FTZ and FTZAHD in an ectopic scenario involves only a subset of FTZ wild-type regulatory interactions. It is also possible that, Iike autoregulation, al1 of the other regulatory interactions of FTZ can be carried out by FTZAHD but that FTZAHD is hlly functional only when expressed at the higher levels that are normally achieved following autoregulation (or fiom a heterologous prornoter). This might happen if FTZ functions as part of a multi-protein complex, At Iower levels the FTZ homeodomain-DNA interaction may provide contacts required to nabiIize the protein cornpiex on target gene promoten. At higher levels, contacts with other proteins, and theu abiIity to bind DNA, may be suficient to drive complex formation without the need for FTZ homeodomain-DNA 6 O contacts. According to this rnodel, FTZAHD can execute all of the regulatory functions of FTZ, except for the ability to act at low protein levels, which are typical for the initial phase of& expression. As a result, FTZAHD would not autoregulate eEciently and would fail tu increase its leveis. Imrnunocytochemistry carried out in embryos expressing FTZ proteins with defective homeodornains confirms that the levels of these proteins are indeed low. [nterestingly, mutational analysis of the)= autoregulatory element in embryos expressing either FTZ or FTZAHD revealed that HD-dependent and HD-independent activities recognize the same or closely overlapping cis- regdatory motifs (Hyduk and Percival-Smith 1996). This is consistent with the idea that FTZ functions as part of a rnultiprotein cornplex and that HD-independent activity arises as a consequence of such complex formation.

1.2.4.5.4. FTZ Cofactors

The ability of FTZAHD to specifically recognize and regulate FTZ target genes demonstrates that the DNA-binding of FTZ per se does not account for functional specificity in vivo (Schier and Gehring 1993b; Copeland et al. 1996; Hyduk and Percival-Smith 1996) and that this specificity depends on interactions with other factors. The dependence of FTZ regulatory functions on cofactors is also demonstrated by the finding that an artificiat enhancer element that only contains oligomerized consensus binding sequences of FTZ does not direct FTZ-dependent expression in vivo (Vincent et al. 1990). Other evidence that cofactors modulate FTZ regulatory activity comes fiom the observation that FTZ fails CO activate its own enhancer in the nervous system (Hiromi and Gehring 1987). This suggests that the activity of the upstream enhancer depends not only on FTZ but also on genn layer-specific transcription factors (Pick et al. 1990). These observations are consistent with the general notion that interactions with cofactors are required for specific target gene selection and regulation by homeodomain proteins (see below). Two FTZ cofactors have been identified. They are encoded by the pair-rule gene paired @rd) (Copeland et al. 1996) and the nuciear hormone receptor genejkgI (Guichet et al. 1997; Yu et al. 1997).

-PRD As mentioned above, experirnents in Drosophila cultured cells demonstrated that FTZ and PRD activate transcription synergistically from a promoter containing binding sites recognized by their homeodomains (Han et al. 1989: Ananthan et al. 1993). Deletional anaIysis showed that the FTZ homeodomain is dispensable for this activation and that synergism between FTZ and PRD depends on the N-terminal 171 amino acids of FTZ (Ananthan et al. 1993). A functional interaction berneen FTZ and PRD also takes place in developing embryos, where it is required, for example, for FTZ-dependent regulation of wg (Copeland et al. 1996). PRD is a positive regulator of wg. The expression patterns of these two genes overlap flngharn et al. 1988) and wg sûiped expression is severdy compromised in ernbryos mutant for prd (Copeland et al. 6 I 1996). Ectopic PRD expands ivg stripes by 1-2 cells aateriorly (Copeland et al. 1996). This is in contrast to the negative regulation of wg by FTZ (Ingham et al. 1988; Ish-Horowicz et al. 1989). FTZAHD is aIso capable of repressing wg (Fitzpamck et ai, 1992: Copeland et aI. 1996; Hyduk and Percival-Smith 1 996) indicating that the FTZ homeodomain is dispensable for FTZ-dependent regulation of wg. In vitro assays demonsûated that FTZ physically binds to PRD and that this binding depends on the N-terminal portion of thefrz protein containing amino acid residues 101-150 (Copeland et al. 1996). FTZ with amino acids 101-150 deleted (FTZAN)is no longer capable of repressing wg (Copeland et al. 1996). These results suggest that a direct interaction between FTZ and PRD is required for FTZdependent regulation of wg. Indeed. FTZ is incapable of repressing the residual expression of wg in embryos that are mutant forprd (Copeland et al. 1996).

FTZ-FI As mentioned above, the product of thefrz-fl gene is a member of the nuclear receptor gene superfamily (Lavorgna et al. 199 1) that was originally identified as a protein binding to thefiz promoter (Ueda et al. 199üj. Mutations compromising matemal expression offi-fl give rise to cuticfes in which even-numbered parasegments are deleted (Guichet et al. 1997; Yu et al. 1997). This phenotype is oIso produced by embtyos homozygous forftz mutations (Wakimoto and Kaufinan 198 1). Examination offi transcript and protein expression patterns reveakd thatftz expression is not noticeably disturbed in embryos mutant forfrz-fr (Guichet et al. 19971, indicating that thefi-jï phenotype does not arise as a consequence of reduced or abolishedjz expression, as initially expected. Since expression of FTZ target genes such as en or wg is aItered infi=-ji mutant embryos in much the meway as infi mutant embryos (Guichet et ai. 1997; Yu et al. 1997): it has been suggesred that FTZ-FI acts as a cofactor of FE.These two proteins rnay also regulatefi,. expression but in the contcxt of the whole fc promoter the contribution of FTZ-FI appears dispensable. Immunoprecipitation experiments using Drosophila ernbl extracts showed that FTZ and FTZ-FI are associated wirh each other in Drosophila embryos (Yu et al. 1997) and biochemical studies demonstrated that FTZ-FL binds directly to FTZ in vitro (Guichet et al. 1997; Schwartz et al. 200 1). This interaction depends on the FTZ N-terminus which contains an evolurionariiy conserved LRALLT motif (see Figure 1.9). This motif comprises a nuclear receptor box (LXXLL), which has been previousl identifieci in a particular cIass of coactivators of nuclear receptors (Heery et al. 199i). Men the FTZ nuciear receptor box is deleted, in vitro binding to FTZ-FI is severeiy comprornised (Schwartz et al. 2001). When expresseci in vivo under control cf thejz promoter, FTZ with a deleted nuclear receptor box (FI721 12-1 18) is unable to activate even-numbered en stripes. to repress odd-numbered wg stripes, or to rescuefc-dependent segments, despite the abiIity to nomally autoregdate its own expression (Schwartz et al. 1001)- The simiIariQ of this phenotype to that generated by loss offi=-fr (Guichet et al. 1997) strongly suggests that the nuclear receptor box mediates a direct in vivo interaction between FTZ and FTZ-FI. 62 Similar deletional analysis of the FTZ-FL protein identified the C-terminal AF2 Uctivation -Function 2) motif as a key domain required for an interaction with FTZ (Schwartz et al. 2001). This motif is an a-helk found at the C-terminus of the evolutionarily conserved ligand-binding domain (LBD)(Durand et al, 1994). AF2s have been demonstrated to provide essential contact surfaces for the binding of nuclear receptor box-containing coactivators (Darimont et al. 1998; Nolte et al. 1998; Shiau et al. 1998). This domain also seems to be required for interaction with FïZ in vitro and in vivo. This is based on the observation that a deletion of AF2 compromises binding to FTZ and fails to rescue thejk-fr mutant phenotype (Schwartz et al. 200 1). This rescue is however restored when the AF2 motif is added back or when it is substituted with highIy conserved AF2 domains frorn venebrate FTZ-F 1 homologues (Schwartz et al. 200 1). Both FTZ-F 1 and PRD are required for FTZ-dependent regulation of ivg (Copeland et al. 1996; Guichet et al. 1997). Since prd expression seems unaffected infi=-fl mutant embryos (Guichet et al. 1997), both PRD and FTZ-FI seem to act as cofactors of FTZ. This interaction could involve either simultaneous or competitive interactions arnong the three proteins. Although both PRD and FTZ-FI interact in vivo and in vitro with the N-terminal portion of FTZ, they seem to recognize ditierent domains within this region. This is based on the finding that although deletion of the nuclear receptor box within the FTZ protein has a drarnatic effect on FTZ-FI binding, binding to PRD is unaffected (Schwartz et al. ZOO 1).

How do cofactors affect FTZ function? The ability of FTZAHD to specifically recognize and regdate FTZ target genes (Fitzpatrick et al. 1992; Schier and Gehring I993b: Copeland et al. 1996; Hyduk and Percival-Smith 1996) suggests that a major mle of f'TZ cofactors is to ncruit FTZ to specific mget gene promoters. This contrasts with actions of cofactors that were proposed based on in vivo UV-crosslinking experiments (Walter et al. 1994; Biggin and McGinnis 1997). These experiments have €ound that FTZ and EVE are bound at similar relative levels to al1 DNA fragments tested, suggesting that these and other homeodomain proteins are distributed on chromatin in a widespread and equivalent manner. This distribution would be consistent with the in vitro binding properties of these proteins, which have been show to have broad and similar DNA-binding specificities (Desplan et al. 1988; Hoey and Levine 1988; Kalionis and O'Farrell 1993; Ekker et al. 1994). These similar in vivo DNA-binding properties cannot explain the unique developmental functions of each protein. Hence, it was proposed that cofactors mus specificaIIy modi@ the transcriptiona1 activities of these widely dispersed proteins (Biggin and McGinnis 1997). Each homeodomain pmtein would uniquely respond to a set of cofactors bound at each promoter. According to this model, FTZ-FI would not contribute to the recmiment of FTZ to regulated promoters, but would only rnodulate the cranscriptionai potential of FTZ- 1.2.4.5.5. Phosphotylation of FTZ

In addition to protein-protein interactions, FTZ function in developing embryos is also affected by phosphorylation, As mentioned above, FTZ contains a larse number of serine and threonine residues (see Figure 1.9), many of which are phosphorylated in vivo (Krause et al. 1988). The analysis of FTZ phospho-isoforms revealed an average of 8-9 and as many as 16-18 phosphates per molecule (Krause et al, 1988; Krause and Gehnng 1989). Interestingly it has been shown that the number and position of phosphates is stage- and tissue-specific (Krause and Gehnng 1989). This suggests that tissue-specific phosphorylation may play an important roIe in changing the regulatory properties and roles of FTZ during different stages of its expression. Phosphorylation of threonine 263 ('P,63), located in the N-terminal portion of the FTZ homeodomain (see Figure 1.9), has been studied in detail (Dong et al, 1998; Dong and Krause 1999). T263 is part of an evolutionari\y conserved motif rnatching PKA (CAMP-dependent protein kinase) recognition sites, It has been shown that T263 is phosphorylated in vitro in a CAMP-dependent fashion by Drosophila embryo eamcts and by punfied PU,To test whether T263 is also phosphorylated in vivo, and whether this phosphoryIation event is required for FTZ activity, T263 was mutated to Ala and Asp to rnimic the unphosphorylated and constitutively phosphorylated States. The mutant proteins were tested for their ability to rescue thefi phenotype. These studies showed that the Ala-substituted FTZ protein was unable to rescuepz mutant embryos but that Asp substituted protein worked well (Dong et al. 1998). These results strongly suggest that T263 is phosphorylated in vivo and that phosphorylation of this site is required for normal protein activity. Interestingly, phosphorylation of this site seems irrelevant fcr FTZ function in the CNS (Dong and Krause 1999). This is based on the observation that FTZT263A and FTZT363D were both capable of regulating FTZ target genes in this tissue. Moreover, both mutant proteins directed FTZ- dependent differentiation of RP2 neurons. This indicates that there are tissue specific requirements for FTZ phosphorylation. What molecular properties of FTZ are affected by T263 phosphorylation? Evidence has been presented that phosphorylation of this site neither affects protein stability nor subcellular localization (Dong et al. t998). Also, the presence of either Thr, Ala or Asp at position 263 makes no difference in the binding of FTZ to consensus binding sites in vitro or in Drosophila cultured ceils (Dong et al. 1998). In addition, T263A and ï263D mutations had no observable effect on the general transactivation activities of FTZ, as measured with previously tested reporter genes in SZ cells (Dong et al. 1998). Hence, it has been proposed that phosphorylation of T263 may affect protein-protein interactions with specific FTZ cofactos (Dong et al. 1998: Dong and Krause L999). These cofactors are untikely to be FTZ-Fi and PRD, since in vitro interactions with these proteins are not affected by T363 phosphorylation. 61 1.4. DETERMINANTS OF DEVELOPMENTAL AND RECULATORY SPECIFICITIES OF CIOMEODOMAIN PROTEINS

1.4.1. SEQUENCESPECIFZC DNA BINDING OF THE EOMEODOMAIN

nie first clues regarding the function of the horneodomain came Frorn comparative protein sequence analyses. In these studies, a mal1 but significant degree of homolo~was found between the homeodornain and the yeast rnating-type proteins MATal and MATa2 (Shepherd et al. 1984). The yeast proteins bind to regulatory sequences of target genes in a sequence-specific manner (Johnson and Herskowitz 1985) suggesting chat the homeodomain may alsa mediate sequence-specific DNA binding (Laughon and Scott 1984; Shepherd et al. 1984). This was further supported by weak homologies behveen the homeodornain and the helix-tum-helix motif of prokaryotic regulatory proteins (Laughon and Scott 1984; Pabo and Sauer 1984: Shepherd et al. 1984). In vitro DNA- binding studies dernonstrated that horneodomain proteins are indeed capable of sequence-specific DNA binding (Desplan et al. 1985; Fainsod et al. 1986; Beachy et aL 1988: Cho et al. 1988; Hoey and Levine 1988: Laughon et al. 1988; Muller et al. 1988; Odenwald et al. 1989) and that the isolated homedomains retain sequence-specific binding activity (Hoey et al, 1988; Mihara and Kaiser 1988; Muller et al. 1988: Percival-Smith et al. 1990; Ekker et al. 1991; Florence et al. 1991). These studies revealed that, unlike prokaryotic regulaton which bind as homodimers to their operator sites, homeodomain proteins generally bind to DNA as monomers (Affalter et al. 1990). The sequence 5'- TAATNN-3' was shown to be a high affinity binding consensus site for the rnajority of these proteins (Odenwald et al. 1989). The three-dimensional structure of the homeodomain. both free and bound to DNA, was determined in structural studies (Otting et ai. 1988; Qian et ai. 1989; Kissinger et al. 1990; Ottîng et al. 1990; Wolberger et al. 1991: Billeter et ai. 1993; Klernrn et al. 1994: Qian et al, 1994). These studies revealed that the homology between the homeodomain and the helix-turn-helix motif extends to their tertiary structures. A characteristic feature of the homeodomain structure is its overall

belobular conformation with a hydrophobic core containing highly conserved hydrophobic amino-acid side chains. The structuralIy well-defined portion includes 3 a-helices. There is also an N-terminal arm that is without secondary structure. Helices [ and II are separated by a 5-amino-acid loop. Helix III is arranged approximately perpendicular to helices 1 and II and lies in the major groove of bound

DNA molecules (roughly parallel CO the grove). Base-specific DNA contacts are established by helix III (the recognition helix) and by the N-terminal am, which contacts the minor groove on the other side. The two first base pairs of the 5'-TAATNN-3' consensus sequence are contacted by the N- terminal am, while the fast four base pairs are contacted by helix [II. X simple way to explain the different regulatory specificities of horneodomain proteins is to postulate differences in their abilities to recognize different sequences in target gene promoters. It appears that the amino acid residue at position 30 of the horneodomain is the principal source of 6 5 differential sequence-specific DNA binding amongst different homeodomains (Hanes and Brent 1989; Treisman et al. 1989; Kissinger et al. 1990; Oning et al. 1990: Percival-Smith et al. 1990; Hanes and Brent 199 1; Schier and Gchnng 1992). This residue is in heh III and recognizes the two bases immediately 3' to the TAAT core of the 5'-TAATNN-3' consensus binding site. In vitro studies demonstrated that proteins such as FTZ (Desplan et al. 1988; Pick et al. 1990), ZEN (Hoey and Levine 1988; Treisman et al. 19891, ANTP, Distai-less (DLL),EVE (Hoey and Levine 1988) and EN (Desplan et al. 1988; Hoey and Levine 19881, al1 of which have Gln at position 50, bind with high afinity to 5'-TAA'ITG-3' but only weakly to 5'-TAATCC-3'. In contrast, BCD (Driever and Nusslein-Volhard 1989) and Orthodenticle (OTD),both of which have Lys at position 50, specitically recognize 5'-TAATCC-3'. tt has also been demonstrated that Paired (PRD), which contains Ser at position 50, binds preferentiaily to 5'-TAATCG-3' (Treisman et al. 1993). Thus, the amino acid at position 50 seems to be highly predictive of binding specificity of the homeodomain. Inspection of structures of complexes between different homeodomains and oligonucleotides containing the TAAT core sequence found that the residue 50 indeed makes important contacts with bases in the major groove (Qian et al. 1989; Kissinger et al. 1990; Oning et al. 1990). Genetic studies camed out in yeast and in flies provided funher support for this conclusion (Hanes and Brent 1989; Schier and Gehring 1992).

1.4.1.1. DNA-binding specificity of tbe homeodomain is not sunicient to account for unique developmental functions of homeodomain proteins

Although different DNA-binding preferences could account for unique functional specificities of some homeodomain proteins. the homeodomain is highly consewed both between different species and within a given species, and it has been demonmted that multiple homeodomain proteins with different effects on development bind identical sites in vitro with similar affinities (Desplan et al. 1988; Hoey and Levine 1988; Kalionis and O'Farrell 1993; Ekker et al. 1994). In addition, some homeodomain proteins containing quite distantly related homeodomains have also been demonstrated CO recognize the same sequence elements in vitro (Desplan et al. 1988; Hoey and Levine 1988) and to regulate the same target genes in transient transfection assays (Jaynes and O'Farrell 1988: Thali et al. 1988; Han et al. 1989). Therefore, it seems unlikely that the DNA- binding specificity of the homeodomain is suficient to account for al1 of the distinct developmental tùnctions of homeodomain proteins. How then do homeodomain proteins with similar or even identical in vitro DNA-binding properties achieve their unique fbnctional specificities in vivo? This issue is particularly relevant to the HOX famiIy of homeodomain pmteins. This is a highly conserved family required for specifLing segmental identity and the body plan along the anterior-postenor auis of many animals (McGimis and Krumlauf 1992; Scott 1993; Burglin 1994; Kenyon 1994; Krumlauf 1994; Manak and Scott 1994). Most of these proteins are encoded by genes organized in gene complexes and aligned along the chromosome in the same order as they are 66 expressed along the anterior-posterior axis (Lewis 1978; Sanchez-Herrero et al. 1985; Acampora et al. 1989; Duboule and Dolle 1989; Graham et al. 1989; Kaufman et al. 1990; Duboule 1992; McGinnis and Knimlauf 1992; Clark et al. 1993; Wang et al. 1993). Genes from different clusters can be classified into subfmilies (paralog groups) on the basis of structural similarities in, and in some cases around, the homeodomain (Sharkey et al. 1997). Al1 HOX gene products contain a Gln residue at position 50 of the homeodomain (Q50 homeodomain proteins) and display very similar DNA- binding properties in vitro (Desplan et al. 1988; Hwy and Levine 1988; Kalionis and O'Farrell 1993; Ekker et al. 1991). In vivo, however, HOX proteins instruct unique developmental processes. How is the developmental specificity of HOX proteins attained? Since al1 HOX genes have unique spatial and temporal expression patterns (White and Wilcox 1985a; Celniker et al. 1989; Delorenzi and Bienz 1990; Karch et al. 1990; Macias et al. 1990), producing a homeodornain protein in sorne cells but not in others may have specific effects on morphology. This, however, could not be the only factor contributing to their functional specificity given that ectopic expression of different homeodomain proteins in the same group of cells leads to the specification of different developrnental pathways (Schneuwly et al. 1987; Kuziora and McGinnis 1988; Kuziora and McGinnis 1989; Gibson et al. 1990; Gonzalez-Reyes and Morata 1990; Mann and Hogness 1990). To address how the functional specificity of HOX proteins is achieved, chimeric HOX proteins with exchanged homeodomains andlor flanking regions were prepared, and their capacity to direct different developmental pathways were assessed in in vivo assays (Kuziora and McGinnis 198% Gibson et al. 1990; Kuziora and McGinnis 1990; Mann and Hogness 1990; Dessain et al. 1992; Lin and McGinnis 1992; Chan and Mann 1993; Furukubo-Tokunaga et al. 1993: Zeng et al. 1993). These studies demonstrated that both the homeodomain as well as regions outside the homeodomain contribute to functional specificities of these proteins, For e.uample, replacing the ANTP homeodomain with that of SCR resulted in a protein that was able to produce developmental transformations specific IO the full-length parental Scr protein (Gibson et al. 1990; Furukubo- Tokunaga et al. 1993; Zeng et al. 1993). Likewise, studies with mouse homologues of the Antp and Scr proteins, HOXb6 and HOXa5, also suggested that homeodornain sequences are suficient to specitj some ANTP and SCR functions in the embryo (Malicki et al. 1990; Zhao et al. 1993). The sarne was observed for DFD (Kuziora and McGinnis 1989; McGinnis et al. 1990; Lin and McGinnis 1992). In this case. ectopic expression in Drosophila of a human homologue of DFD, HOXJb, which is nearly identical within the homeodomain and virtually unrelated outside, resulted in the formation of head structures in the thorax of the Drosophila larva, a phenotype specific to ectopic expression of DFD (McGinnis et al, 1990). In another experiment, precisely substituting the UBX homeodornain with that of DFD resulted in ectopic induction of thoracic structures (UBX-specitic stnictures) instead of head structures (Kuziora and McGinnis 1989). In addition, DFD with the UBX homeodomain also regulates the Anrp gene (Lin and McGinnis 1992), a normal target of UBX and not DFD (Hafen et al. L981b). A similar series of swaps have ken made between the Ubx and Antp proteins (Mann and Hogness 1990: Chan and Mann 1993). As in the above cases. substituting the 67 ANTP homeodomain for the üEX homeodomain in [IBX successfilly switched the resultant chimeric protein from UBX to ANTP specificity. However, this effect only took place when a portion of the Ubx protein C-terminal to the homeodomain was deieted. In the presence of rhese residues, swapping the homeodomain from UBX to ANTP resulted in a protein that retained UBX specificity. This suggests that not only the homeodomain but also the regions outside this highly conserved domain rnay play an important role in determining the functional specificity of homeodomain proteins. Further mdies provided evidence that the N-terminal arm of the homeodomain is critical for determining the specific developrnental and regulatory effects of HOX proteins (Lin and McGinnis 1992; Furukubo-Tokunaga et al. 1993; Zeng et al. 1993). For example, it was demonstrated that six amino acids in the N-terminus of the homeodomain are both necessary and suficient to switch functions fiom Dfd to Ubx (Lin and McGinnis 1992). Four of these changes were found at positions identical to those suftïcient to swap dnrp and Scr activity (Zeng et al. 1993). Because the N-terminal m makes sequence-specific interactions in the minor groove, it is possible that the differences in this region of the homeodomain translate into slightly different binding preferences, which in turn contribute to different biological functions. In vitro binding studies demonstrated that slight differences among these different HOX proteins can indeed be obsemed (Ekker et al. 1992). It is not clear, however, to what extent these small differences are exploited in the embryo, Alternatively, the N-terminal arm of HOX homeodomains may conmbute to the functional speciticity of HOX proteins through establishing selective protein-protein interactions. Indeed, based on the homeodomain-DNA crystal structure. it can be infened that these specificity-determining amino-acid residues of the N-terminal arm are not in position to make contacts with DNA but could form a protein-protein interaction interface (Billeter et al, 1993). Similar conclusions came from a study in which evolutionarily conserved residues defining different HOX paralog groups were identified (Sharkey et al. 1997). Many of these residues are located on surfaces oriented away fiom DNA. This suggests that protein-protein interactions are important detenninants of functional specificity.

1.1.2. CONTRIBUTIONS OF COMBINATORLU, INTF;RACTIONS TO FUNCTIONAL SPECIFICiTIES OF HOMEODOMAiN PROTEINS

The view that interactions with cofactors are important determinants of developmental specificities of homeodomain ptoteins is supported by the structure of the response elements mediating their regulatory effects. It has been demonstrated that the consensus binding site of QSO homeodomain proteins, when muitimerized and piaced upstream of a heterologous promoter, is not sufficient to mediate their regulatory effects in developing embryos (Vincent et al. 1990). Typicdly, response elements of homeodomain proteins have complex organizarions reflecting combinatonal interactions with other proteins, For example, as discussed above, thefi,- autoregulatory element contains evolutionarily conserved binding sites for other transcription factors that are required for its 6 8 activity (Schier and Gehnng 1992; Schier and Gehring 1993a; Schier and Gehring 1993b). Similarly, the Dfd autoregulatory unit has been shown to consist of binding sites for different regulatory factors (Zeng et al. 1994: Gross and McGinnis 1996). Combinacorial interactions of homeodomain proteins with other factors are in part achieved through protein-protein interactions. The involvement of homeodomain proteins in protein-protein interactions is suggested by the presence in these proteins of highly conserved non-DNA-binding motifs outside of the homeodomain. These motifs include the YPWM motif (Chang et al. 1995; Knoepfler and Kamps 1995; Neuteboom et al. 1995; Phelan et al. 1995), the Paired octapeptide (Gehring et al. 1994a), the LIM domain (Freyd et al. 1990), the Engrailed homology dornains (Jaynes and O'Farrell 1991; Logan et al. 1992; Han and Manley 1993a; Smith and Jaynes I996), the CVC domain (Svendsen and McGhee 1995), the conserved peptide of NK-2 class homeodomain proteins (Price et al. 1992) and the LXXLL motif (Schwartz et al. 2001). A nurnber of factors that physically interact with homeodomain proteins have been identified. T'hese interactions contribute ro regulatory specificities of homeodomain proteins either by specificaIIy affecting their DNA- binding properties, by modulating their transcriptional potential or both. Several examptes of these interactions are described below.

1.4.2.1 MAT&, MATal and MCMl

One of the best-undentood interactions between a homeodomain protein and its cofrtctors is that involving the yeast protein MATa3. In a-type cells. MATa2 is required to repress the a- specific genes (Herskowitz 1989: Mak and Johnson 1993). This repression is mediated by specific operator sequences (the asg operator) to which MAT& binds with a mernber of the MADS box transcription factor family. MCMl (Hall and Johnson 1987; Keleher et al. 1988; Schwarz-Sommer et al. 1990; Johnson 1992; Smith and Johnson 1992). The physical interaction between MAT& and MCM l and the structure of the operator. which consists of an MCMI dimer binding site flanked by nvo MATcQ binding sites (Smith and Johnson 1992), allows for cooperative binding of the two proteins (Keleher et al. 1988: Keleher et al. 1989: Passmore et al. 1989: Smith and Johnson 1992). Since the DNA binding affinity of the heterotetrarnenc complex is much greater than tfiat of MATa2 to its monomeric binding site, MAT& preferentially occupies sites that are adjacent to those recognized by MCM1. This overcomes relatively poor seguence discrimination by MAT& monomers. In diptoid da cells. MAT& is aIso required for repression but of a different set of genes, the haploid-specifïc genes. Repression of these genes depends on an interaction between MAT& and MATal, another homeodomain-containing protein (Herskowitz 1989; DoIan and Fields 1991). The MAT& and MATal heterodimer specifically recognizes the hg operator iocated upsûeam of haploid-specific genes (Goutte and Johnson 1988; Dranginis 199% Goutte and Johnson 1993). Thus, 6 9 as in haploid a cells, regdatory targets of MAT& in diploid cells are also detennined through protein-protein interactions (Smith and Johnson 1992). Atthough the interaction of MATclZ with MCMl or Malis required for repression of MATa2 target genes, these complexes alone are not suficient to negatively regulate transcription- Repression of MATa2 target genes also depends on the interaction of MATa2 with the general tepression factor SSN6/TUP1 (Keleher et al. 1992; Komachi et al. 1994: Tzarnarias and Struhl 1994). interestingly, unlike MCMl and MATal, which both affect the DNA-binding specificity of MAT&, SSN6 and TUPI modulate the transcriptional potential of MATa2 such that MATcr2 becomes a transcriptional tepressor.

1.1.2.2. ROX and PBC proteins

Protein-protein interactions also play a major role in determining the distinct developmental specificities of HOX gene products. The HOX c~factorsthat are best characterized are encoded by the Drosophila e-rrradenricle (erd), C. elegans ceh-20 and vertebrate pre-B cell homeobox I @bx) genes (Mann and Chan 1996). ïhese genes encode a class of homedomain proteins collectively referred to as PBC proteins (Kamps et al. 1990; Nourse et al. 1990; Monica et al. 1991; Burglin and Ruvkun 1997; Flegel et al. 1993: Rauskolb et al. 1993: Burglin 1994). The senetic characterization of exd demonstrated that mutations in this gene Iead to a compound phenotype that incorporates diverse homeotic transformations, indicating that exd is important for HOX genes to execute their specific functions (Jurgens et al. 1984: Peifer and Wieschaus 1990). Since exd mutations do not alter the expression patterns of the relevant HOX genes (Peifer and Wieschaus 1990: Chan et al. I993b: Rauskolb and Wieschaus 1994). EXD could act either in parallel or downstream of the HOX genes. Given that e-rd regulates some of the same target genes that are regulated by the HOX genes (Chan et al. 1994b; Rauskolb and Wieschaus 1994) and that EXD as welI as PBX proteins bind cooperatively to DNA with HOX gne products (Chan et al. l994b: van Dijk and Mune 1994; Chang et al. 1995; Lu et al. 1995: Neuteboom et ai. 1995: Phelan et al. 1995: Popperl et al. 199): van Dijk et al. 1995). it was suggested that the exd and pbx genes encode HOX cofactors. In vitro binding studies and structural studies revealed how PBC and HOX proteins cooperate in binding to DNA (Chan et al. 1994b: van Dijl and Murre 1994; Chang et al. 1995: Knoepfler and Kamps 1995: Lu et al. 1995: Neuteboom et al. 1993; Phelan et al, 1993'; Popperl et al. 1995; van Dijk et al, 1993: Lu and Kamps 1996: Passner et al. 1999: Piper et al. 1999). According to these studies. the homedomains of HOX and PBC proteins bind to opposite sides of the DNA double hek ïheir cooperative binding depends on a conserved hexapeptide motif, located to the N- terminal side of HOX protein homeodomains. This hexapeptide is on the end of a long linker arm which enables it to reach around to EXD or PBX on the other side of the DNA molecule. The hexapeptïde inserts into a pocket that is created partiy by a tripeptide loop between heiices 1 and 2 70 of the EXDIPBX homeodomain (a feature characteristic of members of the TALE -&ee -mino acid 'oop extension- family of homeodomain proteins) (Burglin 1994; Burglin L997; Berthelsen et al. 1998). Since both homeodomains make sequence-specific contacts with DNA bases and backbone atoms, the binding afflnity and specificity of the complex is enhanced. Given that only a subsst of HOX proteins interacts with EXD/PBX (van Dok and Murre 1994; Chang et al. 1995), this cooperative binding could distinguish the DNA-binding specificity of some HOX proteins from others, contributing to their unique developmental finctions. Additional functional specificity may come fiom the ability of different HOWPBC complexes to selectively fom heterodimers that recognize subtle differences within DNA-binding sites (Chan et al. 1994b; Popperl et al. 1995; Chan and Mann 1996; Chang et al. 1996; Shen et al. 1996). In panicular, in vitro binding studies demonsmted that when the 5'-TGATïGATGG-3' binding site was used, UBX and Labial (LAB)both formed compIexes equally well with EXD; when 5'-TGATTTATGG-3' was used, complexes of WU)-UBX, but not EXD-LAB were fomed; and when 5'-TGATGGATGG-3' was used, complexes of EXD-LAB, but not EXD-UBX,were formed (Chan and Mann 1996). These studies also demonstrated that the specificity of a HOX monomer is different from its specificity as a heterodimer with PBC proteins. For example, on its own UBX has a strong preference for the sequence 5'-TAATGG-3' over 5'-TTATGG-3' (Ekker et al. 1991). However? as a heterodimer with EXD, both sequences are bound by UBX with high affinity (Chan and Mann 1996; Chang et al. 1996). The difference in these sequences lies in base pairs that are predicted to interact with the UBX N-terminal am. Thus, it appears that interaction with EXD does not simply increase HOX-binding affïnity but alters how HOX proteins contact DNA and uncovers cryptic DNA-binding specificities that are built into HOX homeodomains (especially their N- terminal arms). The evidence that these in vitro binding properties of PBC and HOX proteins indeed help HOX proteins differentially select correct binding sites in vivo cornes fiom the finding that a 20-bp oligonucleotide containing the LAB-specific and EXû binding sites generate [ab- and Hoxbl- (the mammalian ortholog of the labial gene) dependent expression patterns in Drosophile and mouse ernbryos, respectively (Popperl et al. 1995: Chan et al, 1996). It has been demonstrated that altering only two basepairs within this sequence changes its in vivo recognition by LAB and DFD (Chan et al. 1997). In addition, it has also been found that a PBX-HOXbl binding site is present within, and is necessary for the function of, an autoreguiatory enhancer of the murine Hoxbl gene (Popper1 et al- 1995). A similar sequence was found in an autoregutatory enhancer (lab.550) from the Drosophila lab gene (Grïeder et al. 1997). PBC-HOX binding sites have also been found in other naturai target genes of HOX proteins (Ryoo and Mann 1999). An example is a 37-bp element (fkh250) derived hmthe fork head @h) promoter, a natural target of Scr (Ryoo and Mann 1999). In vino, SCR and EXD bind cooperatively to fkh250 and activation of this element in vivo requires both Scr and exd. Oîher EXD-HOX heterodimers do not bind this element with high &nity in vitro, and do not activate this element in vivo. However by mutating two base pairs in this element, it can be converted from an 7 1 EXD-SCR binding site to an EXD-HOX consensus site that binds several different EXD-HOX heterodimers in vitro and that is regulated by severai different HOX gene products in vivo (Ryoo and Mann 1999). This dernonstrates that subtly different PBC-HOX binding sites have the potential to direct distinct HOX-dependent transcriptional activities in vivo, supporting the idea that PBC proteins selectively enhance the in vivo DNA-binding specificities of HOX proteins. In addition to its role in confemng additional DNA-binding sdectivity to HOX proteins, EXD was also proposed to regulate the transcriptiona1 activities of these proteins through changing, at lest some of the tirne, their regulatory sign - frorn repressors to activators (Biggin and McGinnis 1997: Pinsonneault et al. t997). This is based on the observation that rnost EXD- and HOX- regulated enhancers acrivate transcription and, converseIy, enhancers repressed by HOX proteins appear to do so without .md (Pinsonneault et al. 1997). Additional support for HOX transcriptional activity regulation by EXD cames from in vivo experiments in which the activity of a DFD homodimer binding site was cornpared with an EXD-DFD heterodimer-binding site (Li et al. 1999a). These experiments dernonstrated that, despite similar affinhies for corresponding binding sites, the EXDfDFD heterodirner is a potent transcriptional activator. while DFD homodimers are not capable of activating transcription. Based on these results and those of transfection assays, it was proposed that the interaction of DFD with EXD is required to release the coven transcriptional activation function of DFD (Li et al. 1999a). Although the nvo proposed roles for EXD are not mumally exclusive, it has been argued that EXD is primarily required for transcriptional activity regulation rather than HOX DNA binding (Biggin and McGinnis 1997). This is based on the iinding that DFD appears to be capable of occupying monorner HOX binding sites in Drosophile embryos in the absence of obvious cofactor binding sites (Li et al. 1999a). It is also consistent with the results of in vivo cross-linking experiments (Walter et al. 199.1: Biggin and McGinnis 1997: Liang and Biggin 1998; Cmand Biggin 1999), which suggest that transcriptional activity regulation. and not effects on DNA binding, contributes to different developmental roIes of QSO homeodornain pmteins. The latter idea is further supponed by experiments with a Ubx pmtein that has an enhanced ability to activate. This protein specifies an Antp-dependent cuticular phenotype. suggesting that the functional difference between the Ubx and dnrp proteins in diversifihg dentide patterns may reside in differences in activation and repression strengths on similar target genes cacher than in differences in target occupancy. Interactions with PBC proteins may in some cases be insufficient for HOX functional specificity. It has been dernonstrated, for example. chat the influence of EXD on HOX specificity can be superseded on complex enhancer elernents intepring a variety of regulatory information from multiple pathways (Grieder et al. 1997: Li et ai. 1999b). This can be mediated, in pan, by protein-protein interactions with a variery of cofactors affecting stability anàior activity of the PBC-HOX heterodimers. Several PBC-interacting proteins have ben identified, They include murine Meis gne products and the Drosophila homothorm- (hrh) gette product, which is a Meis ortholog (Moskow et al. 1995; Nakamura et al. 1996; Rieckhaf et al. 1997: Steelman et al. 1997; Kumt et al, 7 2 1998; Pai et al. 1998) as well as a protein that is encoded by the murine Prepl gene (Berthelsen et al. 1998).

1.4.2.3. OCT-1 and OCT-2

OCT-1 and OCT-2 are members of the POU-family of homeodomain proteins (Herr et al. 1988) that recognize a highly conserved 8-bp DNA sequence (ATGCAAAT), temed the octamer motif (Parslow et al. 1984: Singh et al. 1986). This motif is found in a large number of pmmoters (Heintz et al. 1981; Mackem and Roizman 1982; Falkner and Zachau 1984; Mattaj et al. 1985: Davidson et al. 1986: Mangin et al. 1986; LaBella et al. 1988; Cume and Roeder 1989; Meyer and Neuberger 1989). Despite similar DNA-binding properties, the regulatory specificities of OCT-1 and OCT-2 are different, This is, in part. reflected in differential promoter activation by OCT-1 and OCT-2 (Tanaka et al. 1988: Tanaka and Herr 1990: Tanaka et al. 1992). In particular, it has been dernonstrated rhat OCT-1 selectively activates snRNA promoters such as the human U2 promoter, whereas OCT-2 specifically induces mRNA promoters such as the hurnan B-globin promoter (Tanaka and Herr 1990: Tanaka et al. 1992). In both cases. transcriptional activation was mediared by the same octamer motif-containing enhancer. This indicates that OCT-1 and OCT-7 have inherently different activation potentials. It has been proposed that each of these proteins contains unique domains that specifically interact with particular core pmmoters and stimulate transcription by an assembly of different kinds of RIVA polymerase Il initiation complexes. Indeed. deletional snidies as well as experiments with OCT-IIOCT-3 chimeras dernonstrated that OCT-I and OCT-2 have different promoter-specific activation dornains located in the C-terminal portions of these two proteins and that the reciprocal msactivation function can be transferred between OCT-1 and OCT-2 upon exchanging these specific dornains (Gerster et al. 1990: Tanaka and Herr 1990: Tanaka et al. 1992). Thus. inmnsic reglatory differences between OCT-1 and OCT-7 provide an e.lrampIe of how differences in transcriptional potential rather than DNA binding can confer distinct regulatory specificities. Additional differences between OCT-1 and OCT-2 arise hmselective protein-protein interactions with auxiliary factors. For example, OCT-1 but not OCT-2 associates with the herpes simplex virus (HSV) activator VP16. VP16 is a multifünctionai protein that serves as both a structural component of the HSV virion (Batterson and Rohan 1983: Campbell et al. 1984: Ace et al- 1988) and an activator of HSV IE transcription (Goding and O'Hare 1989). The VPI6-OCTI complex aIso contains a second factor tenned HCF (Host Ce11 Factor) (Katan et aI. 1990; Xiao and Capone 1990: Stem and Herr 199 1 ). VP 16 binds poorIy ro DNA on its own (Mmden et al. 1987; Krisrie and Sharp 1990), however. when associateci with OCT-1 and HCF. VP16 is able to bind to DNA avidly. The binding specificity of the complex is different fiom that of OCT-I as OCT-1 is recruited to a new cis-regdatory site that in the absence of VP16 is not responsive to either OCT-1 7 3 or OCT-7 (Baumntker et al. 1988; Preston et al. 1988; apRhys et al. 1989). The interaction with VPl6 enables OCT-L to activate mRNA promoters similar to those regulated by OCT-2. This is due to the potent carbolcyterminal 80-amino-acid acidic activation domain of VP16 (Sadowski et ai. 1988: Triezenberg et al. 1988; Cousens et al. 1989; Greaves and O'Hare 1989). Thus, VPL6 affects tbe reçulatory rpecificity of OCT-1 through altering both OCT-1 transcriptional potential and irs DNA-binding properties. The protein-protein interaction beween OCT-1 and VP16 has been mapped to the homeodomain of OCT-1 (Stem et al. 1989). This observation, together with the finding that OCT-2 is unable to interact with VP16, suggested that amino-acid-sequence differences between OCT-1 and OCT-2 homeodomains could account for the inability of OCT-2 to interact with VP16. Surpnsingly. the homeodomain is the region of highest similarity between OCT-1 and OCT-7. The homedomains of OCT-1 and OCT-2 differ only in seven arnino acids. It has been demonsuated that substitution of a single amino acid, fiom alanine to glutamic acid at position 22 in the end of helix L of the homeodomain. confers on OCT-2 the ability to interact successfully with VPL6 (Lai et al. 1992; Pomerantz et al. 1992). This demonstrates that the homeodomain can confer regularory specificity via interactions with specific proteins as well as through direct contacts with DNA.

1.4.2.4. 1-POU and Cfi-a

Drosophila POU-domain proteins have been isolated based on homo1ogy with marnmalian proteins. Investigation of Drosophila POU-homeodornain proteins in the developing nervous system has revealed a functionally important interaction between I-POU and Cfl-a,both of which are POU proteins. Cfl-a binds to a specific tegulatory element in the Dopa decarbo.ylase (Ddc)gene and activates its transcription (Johnson and Hinh 1990). 1-POU forms a heterdimer with Cfl-a in soIution and affects Cfl-a function by inhibiting its binding to the Ddc reguIatory element (Treacy et al. 1991). The protein-protein interaction between 1-POU and Cfl -a is highIy specific and depends upon a cluster of basic amino acids at the N-terminus and on the fim two helices of the homeodomain (Treacy et al. 1992). 1-POU lacks residues 3 and 4 of the homeodomain and therefore does not bind to DNA. A splicing variant transcript, referred to as nivin of 1-POU. restores these NO amino acid residues (Treacy et al. 1992). Twin of [-POU is no longer capable of interacting with CfI-a, but its capacity for specific DNA binding is restored.

l.l.2.5. Unc-83 and Mec3

Caenorhabdiris elegans UNC-86 is the founding member of the POU-cIass of homeodomain proteins (Herr et al. 1988). It is requkd for establishimg and maintainhg touch-cell fates (Xue et al. 1992). Evidence has been presented for the cooperative action of WC-86 with MEC-3, a LIM homeodomain protein required for the proper differentiation of touch ceIls, once they have been 74 iormed (Xue et al. 1992; Xue et al. 1993). Both proteins are required for the correct regulatian of the mec-3 gene itself (Xue et al. 1992). DNase footprinting of the mec-3 autoregulatory element revealed a number of ovcrlapping binding sites for the nvo proteins (Xue et al. 1992). On their own, each protein binds weakly to these sites. but together they bind cooperativeIy (Xue et al. 1992; Xue et al. 1993). Since the two proteins can be CO-imrnunoprecipitated(Xue et al. 1993), this cooperativity seerns to involve direct protein-protein interactions. Deletional analysis has narrowed down the minimal interaction domains of these pmteins to the POU domain of UNC-86 and the homeodomain plus I6 C-terminal amino acids of MEC-3 (Xue et al, 1993). The in vivo significance of the interaction between MEC3 and UNC-86 is emphasized by two mec-J alleles. One of these alleles, MEC-3 (C779Y), is a missense mutation in the C-terminal UNC-86 interaction domain that substitutes tyrosine for cysteine 779. This mutant protein is still able to bind DNA, but is no longer capable OP binding cooperatively with UNC-86,and consequently acts as a loss-of-hnction ailele. The other allele. MEC-3 (Q266C), is a missense mutation in the homeodomain that substitutes cysteine for giuiamine 266, The mutant protein has a IO-fold Iower afinity for DNA than wild-type. Still. this protein is able to bind DNA cooperatively with üNC-86 and to act as effectively as the wild-type prorein in vivo (Xue et al. 1993).

1.43. OTHER DNA-BlNDmC DOMAINS

A number of homeodomain proteins achieve additional DNA-binding specificity by the coupling of homeodomains with other DNA-binding domains (Bopp et al. 1986: Sturm and Hem 1988: Freyd et al. 1990; Karlsson et al. 1990: Kristie and Sharp 1990: Treisman et al. 199 1 ). For esample. PRD, in addition to the horneodomain, also contains a 128-amino acid Paired domain (Bopp et al. 1986: Treisman et al. 1991). The Paired domain can act independently of or through intramolecular cooperativity with the homeodomain to enfiance the specificity of PRD binding to DNA (Treisman et al. I99 1). Another example of a homeodomain-associated DNA-binding dornain is the POU-specific domain found in POU-class proteins (Hem et al. 1988; SN^ and Herr 1988: ingraham et al. 1990: Verrijzer et al. 1990). This domain is always associated with a POU-type homeodornain. It contacts DNA next to the horneodomain site and is unable to bind in the absence of the homeodomain (Kn'stie and Sharp 1990: Vemjzer et al. 1990). The homeodornain can also be associated with a 60 amino acid repeat caIled ~heLiM domain (Freyd et al. 1990; Karlsson et al. 1990), or with a Cys-His motif reminiscent of the zinc finger (Freyd et al. 1990: Karlsson et ai. 1990). or other homeodomains (Fortini et al. 1991). Sorne homeodomain proteins contain multiple DNA-binding domains. The most exrraordinary examples are the Orosophila ZFH-1 and ZFH-2 (Fortini et al. 1991). WhiIe ZFH-I has a homeodomain and nine potentid DNA-binding zinc fingers, ZFH-2 contains three homeodomains and 16 zinc fingers (Fortini et al. 1991). Their putative human homolog ATBF-1 conrains four homeodomains and 17 zinc ftgen (Mocinaga et al. 1991). 1.4.4. PROTEIN EXPRESSION LEVELS

Levels of pmtein expression also seem to have an important effect on morphoregulatory functions of homeodomain proteins. tt has been demonstrated, for example, that pattern specification by BCD is concentration-dependent (Driever and Nusslein-Volhard I988a). This paraliels the regulatory properties of BCD, which is capable of activating different target genes at different concentration thresholds (Diever and Nusslein-Volhard 1989; Driever et al. 1989). It has been proposed that critical threstiold concentrations are determined by afinities of BCD binding sites in target gene promoter elements (Driever and Nusslein-Volhard 1989; Driever et al. 1989). The capaci~yof concentration-dependent gene regulation has also been reported for EVE (Manoukian and Krause 1992; Fujioka et al. 1995). [n this case, it has been demonstrated that several negatively regulated target gens respond to difierent levels of EVE. In particular, odd and wg are repressed by very low levels of EVE;fiz, sslp and nrn by intermediate levels; and prd only by very high levels. This concentration-dependent repression is apparently required for establishment of odd- and even- nurnbered stripes of en expression (Manoukian and Krause 1992; Fujioka et al. 1995). The function of other homeodomain proteins may also depend on their concentration. For example, the level of Clbr, which is expressed at very high, almost unifonn Ievels in PS6, and at a much lower level and in only a subset of cells in PSS (White and Wilcox 1985a). specifies distinct characteristics benveen PS5 and PS6. It has been found that imagina1 histoblast nests, normally present in PS6 but absent in PS5. are present in both parasegments in Iarvae carrying extra doses of Ubx (Frayne and Sato 199 1). In addition, the identity of the PS5 denticle belt is transformed in these larvae into the identity of PS6 (Smolik-Udaut 1990). Finally, expression of UBX at a reduced level in PS5 seems to be required for the initiation of DI1 expression in the limb primordia and the formation of Keilin's organs. Both Dl1 and Keilin's organs are suppressed by high UBX levels and are therefore not observed in PS6 (Castetli-Gair and Akam 1995). Like UBX, other horneotic gene products are also knoivn to be expressed at different levels in different celis (White and Wilcox 1985b; Carroll et al. 1986a; Win et al. 1986; Celniker et al. 1989; Karch et al. 1990; Deguchi et al. 1991) and their concentration-dependent finctions cm be inferred hmtheir haplo-insufficiency mutations. For example, if only one dose of Scr replaces the normal two doses, a partial transformation of segments is observed (Lewis 1963; Kauhan et al, 1980). This demonstrates that alterations in a homeodomain protein concentration can bring about a morphotogical change. Distinct rnorpho1ogies at diffïerenc levels of the horneotic gene products could result. in part, from competitive functiooal interactions among these proteins. Such competitive interactions have been previously proposed to account for the phenomenon of phenotypic suppression in which one HOX protein cari dorninantly suppress the tùnction of other coexpressed HOX proteins (Gonzalez-Reyes et al. 1990: Lutkin et al. 199 1). It has been proposed that cornpetition of HOX proteins for DNA binding sites is responsible for this phenomenon (Gonzalez-Reyes et al. 1990; Lufkin et al. 1991). 7 6 In my studies, 1 have investigated the activity of FTZ in developing Drosophila embtyos. I have attempted to address which of the genetically identitied FTZ downstream genes are directly and which are indirectly regulated by FTZ (chapterl). In addition, I have been interested in how coCactors involved in FTZ-dependent regulation affect FTZ hnction (chapter3). My goal was to assess whether the main role of these cofactors is to rcgulatc FTZ transcriptional activiy or to control DNA binding, or both. Chapter 2:

Kinetic anaiysis of segmentation gene interactions in Drosophila em bryos

A similar report was published in Andnej Nasiadka and Henry M. Krause, Development: 126, 1515-1526 (1999) A major challenge for developrnental biologists in coming years witi be to place the vast nurnber of newly identified genes into precisely ordered genetic and molecular pathways. This will require efllcient methods to detennine which genes interact directly and indirectly. One of the rnost comprehensive pathways currentIy under sîudy is the genetic hierarchy that controls Drosophilu segmentation. Yet, many of the potential interactions within rhis pathway remain untested or unverified. Here, we look at one of the best-characterized components of this pathway, the homeodomain-concaining transcription factor Fushi tarazu (FTZ), and analyze the response kinetics of known and putative targer genes. This is achieved by providing a bnef pulse of FTZ expression and measuring the tirne required for genes to respond. The time required for FTZ to bind and regulate its awn enhancer, a welI-documented interaction, is used as a standard for other direct interactions. Surprisingly, we 6nd that both positively and negatively regulated target genes respond to FTZ with the same kinetics as autoregulation. The rate-iimiting step benveen successive interactions (

The involvement of transcriptional cascades in deveropment is becoming increasingly apparent. Well-known exampies inchde the hierarchical interactions underlying hematopoiesis and adipogenesis in vertebrates (reviewed in Shivdasani and Orkin 1996; Fajas et ai. 1998), and the ecdysone and segmentation gene pathways in Drosophila (reviewed in Rivera-Pomar and Jackle 1996; Thumrnel 1996). Gene expression within these cascades is predominantly controlled at the level of transcript initiation, and is based on interactions between sequence-specific transcription factors and their cis-acting response eiemenis. Two spes of regulatory relationships, direct and indirect, can be defined. Direct interactions occur independently of intermediary gene regulation but need not involve direct moIecular contact between the regulator and its carget gene promoter. Indirect interactions require the activation or reptession of intermediary genes, the products of which act on the target gene in question. The segmentation gene hierarchy of Drosophila is one of the best-characterized deveiopmental cascades (reviewed in In&m and Martinez Arias 1992; St Johnson and Nusslein- Vohard 1992). Its rote is to pattern the anterior-posterior body axis by converting materna1 information, targeiy in the form of transcription factor gradients, into repeating segmenta1 units. 7 9 This requires the sequential regulation of several gene classes. First, gap genes respond to the graded distribution of materna1 products, fonning bands of expression that span several segment primordia. Then, gap gene proteins regulate expression of the pair-rule genes generating periodic srripes, approximately a segment in width, in different portions of altemate segmental primordia. Finally, pair-rule proteins interact in various combinatorial codes to direct the expression of segment poIarity genes. Segment polarity genes are expressed in stripes in different sub-regions of each Future segment. These stripes provide the blueprint for segmenta1 patteming. In addition to these hierarchai interactions, there are a large number of interactions that occur within each class of segmentation genes and substantial feedback occun between ciasses (e.g. Harding et al. 1986; Jackle et al. 1980: Ingham and Gergen 1988; Hulskamp et al. 1990; Knut and Levine 1991a). This cross-regulation significantly increases the complexity of the regulatory network. Genetic studies have provided a qatdeal of information on the function of this hierarchy. In particular, genetic screens are believed to have identified most of the loci involved (Nusslein- Volhard and Wieschaus 1980; Schupbach and Wieschaus 1986b; Pemmon et al. 1989), and subsequent epistasis studies have defined the regdatory relationships between many of these genes. Afthough the contribution of these studies k profound, many important issues remain unclear. For example, in very few cases do we know whether genetic interactions represent direct or indirect regulatory interactions. The approach rnost ofien used to distinguish between direct and indirect interactions is to identifi binding sites for the putative regulator within the promoters of genetically detined target genes, and to then analyze expression of normal and mutated forms of the promoter in vivo (e.g. Jiang et al. 1991; Schier and Gehring 1992; Schier and Gehring 1993a; Capovilla et al. 1994). Although this combination of rnolecular and genetic approaches has ken used successfully, it is quite tedious, and problems in its execution and interpretation have been encountered. For example, the functional specificity of transcription factors in vivo cannot always be predicted by the DNA-binding properties that they exhibit in vitro. in fact, there is increasing evidence that protein-protein interactions often play a decisive role in detemiining binding site specificity (Chan et al, 1994b: Copeland et al. 1996; Reichardt et al. 1998). Additional complications encountered in these studies include redundancy of cis-acting binding sites and their irans-acting factors, and the size and complexity of most developmentally regulated promoters (e.g. Han et al. 1989; Howard and Stnihl 1990; Tautz 1992; Klingler et al. 1996). These drawbacks emphasue the need for aitemative methods capable of distinguishing between direct and indirect interactions in an in vivo setting. Exarnples of currently employed methods include in vivo cross-linking of proteins to their target sites (Gilmour and Lis 1986; Gould et al. 1990; Walter et al. 1994), in vivo fooîprinting (Huibregtse and Engelke 1989; Mirkovitch and Darnell 1991) and, in Drosophila, visualizaiion of pmteins bound to polytene chromosomes (Zink and Paro 1989; Umess and Thumrnel 1990). Another approach has been to monitor the temporal response of putative target genes foiiowing puIsed expression of the regulator. In this casef the 8 O distinction between direct and indirect targets is based on the assumption that direct targets respond immediately while indirect targets respond with a delay due to the time required for intermediary gene expression. Studies of this type have been used to identie likely targets of several pair-nile proteins (Ish-Horowicz and Pinchin 1987; Ish-Horowicz et al. 1989; Morrissey et al. 1991; Manoukian and Krause 1992; Manoukian and Krause 1993; John et al. 1995; Saulier-Le Drean et al. 1998). However, the time required between direct gene interactions has yet to be rigorously determined. The best-documented example of a direct regulatory interaction amongst Drosophiia segmentation genes is the action of the pair-rule protein Fushi tarazu (FTZ) upon its own promoter. FTZ is a homeodomain-containing transcription factor required for the formation of altemate segmental regions (Wakimoto and Kauhan 1981; Wakimoto et al. 1984) referred to as even- numbered parasegments. The function of FTZ as a direct regulator of its own promoter comes from a number of elegant studies. These include promoter deletional analyses, biochemical studies and rnolecular genetic approaches (Hiromi et al. 1985; Hiromi and Gehring 1987; Ish-Horowicz et al. 1989; Pick et al. 1990; Schier and Gehring 1992; Schier and Gehring 1993a). The most convincing of these studies was the demonstration that mutations in a number of FTZ-binding sites identified in a minimal fi= enhancer element disrupt autoreguIation (Schier and Gehring 1992). More importantly, it was shown that the effect of these mutations could be reversed by expressing a mutated version of FTZ that preferentially binds the rnutated sites. Molecular and genetic data indicate that the segment polarity gene rngraiied (en) is also likely to be a direct target of FTZ (Howard and Ingham 1986; DiNardo and O'Farrell 1987; Desplan et al. 1988: Han et al. 1989; Florence et al. 1997). FTZ-dependent enhancer elements have been identified in both Y-flanking and intron sequences, and both contain FTZ-binding sites. Deletion of the FTZ-binding sites in an intron-derived reporter abolished fi:-dependent expression in vivo (Florence et al. 1997). Genetic data, although less conclusive, suggest that another possible target of FTZ is the segment polarity gene wingiess (wg) (Ingham et al. 1988). Interestingly, the action of FTZ on this gene is negative, suggesting that FTZ might be able to function as both an activator and repressor of transcription. Other possible targets, suggested by genetic analyses, include the pair-rule genes paired @rd) (Baumgartner and Noll 1990), even-skipped (eve) (Kellerman et al. 1990) and runr (mn)(Klingler and Gergen 1993). Here, we use pulses of ectopic FTZ expression in vivo to detenine the kinetics of these and other putative target genes. The response time forfi= autoregulation is used as a standard. We find that the responses of putative target genes fail within two easily disthguished temporal windows, one that overlaps with the autoregulatory response and a second that occurs with a CIO-minute delay. We argue that these different temporal windows teflect the responses of direct and indirect targets. respectively. Interestingly, we fTnd that both activated and repressed target genes respond with the kinetics of direct target genes, indicating that FTZ does indeed have the abiiity to function as both an activator and repressor of transcription. Furthemore, we show that both processes proceed in parallel due to the matching of pmtein synthesis and degradation rates, which are the rate-limiting 8 1 processes between successive gene interactions. Our data also provide important new insights into the complexity of FTZ functions required for the specification and differentiation of even-numbered parasegments.

23 MATEXiALS AND METHODS

Fiy stocks and mrriant alleies

rd Two HSFTZ lines, a heterozygous Iine (245-A) canying the inducible transgene on the 3 chromosome (Fitzpatn'ck et al. 1992) and a homozygous line (hsfl) carrying the transgene on the

znd chromororns (Struhl 1985). were used in this study. Two nuIl alleles 0€Jz,jz9H34 (Jurgeenr et al.

1984) andjzW2' (Lewis et al. 1980a; Lewis et al. 1980b: Wakimoto and Kauhan 198 1). were used to examine target gene expression patterns in a& mutant background. Both caused identicai changes in the expression of analyzed targets. To ÏdentiQ homozygous mutant embryos, mutant chromosomes were baIanced over a TM3 balancer chromosome marked with a hunchback &gai '.45./ 7 reporter (Driever et al. 1989). The allele of prd used for mutant analysis was prh (Frigerio et

al. 1986). pd, HSFTZ embryos were obtained as segregates from a stock of phenocype prd.4jbf7~~y~; HSFTZfTM3, Sb. To idenri@ embryos homozygous for '', a CyO balancer carrying the hunchback &gai reporter was used (Driever et al. 1989)-

rnRNA and protain Iocaluution

Except forfrz, probes for in situ hybridization were prepared by PCR in wo steps. The first step was to ampli@ a cDNA portion of the target gene from a plasmid template. The product of this reaction was then used as template? along with the 3' primer and nucleotides to generate ruitisense probes. Nucleotides used in the latter reaction contâined digoxygenin (DE)-labeled dUTP (Boehringer Mannheim). The following plasmids were used to prepare probes: pSKrun (provided by P. Gergen), pJGH4 containhg haiiy cDNA (provided by D. [sh-Horowicz), pKSll+evel7/ and prdORFZ (provided by C. Desplan), odd 7.4 Il-F (Coulter et al., 1990), pBSslp2 (provided by W. Ghering), pwg-cl4 (provided by N. Baker), pBSen and pNB4Oopa (provided by S. DiNado), and pBSgsb (provided by S. Cote), l7 and T3 promoter primers (Promega Corporation) were used to prepare probes for runt, en. eve, odd. slp and wg. Other probes were prepared with the following primers: 5'-GCTCGCCACTCCMTTGG-3' and 5'-GCAGCTGCTGCTGCnCCGG-3'for haiv, 5'- GCCTCAGTAIIGCCATGTCGC-3 ' and 5'-GGGCGGTGACA TCCAGAGCC-3 ' for paired 5'- GATGAACGCCTCATTGAGC-3' and 5'-TTGACCAGCWGTACTTGGC-3' for opa To prepare the fi= probe, the pGEMF-I plasmid, made by subdoning a Safi-AvaIi Fragment of the#= transcription 82 unit into pGEM-I vector (Promega Corporation), was fim digested with Fspl. The digested DNA was then used as a template, together with a Ti promoter primer (Promega Corporation), to obtain DIG- labeled anti-sense cDNA by PCR in situ hybridization to whole-mount embryos using DIG-labeled DNA probes was performed as follows. Just prior to in situ hybridization, dechorionated and f~vedembryos were ftved again for 20 minutes with 5% fonnaldehyde solution prepared in PBT (IXPBS (phosphate-buffered saline solution) plus 0.1% Tween-20). Thereafler, the embryos were rinsed three times with PBT and incubated for 3 minutes in the solution of non-predigested Proteinase K (50 pg/ml) in PBT. The proteinase digestion was stopped by two rinses in 2 mg/ml glycine solution prepared in PBT, followed by nvo rinses in PBT solution. AAer Proteinase K treatment, the embryos were fixed again with 5% Formaldehyde solution (in PBT), and rinsed extensively (5 times) in PBT (to remove al1 traces of fixative). Thereafter, the embryos were washed once with 50% PBT and 30% DNA hybridization solution (50% deionized formamide, SXSSC, 100 pdml tRNA, 50 pdml heparin, 0.1% Tween ?O), and once with 100% hybndization solution. Then. thembryos were prehybridized (in 100% hybndization solution) for 2 hours at 48" C. AAer prehybridization, DiG-labelled anti-sense cDNA probe was added to the embryos in LOO% hybridization solution and hybridization was carried out at 48'C for 12-16 hours. Thereafier. the embryos were washed extensively. First, in 100% hybridization solution, followed by a wash in 50% hybridizarion solution and 50% PBT, and five washes in PBT. First three of these washes were carried out rit 48OC. and the remaining two washes at room temperature. Duration of each of these washes was approximately 30 minutes. The embryos were then incubated for two hours at room ternperarure with anti-D[G antibody conjugated to AP (alkaline phosphatase). After extensive washes (four washes for 20 minutes) in PBT, the embryos were briefly rinsed with AP solution (100 rnM NaCL 50 mM MgCIi IOOmM Tris pH 9.5. ImM Levamisol, 0.1% Tween 20). In the next step, hP soluble substrates were added. In the course of the enzymatic reaction, these substrates are convened into insoluble precipitating products. Immunolocalization of proteins, and double-labeiing of protein and mRNA were performed as previously described (Manoukian and Krause 1992). Double-Iabeling using anti-&galactosidase antibodies was used to detect embryos heieterozygous and wild-type forfi,- and prd. Anti-$-gal antibodies were obtained from Promega Corporation and PRD antibody was provided by C. Despian.

Determinittg target gene response kinerics

HSFTZ embryos were colIected on apple juicdagar plates for 40 minutes and aged for 7 hours and 40 minutes at ?j°C. Embryos were then heat-ereated for 8 minutes by submersion in water preheated to 36.5' C. Following heat-treatment embryos were split into 8 batches and aliowed to recover at 25' C. Embryos were dechorionated and Exed at 5 minute intervals thereafter, starting with a 3 minute recovery and ending with a 40 minute recovery. Once fked, embryos were stored in MeOH at -20' C. This procedure was repeated a number of rimes, and the embryos representing each 8 3 5 minute recovery time pooled. Once sufEïcient numbers of ernbryos were collected, embiyos representing each 5 minute intervai were once again divided into aliquots for hybridization with different probes. Forftz, en, wg and prd, the experiment was repeated and the values obtained tkom each experirnent were averaged. Following hybridization and mounting, expression patterns were observed by microscopy and the number of altered patterns scored. For ectopically activated targets, responses were first scored as positive when ectopic expression was about 2550% of endogenous levels. For repressed targets, unaffected smpes served as interna1 controls. Hence, early responses could be scored with more certainty. Stripes with about a 20% drop in relative expression levels were eady detected. To help with counting, grids were drawn on each slide. ApproxirnateIy 200-300 embryos were mounted per slide and two slides were prepared for each probe/recovery tirne. Percentages of embryos showing altered expression patterns were detennined and plotted in relation to their time of recovery. Curves calculated in this manner overlap closely with curves generated in previous snidies, where target gene responses were quantitated by enzyme reactions and colorimetry (Manoukian and Krause 1992). Kinetic curves were obtained for transcripts encoded by theh en. ivg, and prd genes, and for proteins encoded by the prd and en genes. AI1 other target gene responses were simply observed at 20 and 40 minutes post-heat shock. For heat-sensitive loci such as m. eve. gsb and wg, experiments were repeated using the heat-inducible FTZ line hst2 (Smhl 1985)- Responses were anaiyzed with 2, 4, 6 and 8 minute heat shocks. Al1 FTZ-specific responses could be reproduced with only a 2-4 minute heat shock. Non-specific responses due to heat treatment were not observed with these durations of heat shock. Specific venus non-specific effects were determined by analyring the response of al1 loci in heat-shocked and non-heat-shocked Oregon R controls, treated in parallei.

2.1 RESULTS

Eiperimental approach

To monitor transcriptional targets of FTZ, expression patterns of putative target genes were monitored by in situ hybridization in heat-inducible FTZ (HSFTZ) transgenic embryos. HSFTZ embryos carry afl transgene in which thefi cDNA is transcribed under control of the hp7û heat- inducible promoter (Sûuhl 1983'). When heat-pulsed, these ernbryos express fi throughout the embryo. Two HSFTZ lines (Stmhl 1985; Fitqatrick et ai. 1992), carrying difTerent consmcts inserted at different genomic locations (see Materials and methods), were used in this mdy. The majority of experiments were perfonned using a Iine carrying a heterozygous copy of the transgene on the third chromosome (Fitqamck et al. 1992). Heat pulses used for this line were for 8 minutes at 36.j°C. These conditions are optimal for the regulation off= mget gens and induction of an "anti--. cuticuIar phenotype, as previously described (Stmhl 1985; Ish-Horowicz et al. 1989). Under 8 4 these conditions, the levels of ectopicjlz expressed hmthe transgene are lower than endogenous levels of expression (Smhl 1985; Fiapatrick et al, 1992)- The second line, hsf2 (Stmhl 1985), was used when analyzing the expression of heat-sensitive loci. This line expresses the same levels of protein as the first with half the duration of heat shock (4 minutes at 36S°C). Non-specific effkcts due to heat treatment were revealed by analyzing gene expression patterns in Oregon R embryos heat shocked in parallel to HSFTZ embryos. No changes in target gene expression were obsemed in Oregon R ernbryos foilowing a 4 minute heat pulse.

Kinerics of endogenous fîz induction

Compelling molecular and genetic evidence has been presented previously demonstrating that FTZ directly binds and regulates expression of its own promoter (Hiromi et al. 1985; Hiromi and Gehring 1987; Schier and Gehring 1992). Hence, the temporal response of this interaction was measured and used as a basis of comparison for other possible direct interactions. Previous studies have shown that endogenousjlz stripes respond to ectopic FTZ expression by limited expansion: each stripe expands anteriorly by about one cell (Figure 2.1 8) (Ish-Horowicz et al. 1989). To determine the kinetics of this response, HSFTZ embryos were heat treated for 8 minutes and aliquots were Lied at 5 minute intervals (see Materials and methods). Enough embryos (approximately 50,000 in total) were collected so that each aliquot of staged and fixed embryos could be partitioned for staining with as many as ten different probes. In this way, possible variations in staining due to differing heat shock or fixation conditions were avoided. A minimum of 500 embryos per 5 minute interval was examined by in situ hybridization for changes infi expression. The Fraction of embryos e-xhibiting broadened FTZ stnpes was determined. Embryos were scored as positive when the regions of broadening could be visually determined with cenainty (see Matenals and methods). The results of this analysis are presented in the form of a curvz (Figure 2.1C) which plots the percentage of embryos e.x.hibiting expandedjk stripes as a function of the the in minutes foliowing the heat pulse. This curve represents the average of four experiments performed with nvo separate collection series. Embryos for each experiment were counted mice (eight counn in total) with essentially identical numben tallied for each duplicate count. The resultant curve shows that autoregdation, resulting in fc smpe widening, occurs benveen 15 and 20 minutes post heat shock. Afier this tirne, a plateau is reached with about 65% of embryos having responded. Since the HSFTZ line used in this experirnent is heterozygous, a maximum of 75% of the embryos examined would be expected to exhibit this response. FaiIure to reach this theoretical 85 Figure 2.1 Kinetics of eudogenous fa activation. Endogenous fi,. mRNA in wild-type (A) and HSFTZ (B) stage 7 embryos. The embryo in B was fied 20 minutes afler an 8 minute heat shock. Stripes of endogenousfr,. expression are norrnally 1-2 cells wide at this nage (except &pe 7), but respond to a pulse of ubiquitous FTZ expression by expanding anteriorly an additional 1-2 cells per suip. The kinetics of this broadening response is shown in (C). The curve shows the percentage of embryos exhibiting broadened smpes at various tirne points after the end of heat shock, The values shown represent the average of four experirnents (performed with two separate collections). Standard deviations for these values are between O and 2%. I time afier heat shoc k (minutes) maximum at rnost time points is probably due to the presence of embryos that were either inaccurately staged, unfertilized or ineficiently heat shocked. One exception was the value detemined for the 30 minute time point. This was a hi&l reproducible 72%, after which the number of embryos with widened stripes appeared to steadily decrease. While this suggests the possibility of complex influences on the ability of FTZ to autoregulate in these cells, we are reluctant to make conclusions based on a single the point. engrailed (en) activation follows the sume kinetics as fîz activarion

The second best-characterized target of FTZ, and hence the nea target that we examined, is the segment polarity gene engrailed (en). Stripes of en initiate at the anterior edge of each parasegment (Figure 2.2A). Every second stripe overlaps with the anterior portion of eachjk smpe. and is lost inftz mutant embryos (Howard and tngharn 1986: DiNardo and O'Farrell 1987; Lawrence et al. 1987). As previously shown, ubiquitous expression of FTZ causes a broadening of expression of thesefi-dependent en smpes (Ish-Homwicz et aI. 1989 and Figure XB), making them 3-3 cells wide instead of 1-2 cells wide. This expansion occurs in the same celis as endogenousfi stripe expansion. The kinetics of this response was detemined in the same way as described above for endogenousfi, using embryos fiom the same set of collections. As shown in Figure 3.2C, the curve generated for en overlaps very closely that of the endogenousfi response. suggesting that both genes are direct targets of FTZ. If en were regulatrd indirect-, a delay would be eiipected reflecting the time required for intermedia. gene products to be expressed. to accumulate and to elicit a response.

Two responses of wingless (wg) in HSFTZ embryos

In contrast to en. the segment polarity gene wfitgless (wg) has been identified genetically as a negative target of FTZ (Ingham et al. 1988). This negative interaction has also been demonstrated in HSFTZ embryos (Ish-Horowicz et al. 1989). Althou& al1 wg stcipes are repressed in these embryos, the predominant effect is on odd-numbered stnps which. as shown in Figure 2.3B, are completeiy repressed following FTZ induction. Repression of even-numbered stripes is much less efficient. To assess whether this repression is direct, we detemined the kinetics of repression as measured above for fc and en activation. The differential repression of odd- versus even-numbered smpes of wg was a helpful tool and interna1 contrd for recogizing affected embryos. The kinetic curve showing the percentag of embqos affected following heat shock is presented in Figure 2.3D. Again, this curve follows very cIoseIy that offl autoreguiation. with the midpoint of both curves occumng at 18 minutes pst-heat shock. This indicates that repression of wg by FTZ is also Iikely to be direct, and that FTZ can act as both an activator and cepressor of transcription. 88 Figure 2.2 The kinetics of en and ftz activation are similar. (A) Expression of en in a stage 7 ernbryo. Fourteen evenly spaced stripes are present and are each 1-2 cells wide. The first six stripes are numbered. (B) in HSFTZ ernbryos, fmed 20 minutes after an 8 minute heat shock, even- numbered en stripes have expanded anteriorly by 1-2 cells. The kinetics of this response (C) overlaps that of endogenousfiz activation. Values for thefi curve are marked as squares, and those for the en cume (bmwn) as brown circles. As for&, the values calculated for en represent the average of two separate experiments. Standard deviations for these vahes range berneen O and 2%. C .S- 80 T m -fe mRNA induction -2' 70 1 a- en mRNA induction

I g O 10 15 20 2 5 3 O 3 5 4 [ .L - time afier heat shock (minutes) 9 0 Figure 2.3 Direct and indirect responses of wg. Expression of wg is shown in wild-type (A) and HSFTZ (B,C) stage 6 embryos. Two different changes in expression were observed in response to ectopic FTZ. The earliest response is characterized by repression of odd-numbered wg smpes (B). Even-numbered sûipes (designated with numbets) are also weakly repressed. in ernbryos fked later (35 minutes post-heat shock) (C), odd-numbered sûipes appear to expand anreriorly into odd- numbered parasegments. The kinetics of wg repression (green triangles) and wg activation (purple diarnonds) are shown in (D) and compared with the response of endogenousfi (black squares). The mid-points of response curves are indicated by vertical dashed Iines The values indicated represent the average of hoseparate experiments. Standard deviations for these values (not shown) range between O and 2%.

92 Repression was not the only response exhibited by wg in HSFTZ embryos. Weak activation within most of each odd-numbered parasegment was also detected (see Figure 2.3C). Figure 2.3D shows that the kinetic cuwe of this activation response is considerably delayed relative to the kinetics of the other three responses measured thus far. This suggests that ivg activation resutts from an indirect genetic interaction. A likely intermediary gene in this response is the paired gene @rd). prd is genetically required for the proper initiation of al1 14 wg stripes (Ingharn et al. 1988; Copeland et al. 1996), and al1 14 rvg stripes expand capidly in HSPRD embryos (Copeland et al. 1996).

PRDfunctiuns as an intermediate between FTZ and wg

To test whether the prd gene acts as an intermediate in the positive response of wg to ectopic FTZ, we determined the spatial and temporal responses of prd in HSFTZ embryos. Ifprd does function as an intennediary gene. its expression should be induced in odd-numbered parasegments where wg activation is later observed. Moreover, the induction of prd transcripts should occur with the same rapid kinetics as the f~.en and early wg responses. Shown in Figure 2.48 is the prd expression pattern detected 20 minutes afier ectopic expression of FTZ. Stripes are significantly wider than those in simiiarly staged wild-type embryos (Figure :.-!A). Using the most posterior stripe of prd as a landmark. it can be seen that each of the expanded stripes has broadened at its anterior edge. These regions of expansion comprise most of each odd-numbered parasegment, which is exactly where ectopic expression of wg occurs (Figure 2.3C). nie cime course ofprd mRNA induction was assessed as described for endogenousfi=. en and wg. Although the slopes of the prd andfi= activation curves differ (Figure 2.4E), the initial responses occur at about the same time. suggesting that the interaction between FTZ and prd is also direct. The differences in the slopes of the two curves are likely due to the autoregdatory nature of thefi response: for a short time. FTZ is expressed hmboth heat shock and endogenous promoters. and then maximal expression is sustained via autoregulation at the endogenous locus. In conmprd activation takes place in regions of the embryo where neitherjk (Ish-Horowicz et al. 1989) norprd (Morrissey et aI. 199 1) autoregulates. Hence, prd transcripts do not accumulate as quickly as those of jiz, and soon disappear due to degradation of the ectopicaily expressed FïZ activator. The role of prd as an intermediary factor in wg activation was tested Merby exarnining wg expression in a HSFTZ:prd background. In the absence ofprd, wg stripes should no longer expand. As shown in Figure L4D. ectopic FTZ does indeed fail to activate ectopic wg in the absence of prd. The expression pattern in HSFTZ;prd embryos is es senti al!^ identical to the pattern of wg expressed in prd embryos (Figure 24C): odd-numbered stripes are weak and even-numbered stripes are essentially absent. This result is consistent with the proposed role of prd as a direct activator of wg and as a genetic intermediate between FTZ and wg during ectopic stripe broadening. 93 Figure 2.4 prd is required for activation of wg. Expression patterns of prd mRNA are shown in wild-type (A) and HSFTZ (B) stage 7 embryos. Secondary stripes of prd expand anterïorly into odd- numbered parasepenci, 20 minutes after an 8 minute heat shock. These prd-expressing cells are the same cells that express ectopic wg 35 minutes after heat shock, consistent with prd acting as an intermediary activator. In prb embryos (C), even-numbered wg stripes are missing and odd-numbered stripes, the first four of which are numbered, are weak. In HSFTZ;prd embryos (D), ectopic expression of wg fails to occur. These results are consistent with PRD acting as a requisite activator of wg. (E) Comparison of the induction curves of prd (brown circles) and wg (purple diarnonds) transcnpts relative to the curve forfi autoregulation (black squares). Also shown is the time required for accumulation of PRD protein within the nucleus (red squares). It can be seen from the curves that prd responds with the kinetics of a direct target, and that the time course of nuclear PRD accumulation precedes that of wg transcript accumuiation by 1-2 minutes. The values represent an average of 2 experiments. Standard deviations of these values (not shown) range between O and 4%. 'raction of embryos, with ectopic repression or cctopic induction (%) Direct reguililion of wg by PRD

Previous studies (Copeland et al. I996), and the results presented above, support the idea that the prd protein (PRD) is a direct activator of wg. To verify this, we analyzed the nature of the temporal delay between prd and rvg activation. Specifically, we examined the temporal accumulation of PRD protein with respect to prd and wg transcripts. If the interaction between prd and wg is direct, then we would expect that much of the interval between accumulation of the two transcripts wouId be occupied by synthesis and nuclear transport of the prd protein. NucIear expression of PRD was monitored immunohistochemically at various intervals following ectopic FTZ induction and quantitated as described above (see also Materials and methods). The kinetic curve for ectopic PRD induction is shown in Figure 2.4E. This curve closely resembles that of wg mRNA activation except that it is shifted by 1-2 minutes to the left (earlier). This indicates that most of the delay observed benveen the accumulation of prd and wg transcripts (about 8 minutes) is consumed by the synthesis and localization of prd protein. The time required (approx. 6-7 minutes) may be fairty typical of other segmentation proteins expressed at this stage. lndeed, the deIay between detection of en transcript and protein responses was also 6-7 minutes, with curves that were virtually identical to those ofprd transcripts and protein (data not shown). These data do not exclude the possibility that there are genes in addition to prd that are required for ectopic activation of wg. However, if such gene products are required, their rates of synthesis or removal do not appear to supercede the temporal limitations imposed by the synthesis of PRD,

Primary pair-rule genes do not respond directly to ectopic FE?

The pair-rule genes hairy (h), runr (run) and even-skipped (eve) were previously designated as primary pair-rule genes, as the? were believed to regulate, and not to be regulated, by the other pair- rule loci (Carroll and Scott 1986; Howard and [ngham 1986; Ingham and Gergen 1988: Pankraa et al. 1990). However, the results of more recent genetic analyses suggest that rn znd run may be targets of FTZ (Kellerman et al. 1990; Klingler and Gergen 1993). Given the conflicting nature of these resuits, we analyzed expression patterns of the primary pair-rule genes in HSFTZ embryos to clarify their regulatory relationships. Since we have consistently observed primary and secondary responses that peak 20 and 35 minutes post-heat shock (this study and Manoukian and buse 1992; Manoukian and Krause 1993; Saulier-Le Drean et al. 1998), we examined the primary pair-mle gene expression patterns at these time points. The more efficient FTZ-expressing line, hsf2 (Smhl 1983, was used due to the sensitivity of eve and run to longer (>4 minutes) heat shocks. Figure SBshows a typicd example of eve expression 20 minutes aller induction of ectopic ETZ. This pattern shows no obvious differences From the wild-type pattern show in Figure 2.5A. 96 Figure 2.5 Expression of eve is una2fected by ectopic FCZ. Expression of eve is show in wild- type (A) and hsf2 (B) stage 5 (late) ernbryos. The embryo in (B) was hed 20 minutes after a 4 minute heat shock. No significant differences in the nvo expression patterns were observed.

98 Likewise, expression of h and run were also unaffected 20 minutes after induction of FTZ (not shown). This was me for al1 stages (5-7) examined. Thus, no evidence was found to support the possibility that these genes are direct targets of FTZ. Somewhat surprisingly, no differences in primary pair-rule gene expression patterns were apparent at the later recovery time either (data not shown), indicating a lack of both direct and indirect responses. Similarly, no primary or secondary responses were observed for another pair-rule gene odd-paired (data not shown). Thus, only a subset of genes investigated in this study respocd to ectopic FTZ, further demonstrating the specificity of this assay and FTZ function.

prd, odd-skipped (odd) and sloppy-paired (slp) are direct targets of FTZ

To determine the regulatory relationships between FTZ and the other non-primary pair-rule genes, we analyzed the expression of odd-skipped (odd) and sloppy-paired (slp) in HSFTZ embryos fixed 20 and 35 minutes post FTZ induction. Expression patterns of these genes were also examined in@ mutant embryos to obtain genetic confirmation of the interactions observed. Likefiz en and prd, odd appears to be directly activated by FTZ. In stage 5 embryos, ectopic expression of FTZ causes rapid expansion of odd, from its initiating pattern of six stripes (Figure 1.6A), to near homogeneous expression across the germband (Figure 2.68). In stage 6 embryos, odd is nonnally expressed in 14 evenly expressed stnpes (see Figure 2.6D). Ectopic FTZ causes an intensification of the primary odd stripes at this stage (Figure 2.6E). These stripes are derived from the original 7 stripes that overlapfr; stripes. In stage 7 embryos, these prima. stripes are not only intensified. but expand anteriorly as well (from about 1 ce11 wide to 7 cells wide; data not shown). The percentage of embryos responding to ectopic FTZ, at al1 stages tested. was about the sarne as the percentage of embryos that show ft,. autoregdation, en and prd activation and wg repression. Thus, FTZ appears to be an activator of odd at ail Rages tested. This positive relationship between FTZ and odd is consistent with the differences in odd expression obsewed in ji,- mutant embryos. Stripes of odd appear to be diminished in intensity in stage 5 embryos (Figure 2.6C), and primary stripes are weak or missing in stage 6 (Figure 2.6F) and 7 (not shown) embryos. Unlike prd and odd, the pair-rule gene slp is negatively regulated by FTZ: ectopic expression of FE! results in the differential repression of secondary slp stripes (Figure 2.6H). Again, the penerrance of repression at the 20 minute recovery the was about 60%. as has been measured for the other genes exhibithg direct responses. As might be exgected. slp stripes expand in fi= mutant embryos, filling the regions where FTZ is normally exptessed (Figure 2-61). Thus, as with wg, FTZ 99 Figure 2.6 FTZ activates odd and represses slp. Transcript patterns are shown, from top to bottom, for the pair-mle genes: early odd (stage 5: A-C), late odd (stage 6; D-F), and slp (stage 7: G- 1). Embryos on the lefl (A,D,G) are wild-type, embryos in the middle (B,E,H) are HSFTZ embryos heat-treated for 8 minutes and recovered for 20 minutes, and embryos on the right (C,F,I) arejïz- embryos. Note that the effects of ectopic FTZ are opposite to those of loss ofjz. Anows point to the affected stripes.

101 appears to act as a direct repressor of slp. This effect is likely exerted through the response elements or irans-acting factors that regulate secondary smpe expression.

Gooseberry (gsb) k an indirect target of FTZ

Genetic studies have suggcsted that the segment polarity gene gooseberry (gsb), Iike rvg, may be repressed by FTZ. In ftr mutant embryos, gsb stnpes expand into the regions where FTZ is normally expressed, fusing to form seven wide smpes (Figure 2.7B). To test whether this interaction is direct, we examined gsb expression in hsf2 embryos fied 20 and 35 minutes after a 4 minute heat shock. As shown in Figure 2.7C no response was observed 20 minutes post heat shock. However, changes were detected in embryos fixed 35 minutes post heat shock; stripes expanded into the ventral regions of odd-numbered parasegments (Figure 2.7D). Interestingly, this response is a positive one, in convast to the negative response predicted from expression infi' embryos. This apparent contradiction can be reconciled by postulating that gsb is indirectly regulated by FTZ, and chat different intermediary factors are involved in each case. The delayed nature of the response in HSFTZ embryos is consistent with this interpretation, and a likely intermediary factor is PRD. since gsb. Iike wg, appears to be activated by PRD (Cai et al. 1994). Infi' embryos, a Iikely intermediary activator is ivg, since wg also expands infi:'embryos (ingham et al. 1988) and appears to function as an activator of gsb (Li and NoIl 1993). An alternative explanation is chat FTZ has the ability to repress gsb directly. but that this effect is spatially limited to regions where fc is normaliy expressed. An exception would have to be made, however, in the anterior-most cell of eachfi stripe, where FTZ and gsb normally overlap.

2.5 DISCUSSION

Using kinetics to dktinguish between direct and indirect gene interactions

Distinguishing between direct and indirect gene interactions is necessary prior to studying the complex molecular interactions chat conûol them. In this study, we used a kinetic approach to identifL direct and indirect targets of ectopically expressed FTZ. An advantage of this approach is that it can be carried out in vivo at the time(s) chat the protein of interest is nonnally expressed. This is crucial for FTZ as, Iike many other transcription factors, it is dependent upon interactions with specific cofactors in order to recognize and regulate specific targets (Copeland et al, 1996; Guichet et al. I99T: Yu et al. 1997). Another advantage of the kinetic approach is that analysis is relatively fast and can be applied to a large number of putative targets with no pnor knowledge of their genetic relationships. When combined with other techniques, such as expression pattern analyses in mutant backgrounds and promoter deletional analyses, cesults can be conthed and mechanisms determined. 102 Figure 2.7 Indirect regulation of gsb by FTZ. Expression of gsb transcripts is shown in WT (A), pz-(B) and hsf2 ernbyos heat-treated for 4 minutes and recovered for 20 minutes (C) or 35 minutes (D).Infiz- embryos, gsb stripes expand into even-nurnbered parasegments (C), suggesting negative regulation byftz. However, no change in gsb expression was observed 20 minutes post-heat shock (C), and the secondary response (D) was positive.

Primary and secondary response windows

The resuits of our analysis with FTZ show that target genes respond to pulses of FTZ expression within two distinct temporal windows. Direct responses are 50% complete within about 18 minutes post heat shock. Indirect responses do not reach the same Ievel of response until 26 minutes post heat shock, Thus, the time behveen successive gene interactions, For the genes studied here. is just under 10 minutes. This tirne interval is shorter than the 10-15 minute approximation of previous studies (Manoukian and Krause 1992; Manoukian and Krause 1993; Saulier-Le Drean et al, 1998). Our analysis of protein and RNA Ievels indicates that the rate-limiting process in this

Factors determning primary and secondaty response limes

Within a transcriptional cascade. the time required between successive gene interactions wiI1 depend upon a multitude of factors. Thesr include the time required for transcription. RNA processing, transport. translation and protein localization. Ttie duration of these events wiII depend, in turn. on the size of the transcription unit and its products. The stabiiity of each gene product is alsa a factor- if there is significant variability amongst genes with regard to -ch oc these parameters, then mget gene responses wiIl €ail into a variety of overlapping temporal windows. This appears not to be the case with the genes investigated in this study. 105 Al1 of the genes investigated here encode transcripts that are -2 kb in length, have few or no introns (e.g. Kuroiwa et al. L984; Ish-Horowicz et al. 1985; Poole et al. 1985; Kilchhen et al. 1986; Macdonald et al. 1986; Baumgartner et al. 1987; Drees et al. 1987; Gergen and Butler 1988) and have similar half-lives (between 5 and 10 minutes) (Edgar et al. 1986; Manoukian and Krause 1992; Manoukian and Krause 1993). In addition, the protein products of each gene are on average -50 kDa, and have similar half-lives (between 5 and IO minutes) (Edgar et al. 1987; Kellerman et al. 1990; Manoukian and Krause 1992). This means that the time required for synthesis of each gene product should be very simila. consistent with the results of this and previous studies. Importantly, the tirnes required for protein import (activated targets) and removal (repressed targets) frorn the nucleus appear to be vimally identical, indicating that protein synthesis and degradation rates are also closely matched. This matching of synthesis and degradation rates may have evolved to ensure chat activation and repression events can proceed in parailel. In pathways such as this, where progression down the hierarchy from one class of genes to the next must be kept in synchrony. and at the same time progress as npidly as possible, this optirnimion and matching of rates may be crucial. Further support for matched rates of activation and repression cornes hma study in which the Hairy protein was converted from a repressor into an activator and expressed ectopically (Jimenez et al. 1996). Activation of target genes that are nomally repressed by Hairy was achieved within the same 10 minute recovel period. Although these results show that rarget gene responses can be readily separated into distinct temporal windows, variations in the curves within these windows may occur. not just because of synthesis and degradation rate differences. but also due to differences in the molecular mechanisms underlying each interaction. For example, the curve representing the response of prd differed from those of the other direct responses, most IikeIy because it is the only response that occurs whereJz autoregulation does not. Where autoregulation occurs, the responses are enhanced and maintained by the endogenous fc gene product. Another important point to keep in mind is that direct interactions. as defined here, need not invotve direct contact between FTZ and its target regulatory elements. Effects could also be relayed via a series of protein interactions, the complexity and nature of which could vlthe speed or magnitude of the response. However. hmthe data obtained here, it seems that these 'Ipes of molecular variations. if they are occumng, cause temporal shifts that are minor in comparkon to the significant intervals of haie required for the transcription, transiahon and Iocalization of intermediary gene products.

Direct turgets of FîZ

Our results suggest that genes directly activated by FTZ include the endogenousfi gene, the pair-nile genes prd and odd, and the segment polarity gene en. Genes that appear to be directly tepressed by FTZ include the pair-de gene sip and the se-ement polarity gene wg. Al1 skloci responded with simiIar kinetics. The spatial and regulatory relationships amongst these genes are summarized in Figure 2.8. The importance ofjz gene autoregulation and en activation is well described in the literanire (Monta and Lawrence 1975; Kornbeg 1981: Hiromi and Gehring 1987), and the data arguing chat both interactions are direct were already quite persuasive (Hirorni et al. I985; Hiromi and Gehring 1987: Desplan et al. 1988; Han et al. 1989; Pick et al. 1990: Schier and Gehring 1992; Florence et al. 1997). There is less support for the activation of prd and the repression of wg (Ingham et al. 1988; Ish-Horowicz et al. 1989; Baumgartner and NoIl 1990; Gutjahr et al. 1994). Regulation of the odd and slp genes by FTZ has not been previousIy noted. Support for the less-well-characterized interactions will be provided below, as well as their Iikely relevance to the segmentation process. Previous observations supgesting a positive regulatory relationship between FTZ and prd include the overlapping expression patterns of the nvo genes, the presence of FTZ-binding sites within the prd promoter, and inappropriate narrowing of prd and prd reporter gene stripes in& mutant embryos (Kilchherr et al. 1986: Baumgamer and Noll 1990; Gutjahr et al. 1994). Our data support the occurrence of a direct interaction krween the two genes and suggest that the most important function of this interaction is to maintain prd expression in the posterior regions of even- numbered parasegments. where prd is required for the activation of wg (this smdy and Ingham et al. 1988: Copeland et al. 1996). Interestingly, prd was the onIy gene that responded to ectopic FTZ in the anterior regions of odd-numbered parasegments. This suggests that the molecular interactions goveming activation of prd rnay diffa from those involved in other FïZ target gene responses. Our data also indicate that FTZ is an effective activator of odd. This is consistent with the coinciding spatial expression patterns of the two genes during stages 5 and 6 (Coulter et al. 1990: Manoukian and Krause 1992), and the reduction in Levels of odd expression infi mutant embryos (Figure 3.K, F). The requi~mentO€ FTZ as an activator of odd beyond stage 5 also suggests a unique mechanisrn for the regulation and kinetics of odd stripe narrowing. During stage 6 and 7. srripes of odd andfi: become narrower as expression is lost rrt their posterior edges (see Figure SB). The narrowing off= stripes occurs first as a result of repression by ODD (Saulier-Le Drean et al. 1998). The resolution of odd srripes. which foliows immediately thereafter, may be due to prior Ioss of FTZ. Thus! ODD appears to be instigating its own removal by repressing its own activator. The participation of other genes rnust be invoked, however? to explain why FTZ faiIs to maintain odd expression in the anterior-most FTZ-expressing cells of each FTZ stripe. and why add continues to be evpressed just behind each FTZ stnpe. well afier FTZ has been removed from those celts. The broadening of wg stnpes in)= mutant embryos (Ingham et al. t 988) and repression of wg in HSFTZ embryos (Ish-Horowicz et al- 1989) had previously suggened that wg rnay be a target of FTZ. However. because of the very different nanue of this interaction, opposite to that of the weli-characterized activation finctions. hrdier confirmation of a direct interaction was sought- 1 O7 Figure 2.8 Direct gene-regulatory interactiaos triggered by FTZ. A schematic representation of three consecutive parasegmental intervals is shown. Each parasegrnent is designated by text and solid black lines at the top of each panel as well as vertical dashed lines. Spatial gene expression is presented in the form of boxes. Odd-nurnbered parasegments are distinguished by expression of even- skipped (me) whereas even-numbered ones expressfishi tarazu Uiz) (blue box). Positive interactions triggered by FTZ are shown in green and the negative ones are shown in red. During early stage 5 (A), FTZ autoregulates its own expression as weIL as positively regulates odd andprd (green solid arrows). These interactions are believed to be direct. Lam, during stage 6 (B), FTZ continues to autoregdate and begins to activate even-numbered stripes of en (green solid arrows). At the same time. it represses (red blunt-ended lines) odd-numbered stripes of wg and secondary stnpes of slp, preventing their expression in even-nurnbered parasegments. 000 # PS EVEN # PS 1 ODDX PS 1

l ftz I I I I odd I I

000 # PS EVEN # PS ODD # PS

I eve I il b n I I I O I O I I I I l I I I 1 I I I I I 1O9 The near identical kinetics of wg repression andfi,- autoregdation are consistent with the ability of FTZ tu function as both an activator and a repressor of transcription- Our ectopic expression studies suggest chat FTZ acts on wg via regdatory elements that apecificaily regulate odd-numbered stripes. and that it is these snipes that expand in fiz mutant embryos. The repression by FTZ of slp suipes, as well as wg smpes. indicates that repression is an important aspect of FTZ function. Negative regulation of slp by FTZ is consistent with the non- overlapping expression patterns of the two genes (see Figure 28B) and with the broadening of stp strîpes thac we observcd in$= mutant embryos (Figure 7.61). Our data suggest that the normal role of FTZ is to prevent posterior expansion of secandary slp smpes into even-numbered parasegments. In summary. FTZ conbols even-numbered parasegment formation and polarity by a number of direct and indirect interactions that span a time interval of at least 30-60 minutes (surnmarized in Figure 1.8). FTZ activates en expression at the anterior edges of each parasegment directly. It also activates en indirectly by repressing the en repressor slp. At the metirne, FTZ exludes expression of wg throughout ail but the posterior edge of each even-numbered parasegment. Facilitation of wg expression in these ceIIs is brought about indirectly by the activation of prd. ExcIusion of en and wg expression within the middle of these parasegments is controlled indirectly by activation of odd. which encodes a repressor of both genes (Manoukian and Krause 1993: Mullcn and DNardo 1995: Saulier-Le Drean et al. 1998). The activation of odd also results infi= stripe narrowing and funher polarization of each even-numbered parasegment. Thus. FTZ has many hnctions throughout each even-numbered parasegment. This conctusion differs tiom that of an earlier study in which it was suggested that the only real requirement of FTZ may be to activate en at the anterior edges of even- numbered parasegments (Lawrence et al. 1987). Five of the eleven genes tested in this study did not respond to ectopic FTZ with the kinetics of direct carget genes. These include the pair-nile genes me, h. nm and opo. and the se-ment polarity -eene gsb. Except for gsb. secondary responses were also absent, This suggests that effects observed in previous genetic studies suggesting that me (KeIIerman et al- 1990) and run (KIingler and Gergen 1993) rnay be regulated by FTZ. are most likely due to indirect eflects. We cannot exclude the pssibility, however. that such interactions do occur and that they were missed here due to an inability of FTZ to regulate these genes outside the normal domains of FïZ expression.

Rinetic assament of genetic hierarchies

A major goal of the pst genome-sequencing era will be to or-eanize ail newiy identified genes into comprehensive genetic circuits. Powerfiil new methods wilI be required to perfom this task efficientiy and accurateiy. One new approach that shows a _gtdeal of potentiaI is the use of high- density oligonucieotide arrays (DeRisi et al. 1997 Wodicka et al, 1997). Like the kinetic approach described here, this is a functional assay in that it has the potentiai to identie gens that respond directly to specific cellular stimuli or programmed neps in differentiation. As currently employed 110 however, changes in gene expression may not represent direct interactions. For example, a recent analysis of genes affected by mutations in components of the general transcription machinecy used a 45 minute recovery time following inactivation of temperanite-sensitive alleles (Holstege et al. 1998). Based on our results, 45 minutes is enough time for several successive gene interactions to take place. To identie direct targets, a kinetic approach, much as described here, would have to be employed. Genes could be tumed on or off using a variety of techniques. and then the responses monitored at intervals thereafier. Genome-wide analyses of this type can then be used to comprehensively solve cornplex hierarchies such as the segmentation gene hierarchy studied here. Once direct circuitry is established, the more time-consurning business of establishing the relevance and molecular mechanisms underlying each interaction can follow. Chapter 3:

Mechanisms regulating target gene selection by the homeodomain-containing protein Fushi tarazu

A similar report was published in Andnej Nasiadka, Allan Grill and Henry M. Krause, Development: 127, 2965-2976 (2000)

*AhGril1 assisted with cuticle preparations and the analysis of mutant phenotypes produced by HSFTZ and HSFTmP 16 embryos. Horneodornain proteins are DNA-binding transcription factors that control major developmental patterning events. Alrhough DNA binding is mediated by the homeodomain, interactions with other transcription Factors play an unusuatly important rolc in the selection and regulation of target genes. A major question in the field iç whetber these cofactor interactions select target genes by modulating DNA binding site specificity [selective binding model), mscriptionaI activiiy (activity reguiation rnodel) or both. A related issue is whether the number of mget genes bound and regulated is a srnall or large prcentage of genes in the genome. In this study, we have addressed these issues using a chimecic protein that contains the strong activation domain of the vin1 VP16 protein fused to the Drosophilu homeodornain-containing protein Fushi tarazu (FTZ). We find that genes previously thought not to be direct targets of FTZ rernain unaffected by FTZVP16. Addition of the VP16 activation domain to FTZ does, however, allow it to regulate previously identified target genes at times and in regions that FTZ alone cannot [t also changes FTZ into an activator of two genes that it norrnally represses. Taken together. the resuhs suggest that FTZ binds and regulates a relatively limited number of target genes. and that cofactors affect target gene specificity primarily by controlling binding site selection, Activity regulation then fine-tunes the temporal and spatial dornains of promoter responses, the magnitude of these responses. and whether they are positive or negative.

Homeodomain proteins are arguably the most important class of transcription factors in early developmental panerning (McGinnis and Knrrnlauf 1992). As with most other DNA binding transcription factors, they require interactions with other cofacton in order to recognize and regulate specific target genes (Mann and Chan 1996; Biggin and McGinnis 1997; Mann and Affotter 1998). These interactions, however, are particularly important for homeodomain proreins because of the relatively low DNA binding specifici~of the homeodomain. Most homeodomains bind sequences with an ATTA core motif (Gehrinp et al. 1994b) and these motifs are abundantly distributed thmughout the genomes of eukayotic organisms (Desplan et ai. 1985 Appel and Sakonju 1993; Walter and Biggin 1996). A classic esample of an interaction that overcomes this problem of limited specificity is the interaction between the yeast homeodornain pmteins MATal and MAT&, and their cofactor MCMI. [n different combinations, these thcee proteins recognize unique binding sites and coordinate the differential gene expression patterns that define alternate yeast maung types (Goutte and Johnson 1988; Keieher et ai. 1988; Keleher et al. 1989; Dranginis 1990). An dternative mategy used by some homeodomain proteins is the incorporation of additional DNA binding dornains (Bopp et al, 1986; Sturm and Kerr 1988; KarIsson et al. 1990; Treisman et al- 1991). Nevertheles, these compound homeodomain proteins also appear to require cofactor interactions in 1 13 order to bind and regulate specific target genes (Knstie et al, 1989; Stem et al. 1989; Kristie and Sharp 1990; Zwilling et al. 1995; Copeland et ai. 1996). In this study, we focus on the Drosophila homeodomain-containing protein Fushi tarazu (FTZ) and cofactors that are required during early ernbryogenesis. FTZ is a member of the Q5O class of homeodomain proteins, meaning that it has a glutamine at position 50 of the horneodomain (Treisman et al. 1992). Q50 horneodornain proteins represent the largest class of homeodornain proteins, and because the residue at position 50 is the main determinant of specificity, these proteins al1 bind in vitro to nearly identical sites (Despian et al. 1988; Hoey and Levine 1988; Laughon et al. 1988; Treisman et al. 1992). In the embryo, however, they control very different aspects of developrnent. ETZ, for exarnple, belongs to the pair-mle class of segmentation proteins (Nusslein- Volhard and Wieschaus 1980; Wakirnoto and Kaufman 1981; Wakirnoto et al. 1984). It con~olsthe patterning of alternate segmental regions by binding and regulating target genes that are expressed in these regions. In culmred cells, FTZ acts as a relatively strong activator of both synthetic and actual target gene promoters (Jaynes and OfFarrell 1988; Han et al. 1989; Winslow et al. 1989). [n the embryo, however, it appears to act as both an activator and repressor of transcription (Ingham et al. 1988; Ish-Horowicz et al. 1989: Nasiadka and buse 1999). Several Iines of evidence emphasize the importance of cofactors in the binding and regulation of these target genes. First, when reporter genes with synthetic prornoters that contain onIy consensus FTZ binding sites are introduced into embryos, they fail to respond to FTZ (Vincent et al. 1990). Second, homeodornain-deleted versions of FTZ that are incapable of binding to DNA are still able to regulate FTZ target genes, and can even rescuefiz-dependent segmentation (Fitzpatrick et al. 1992; Copeland et al. L996 Hyduk and Percival-Smith 1996). Thus, interactions with other transcription factors are both required and suficient for the recognition and regulation of specific target gene promoters. To date. two cofactors have been identified that are required for proper FTZ function in vivo, the nuclear receptor protein FTZ-Factor 1 (FTZ-FI) (Guichet et al. 1997; Yu et al. 1997) and another homeodornain-containing protein Paired (Pm) (Copeland et al. 1996). Loss of FTZ-FI expression during early embryogenesis causes a segmenta1 phenocype that is indistinguishable from that of FTZ (Guichet et al. 1997: Yu et at. 1997). This phenotype is due to an inability of FTZ to regulate both positively and negatively regulated target genes (Guichet et al. 1997). A direct interaction between FTZ and FTZ-FI is required for these prccesses. Interestingly. FTZ-F1 is not essential for the ability of FTZ to autoregulate its own expression (Guichet et al, 1997), indicating the existence of additional cofactors, Two models have recently been put forward to explain how cofactors contribute to functional differences amongst QSO homeodornain proteins. The first mode1 suggests that cofactors target each protein to different sets of genes by conmiuthg to their sequence-specific DNA binding properties (Mmand Chan 1996: Mann and Affoiter 1998). Some classic examples in support of this mode1 include interactions on specific response elements between members of the Extradenticle 114 (EXD)/PBX protein farnily and homeotic selector proteins such as Ultrabithora~(UBX) (Chan et al. 1994b; van Dijk and Murre 1994; Chang et al. 1995). Similarly, it has ken shown that FTZ and FTZ-F1 bind to adjacent sites on a functionai reporter with as much as a 100-fold increase in afinity (Yu et aI. 1997), suggesting that FTZ-FI plays an important rde in recruiting FTZ to these specific sites. Indeed, one of these reporters has also been tested with a homeodomain-deleted version of FTZ (Hyduk and Percival-Smith 1996) and responds appropriately, indicating that cofactor interactions are sufficient for recniitment. The second model proposes thab instead of moduiating DNA binding properties, cofactors alter the ability of homeodomain proteins ta regulate promoters to which they are already bound. These binding sites are thuught to inchde al1 or most of the sites bound in vitro, meaning that these proteins would be extensively bound to most accessible promoters. Specificity of action is achieved by cofactor interactions that modifi the reguiatory potential of particular molecules in a site-specific fashion. Much of the support for this 'activity regulation' model comes from in vivo cross-linking studies to a number of promoters (Walter et al. 1994; Cmand Biggin 1999). The two Q50 homeodomain proteins, FTZ and Even-skipped (EVE), were detccted on most of the restriction fragments tested with similar relative levels of occupancy (Walter et al. 1994). lt was also found that most ernbryonically expressed genes show some degree of pair-rule periodicity in their patterns of expression (Liang and Biggin 1998). consistent with the possibility of direct regdation by EVE or FTZ. Additionai support for this modei cornes from the observation chat the homeodomain protein Deforrned (DFD) can bind to monomer DNA sites in vivo independently of cofactors such as EXD, but that this binding is apparently not suficient for gene activation (Li et al. I999a). A major objective of this study was to search for additional evidence in support of either the selective binding or activity regulation models of FTZ cofactor function. .4 method was sought that wouid a1iow the monitoring of native target genes in an in vivo sening. This was achieved by expressing a chimeric FTZ transcription factor that contains the potent acidic activation domain of the herpes simplex virus protein VP16 in early embryos. The VP16 activation domain is fully modular in tùnction and is capable of activating vinuaiiy any eukaryotic promoter in organisms ranging from yeast to man (Triezenberg et al. 1988). It can even conven transcriptional repressors into activators (Jimenez et al. 1996; Jimenez et al. 1997; Kramer et ai. 1999). The strength and autonomy of the VP16 activation domain is explained in part by its abitity to activate transcription via multiple mechanisms and at multiple Ievels. These include an abiiity to rec~ithistone ricetyitransferases (Utley et al. 199%;[keda et al. l999), to recntit ATP-dependent chrornatin remodeling complexes (Neety et al. 1999) and to interact directly with components of the RNA polymerase holoenzyme (reviewed in Triezenberg 1995; Stargell and Struhl 1996). Taken together, these properties suggest that. when bound to FTZ, the VPI6 activation domain should be able to activate promoters that are within regdatory distance of ET2 binding sites independently of FTZ cofactors. nus, if Fi2 hnction is stnctly regulated at the lever of DNA binding, FTZVPl6 should teplate the same set of senes as FTZ. These genes mi& be activated ta higher levels, but this may 115 have little consequence on development since raising the levels of FTZ by as much as 2 fold has no apparent eff'ct on segmental patteming (Lawrence and Pick 1998; Nasiadka and Krause, unpubrished observations). Conversely, if FTZ is widely dismbuted on most accessible promoters, as suggested by the activity regulation model, expression of FTZVP16 should result in the activation of genes that are not normally regulated by FTZ. Our results suggest that FTZ specificity is primarily regulated at the level of DNA binding. Activity regulation then fine-tunes the cimes and regions that bound promoters can be regulated, as well as the magnitude of responses and whether they are positive or negative.

33 MATERULS AND METHODS

Construction of Fm-VPl6 fusion and control consiiucis

FTZVPI6 was made by rernoving a HaeIII - EcoFü Fragment from the C-terminal end of):? and replacing it with a Salt - EcoRI restriction fragment that encodes the C-terminal activation domain of VP16 (Campbell et al. 1984). An analogous exchange was made using a homeodomain- deleted version of FTZ (FTZAHD) (Fitzpatrick et al. 1992) to make FTZAHDVP16. FTZAC, was made by rernoving the C-terminal HaeIIl - Ecoiü thgrnent offir. FTZANVPI6 was made by removing sequences from the FTZVP16 consmct N-terminal to the Xhot site chat is located in the N-terminal portion of the fi homeobox. For the transfection studies, each of the constructs was inserted into the expression vecror pPac (Krasnow et al. 1989). The same fragments were also insened into the P element vector pHT4 (Schneuwly et al. 1987) for embryonic expression under control of the hsp70 promoter. As a consequence of the cloning stmtegies, non-FTZ amino acids were introduced at the termini of the FTZ polypeptides. In the FTZ, FTZAHD, FTZVP16, FTZAHDVP16 and FTZAC constnicts, the N-terminal sequence is MDPEFIKEEKLTMRDP"(Q3. In FTZANVPI6. the N- terminal sequence is MDPEFELGTRGSSRV-(E)273, Non-FTZ arnino acids at the C terminus of FTZAC are FTZ(E)33 7-GGILV. The sequence at the FTZ-VP16 interface also includes the additional amino acids: FTZ(E)337-GGIR-(T)48OVP16.

Cell culture, transfuctions und CAT arthiîy mqs

Drosophila Schneider Iine 2 (S2) cells (Schneider 1972) were grown as descnbed previously (Di Nocera and Dawid 1983). Transfections were performed ushg the calcium phosphate technique described by Krasnow (1989). DNA mixtures included 2 pg of FTZ-expressing plasmids and 0.4 pg of the reporter plasmid pNP6CAT (Jaynes and O'FarrelI 1988). Celis recovered fiom each of the duplicate plates were sptit, with one half used for western blotting and the other half lysed for CAT activity measurements. CAT assays were performed as described by Seed and Sheen (1988). Ceneration of transgenicflies

Transgenic flies were obtained by injecting the P element constmcts descnbed above into ry'06 embryos as descnbed by Rubin and Spradling (1982). Transgenic Iines obtained for pHT4FTZ and pHT4FTZAHD have been described previousiy (Fitzpatnck et al. 1992). Two independent lines with single copy insertions, one on the first chromosome and the other on the third, were obtained for pHT4FTZVP 16. Both lines are non-viable when homozygous and maintained as heterozygous stocks over either FM6 or TM3. Two independent Iines with single copy insertions were obtained for pHT4FTZAHDVP16; one inserted on the second chromosome and the other on the third. These lines are also homozygous inviable, and are kept over CyO and TM3, respectively. One hornozygous line was obtained for pHT4FTZAC; a single copy insertion on the third chromosome. Finally, two independent homozygous lines were produced for pHTJFTZANVP16, one with a single copy insertion on the first chromosome, and the other with a single copy insertion on the second. For western blots, FTZ polypeptides were detected using either anti-FTZ or anti-VPI6 polyclonal antibodies and then alkaline phosphatase-coupled secondary antibodies.

Cuticle preparation and analysis

Embryos for cuticle preparations were collected on apple juicelagar plates for 30 minutes. Eight collections each of wild-type, HSFTZVPI6 and HSFTZ embryos were made and aged for different periods such that a series of 30 minute intervals was obtained. Each interval overlapped the preceding and the following one by IO minutes (Le. I:40-210, 200-230, 220-250, etc.), spanning from 1:JO to J:30 AEL. After aging at 2j°C for the times indicated, embryos were heat-treated by immersion in a 36S°C water bath, usually for 8 minutes, then rinsed with 2j°C water and transferred to apple juice plates for hrther aging. Cuticies were prepared as described in Saulier-Le Drean et al. (Saulier-Le Drean et al. L998).

In situ hybridization and immunocyiochembhy

Embryos were collected on appIe juice/agar plates for I hour, aged until either 200-3:00 or 230-3:30 hours old, and then heat-treated for either 8 or 4 minutes at 36.3"'. After heat-shock, embryos were allowed to recover for 25 minutes, fixed in 4% formaldehyde and stored in methanol. The effect of the heat treatment was determined by following target gene expression in heat-treated Oregon R ernbryos, treated in paraIlel. In situ hybridization to whole-mount embryos using digoxigenin-labeled DNA probes \vas performed as previously described (Manoukian and buse 1992). PIasmids and primers for probe preparation have also been described elsewhere (Nasiadka and Krause 1999). Following hybridization, 117 probes were visualized using alkaline phosphatase-conjugated anti-digoxigenin antibody (Boehringer Mannheim). Cellularizing embryos were precisely staged by monitoring the position of inward- migrating plasma membranes. Antibody staining was performed as previously described (Manoukian and Krause 1992). Expression of FTZ deletion and fusion consaicts in whole-mount embryos was detected with anti- FTZ polyclonal antibodies (Krause et al. 1988).

3.4 RESULTS

Enhancement of FTZ activity by the VP16 acîivution domcrin

The C-terminal 76 amino acids of FTZ encode a modular, glutamine-rich activation domain capable of eficiently activating transcription in cultured Drosophila S2 cells (Fiapatrick et al. 1992). In the embryo, deletion of this region causes onIy a moderate loss of activity that is largely compensated for by higher levels of protein expression {Kellennan et al. 1990: Fitzpatrick et al. 1992). Hence, replacement of this region with the acidic activation domain of VP16 should result in a fusion protein that retains FTZ specificity while gaining the ability to activate target genes with greater autonomy and effectiveness. If so, then target genes might be activated at times and in places where the non-fused protein is incapable of acting. [n addition, target genes that are normally repressed by FTZ might now be activated. Consmcts generated to test this hypothesis are shown in Figure 3.1A. In addition to the FTZVP16 fusion protein, several control constructs were made, including full-length FTZ, FTZ with the C-terminal 76 amino acids deleted (FTZAC), the VP16 activation domain fused to the C-tenninal 213 of the FTZ homeodomain (FTZûNVP 16) and homeodomain-deleted versions of FTZ (FTZAHD) and FTZVPl6 (FTZAHDVP16). Transient expression in Drosophila S2 cells was used as a first approximation of whether FTZVP 16 retained the same DNA binding specificity as that of FTZ. FTZ constructs were subcloned under control of the acrin jC promoter (Krasnow et al. 1989). The reporter gene NPGCAT, which contains concatamerized FTZ binding sites upsmam of a basal promoter (Desplan et al. 1988; Saynes and O'Farrell 1988), was used to monitor cranscriptiona1 activity. Al1 proteins were expressed at similar levels when assayed by western blot anaIysis (data not shown). Results, shown in Figure 3.IB, show that remova! of the FTZ C-terminal activation domain results in an approximate six-fold drop I I8 Figure 3.1. Structure, activity and expression of FTZ denvatives. (A) Schematic representation of FTZ constnicts: The FTZ poiypeptide is presented as an open box with the homeodomain denoted in black. The VPI6 activation domain is shown as a gray box. Internai deletions span the FTZ amino acid residues indicated by the numbers above. (0) Relative transactivation accivity of FTZ and FTZ derivatives in Drusuphila Schneider line 2 (S2) cells. Cetls were transfected with the CAT (chloramphenicol acetyltransferase) reporter plasmid pNP6CAT alone (-) or together with the FTS-expressing construct indicated. Transactivation activity of FTZ was set 100% and used as a basis for calculating relative activities of FTZVP 16 and other FTZ- derivative constnicts. The activities shawn are the average of three transfections. Correspondhg standard deviations are indicated. Addition of the VPI6 activation domain significantly increases the activity of FTZ, while removat af the homeodomain reduces activity to near background levels. (C) Western blot analysis showing FTZ and FTZVPI6 expression levels in transgenic HSFTZ (lanes L, 2) and HSFTZVPI6 (lanes 3, 4) Iarvûe. Proteins were extracted from non-heat shocked (Ianes 1, 3) and heat shocked (lanes 1, 4) larvae and detected using anti-FTZ (lefi panel) or anti-VPI6 (ri& panel) antisera. FTZ and FTZVP 16 migrate with apparent molecular masses (LM,)of about 70 and 68, respectively. Both polypeptides are expressed at similar levels and are predominantly full-length.

120 in FTZ activity (the activity of FTUC is 16% of the full-length FTZ activity). Replacement of this region with the VP16 activation domain results in a two-fold increase in activity. This increase likely reflects the relative activities of the two activation domains; the FTZ C-terminal activation domain is a relatively strong one (Fitzpatrick et al. L992), while the VP16 activation domain is arguably the most potent domain tested (Triezenberg et al. 1988). As has been previously shown, removal of the FTZ homeodomain drops transcriptional activity in this assay to near basal levels (Fitzpamck et al. 1992). This was also me for the homeodomain deleted control constnicts FTZmVP16 and FTZANVP~~.Similar results were obtained using another FTZ reporter gene (UbxCAT), and no responses were observed when a RpOIter gene with no FTZ binding sites (TATACAT) was used (data not shown). Taken together, these data show that FTZVPl6 mains the DNA binding specificity of FTZ. and that in cultured cells, it is a bener activator of mscription. All constructs shown in Figure 3.IA were next subcloned under control of the hsp70 gene heat-inducible promoter (Schneuwly et al. 1987) and introduced into flies by P element-mediated genn-line transformation (Rubin and Spradling 1982). At least two transgenic lines were obtained for each constnict (see Materials and methods). To ensure that each constmct was correct, and to compare levels of protein expressed, larvae were collected fiom transgenic lines. heat pulsed for 1 hour, allowed to recover for 30 minutes. Iysed in SDS PAGE loading buffer and analyzed by western blotting. Figure 3.lC shows that equivalent levels of FTZ and FTZVP16 are expressed. and that no major degradation products are observed. Each of the other pmteins is also expressed at similar levels (Fitzpatrick et al. 1992: Copeland et al, 1996 and data not shown). Protein levels were also assessed in embryos by irnmunocytochemistry. With an 8 minute heat pulse at 36'C. the standard duration of heat shack for our HSFTZ lines (Nasiadka and Krause 1999), cellular levels of ectopic protein expression were similar to one another. and at lest two-fold lower than the cellular levels of endogenous FTZ protein (data not shown). The same induction conditions were used in al1 of the results that follow, except when monitoring genes whose expression is particularly sensitive to heat shock (even-skipped (rw),mr (run) and goosebeny (gsb)). For the latter, heat pulses of 4 minutes were used.

Effects of FTZVPI 6 on segmental paneming

FTZVP16 activity in vivo was first assessed by inducing ectopic expression at various developmental time points and then exarnining patteming defects in the cuticles of mature embryos. It has been show previously that ectopic expression of FTZ between 2.5 and 3 hours aller egg laying (AEL) gives rise to a pair-rule cuticular phenotype in which regions normally derïved from odd-numbered (@-&dependent) parasegments are missing (Stnihl 1985; Ish-Horowicz and Gyurkovics 1988; Ish-Horowicz et ai. 1989). To compare the patteming activities of FTZ and FTZVP16, hansgenic embryos were heat pulsed befoq during and after this the interval, and then cuucles were prepared and examined. 121 We fm scored the number of wild-type cuticles obtained from each Iine as a tünction of the tirne of heat shock. As can be seen in Figure 328, induction of both FTZ and FTZVP16 causes a substantial decrease in the number of wild-type cuticles, with the mavimal decrease centered at about 2.5 hours AEL. The two curves differ, however, in the number of non-wild-type cuticles obtained at earlier and later times. FTZVP16 is able to induce patterning defects with a modest increase in effïciency at both earlier and later stages of development. Next, we closely examined the mutant cuticles to see if the FTZVP16-induced phenotypes were either similar to those induced by FTZ or novel. The most frequently observed phenotype for both constructs is the "anti-fi-" pair-ntle phenotype previously described by Stnthl (Struhl 1985 and Figure 3.2C). In both the HSFTZ and HSFTZVP16 lines, the frequency of this phenotype (deletion offrz-hdependent parasegments) peaks at about 50% when induction is initiated at about 2.5 hours AEL (Figure 3.2D). The HSFTZVP16 line differs, however, in that this phenotvpe can be induced a little earlier and with a somewhat higher eficiency. Further examination of the FTZVP16-induced cuticles revealed several novel phenotypes not found amongst the FTZ-induced cuticles. These less abundant phenotypes are show in Figure 3 2, panels E, G, 1 and K. The first phenotype (Figure l2E) is characterized by fusions benveen alternate denticle belts. Most of these fusions (appron. 80%) occur in posterior regions. Disruptions of terminal structures are also observed in each of these cuticles. This phenocype peaks when heat pulses are administered between 1:40 and 220 AEL (Figure 3.2F). Although the non-hsed FTZ protein appears to generate a weaker fom of this phenotype, its penetrance and severity are substantially lower than that induced by FTZVPI6, The second FTZVP 16-specific phenotype can best be described as an extreme pair-wise fusion of denticie belts with the remaining denticles arranged in minor-image symmetry (Figure 31G). This phenotype peaks with a fiequency of about 15% at 240-3:10 AEL (Figure 3.2H). The third phenotype (Figure 3.21) exhibits deletions of both naked and denticle belt-containing portions of each segment. As with the previous phenotype, the rernaining portions of each denticle belt appear to be duplicated with mirror image symmeay. This pattern is prevalent when heat pulses are administered at 3:OO-330 AEL with a mauimum fkquency of about 17% (Figure 3.25). The final pattern (Figure 3.X)occurs when heat putses are provided relatively late (3:304:00; Figure 3.3L) in the responsive window. All denticle kltç are deleted in this phenotype. The segmental phenotypes described were not observed in heat shocked control embryos, nor in FTZhNVP16 embryos. FTaC embryos yielded only wild-type and anti-fi=phenotypes, with the latter at a lower penetrance than those induced by tiill-length FCZ. Taken togeiher, these controls 122 Figure 3.2. Cuticular phenotypes caused by ectopic FTZVP16. FTZ and FTZW16 were expressed ectopically in differentially staged embryos (aged between 150 and 430 AEL), and the embryos allowed to develop for 34 hours to secrete cuticle. Cuticular phenotypes obssrved are show on the lefi (dark-field photomicrographs: A,C,E.G,i,K). Curves on the right (B,D.F,H,J,L) show the frequencies (Y axes) at which each of the phenotypes on the left was observed when induced at the times shown (X axes). A minimum of 700 cuticles were scored for each of the indicated times. Values obtained using HSFTZ embryos are indicated by triangles, and dues obtained using HSFTZVP 16 embryos are indicated by squares. Although their effects are similar. FTZVP16 was able to induce novel phenotypes during times at which FTZ had no effect.

124 show that the novel phenotypes generated by FTZVP16 are specific and most likely due to the enhanced abilities of the chimeric protein.

The changes underlying the cuticle phenocypes described above can best be understood by monitoring the eariiest changes in FTZ target gene expression patterns. These were monitored by in situ hybridization in cellularizing and gasuulating embryos. Ernbryos were fxed 35 minutes afier a brief 8 minute hear shock to ensure thar the effects observed are direct (Nasiadka and Krause 1999). The best-characterized target of FTZ is thejk gene enhancer (Hiromi et al. 1985: Hiromi and Gehring 1987; Pick et al. 1990: Schier and Getiring 1992; Schier and Gehring 1993a). Multipte FTZ binding sites are present on the#: enhancer, and reporter genes carrying portions of thepz enhancer faiI to respond when either the sites or FTZ is inactivated. Conversely. ectopic expression of FTZ results in expansion of endogenous fc smpes (Mi-Homwicz et al. 1989) and reporter gene stn'pes controlled by these FTZ-responsive elements (Hyduk and Percival-Smith 1996). Figure 3.3 shows that ectopically expressed FTZ begins tr, effectively activate the endogenous& gne during earIy stages of cellularization (early stage 5. ?00-230 AEL). Endogenous fi expression is expanded hman irregular set of initiating stripes (Figure 3.3A) into a single broad band that fills the trunk portion of the embiyo (Figure 3.3B). FTZVP16 has a similar effect except that induction is more robust and unifam (Figure 3C), and begins about 10-15 minutes eariier (not shown). Neither protein, however. is able to activatejiz in the terminal regions of the ernbryo. As cellularization nears completion,fi= stn'pes are well-established and about 4 cells wide (Figure 3.3D). At this cime (late stage 51, ectopically expressed FTZ can no longer activatefi throughout the trunk: the pattern remains clearly striped (Figure 33E). in contrast. FTZVPI6 retains the abiliy to effectively activatefi throughout the mk(Figure 33). 3y the tirne cellularization is complete (stage 6). endopenousjk srripes are beginning to namw (Figure 33G). When FTZ is expressed ec~opically.eachJz sûipe widens anteriody by approximately one ceIl in width (Figure 3.3H). As observed at the earlier stage, FTZVP16 has a similar but more robust effect. Endogenous sûipes widen by about the same amount anterioriy. but are noticeably more intense (Figure 5.3). in addition, a novel set of weak stripes (marked by arrowheads) appears imrnediately behind each of the stronget smpes. These occupy the posterior regions of even-numbered paraseeents. Similar responses are observed when FEor FTZVP16 is induced during later stages of normal fi= expression. Ectopic FTZ causes only a modes widening of endogenousfe stripes (Figure 23K), while FTZVP16 has the ability to induce additionai fi= smpes in the posterior regions of even- numbered parase-ements (Figure 3.3L). This capaciy perdures somewhat beyond the tirne thatfi,. is normal1 expressed and beyond the thne that ectopic FTZ has any effect (Figure 33M-O). 125 Figure 3.3 Effects of ectopic FTZ and FTZVPl6 on the endogenous frr gene.fr,. expression panems are shown at five different stages: the onset of cellularization (A-C), the end of ceiIularization CD-F), Iate gastruiation (G-1),early genband extension (J-L) and advancrd genn band extension (M-O).Conîrol embryos are show on the left (A,D,G,J,M), HSFTZ embryos in the middle (B,E,H,K,N) and HSFTZVP16 embryos on the right (C,F,I,LIO).fil transcripts were detected using a probe that hybridizes specifically to endogenousfi gene mRNA. Arrows mark the position of the cephalic furrow in gastrulating and early germband extending embryos. kowheads in panels I and L indicate posterior regions of the first three even-numbered parasegments. Asterisks in panel L mark the middIe of the first three even numbered parasegments (L). Note that FTZVP16 is able to induce ectopic fc expression in regions and at times that Fi2 cannot. However, its activity is still limited. WT HSFTZ HSFTZVP 16 To summarize, both FTZ and FTZVPl6 are limited in terms of when and where they can activate the endogenous fi,. gene. FTSVPL6, however, activates fi to somewhat higher levels, and exhibits broader temporal and spatial domains of activity. As seen with the cuticle preparations, tfie effects of FTZAC are similar to those of FTZ, although weaker, and FTïANVP16 has no effects on jk expression (data not shown). Similar results, although rnarginally weaker, were obtained with 4 minute heat shocks (data not shown).

Responses of the engrailed (en) gene

Afterjk, the next best-characterized FTZ target gene is the segment polarity gene engrailed (en).Transcription of en normally begins at the end of celIularization, with its 14, single-cell-wide stripes initiating in an anterior to posterior fashion (Fjose et al. 1985; Komberg et al. 1985 and Figure 3.4D). The even-numbered stripes, which appear prior to adjacent odd-numbered stripes, are FTZ-dependent (Howard and lngham 1986; DiNardo and O'Farrell 1987: Ingharn et al. 1988). These broaden when FTZ is ectopically expressed (Ish-Horowicz et al. 1989 and Figure 3.4E) with the same response kinetics asfi autoregulation (Nasiadka and kause 1999). Binding sites for FTZ and FTZ- FI have been mapped within the en promoter, and mutation of these sites affects reporter gene responses (Florence et al. 1997). tnterestingly, FTZdependent activation of en begins well after the beginning offi gene autoregulation. Consistent with this delayed responsiveness, ectopic expression of FT2 fails IO bring about prernature activation of en (Figure 5-48}.However, as even-numbered stripes begin to initiate. ectopic FTZ expression causes an anterior expansion of about one ce11 in width (Figure 3.4E). This effect can be achieved until about mid-germ band extension (approx. 3:30 AEL). after which time. ectopically expressed FTZ has no effect. The effects of FTZVPl6 on en are again different than those of FTZ and also differ somewhat fiom the types of effects that were observed on the fc gene. As withftz. en responses begin enrlier than those induced by the non-fused FTZ protein. Weak induction is observed in the mnk of the embryo during the earliest stages oPceIlutarization (approx. 210 AEL: Figure 3.4C). Curiously, this early ability of FTZVP16 to activate en in the trunk is quickly lost, such that by mid- cellularization. FTZVP16 is only able to activate en in posterior regions of the head, and more weakly at the posterior tip of the embryo (Figure 3AF). By the end of cellularïzation, a reciprocal pattern of induction is once again detected: en is no longer activated in the terminal regions. and induction of seven pair-rule-like hpes occurs in the tmnk (Figure 3.41). These stripes are formed by the anterior expansion of even-numbered en stripes such that they fuse with the odd-numbered stripes in Front Expression is. however, stiIl excluded hmthe posterior regions of even-numbered parasegments. Interestingly, mZW16 was able to activatefi= within the latter regions. but not in the middle portions of odd-numbered parasegments (compare Figure 3-41 to Figure 3.31). 12% Figure 3.4 Effects of FTZVP16 on engrailal (en). E~pressionof en traoscripts is show at three different stages: during early celluIarization (A-C), mid-cellulafization (D-F) and early gerrnband extension ((3-1). Control embryos are show on the lefi (A,D,G), HSFTZ embryos in the middle (B,E,H) and HSFTZVP16 embryos on the right (C,F,I). Arrows mark the position of the cephaIic hrrow, and smpes are numbered up to stripe 6. Smpe 1 in G-1 is located within the cephalic furrow and is therefore undetected. Note that FTZVP16 overcomes some, but not all, of the spatial and temporal limitations imposed on FTZ.

Thus, the factors that limit FTZVPl6 activity spatially are promoter-specific, and for these two genes, nearly complementary, Effects on en expression by the conml constructs FTZAC and FTZhNVP 16 were essentially the same as observed on&; FTaC had the same effects as FTZ (although weaker), and FTZhNVP16 had no effects at al1 (data not shown). Similar responses were also observed when 4 minute heat shocks were used to induce transgene expression (data not shown). These aends were found with al1 other target genes tested, and will not be described fiirther.

Eflecrs ofFZZVP16 on negativeiy regdated rurget genes

Previous tests conducted on the transcriptional activities of FTZ in vitro and in cultured cells only revealed its ability to function as a transcriptional activator (Jaynes and O'Farrell 1988; Han et al. 1989; Winslow et al. 1989: Ohkuma et al. 1990). In the developing embryo, however, FTZ also appears to function as a transcriptional repressor (Ingham et al. 1988; Copeland et al. 1996; Nasiadka and Krause 1999). The best-characterized negative target of FTZ is the segment polarity gene ivingless (wg). Infi mutant embryos, wg smpes expand to fil1 regions where FTZ is nonnally expressed (Ingham et al, 1988), and when FTZ is expressed ectopically, odd-numbered wg stripes are repressed (Ish-Horowicz et al. 1989) with the same rapid kinetics as observed forpz gene autoregulation (Nasiadka and Krause 1999). If FTZ is indeed a direct repressor of wg, then it is quite possible that addition of the strong VP16 activation domain to FTZ will convert it fiom a repressor of wg into an activator. Figure 3.58 shows the strong repression of alternate wg stripes that is observed 25 minutes afier ectopic expression of FTZ. The effect of FTZVP16 within the same short recovery time was quite the opposite (Figure 3.32). Instead of repressing wg, FTZVPI6 caused a rapid and substantial broadening of wg stripes. Expression expanded into odd-numbered parasegments, and in abdominal regions, into parasegments 8 and 10 as well, The pair-rule gene sloppy-paired (slp) is another gene that appears to be directly repressed by FTZ (Nasiadka and Krause 1999). Like wg, slp svipes expand into FTZ expression domains infk mutant embryos, and are repressed when ETZ is expressed ectopically (Nasiadka and Krause 1999 and Figure 3.5E). As with wgt the repression caused by ectopic FTZ is converted into activation when the VP16 activation domain is fused to FTZ. FTZVP16 causes a rapid expansion of slp stripes into odd- numbered parasegments (Figure 3.5F), much as seen for wg. As observed with the responses offi and en. the regions in which activation of wg and slp occurs are limited both spatially and temporally. Taken together, these results provide furthet evidence that FTZ is capable of acting as a transcriptional repressor, and that wg and slp are direct targets of FTZ repression.

Efleécis of Mt6 on suspected non-target genes 13 I Figure 3.5 Effects of FTZVPl6 on genes that are repressed by FTZ. The effets of ectopic FTZ and FTZVP16 expression are shown on the expression patterns of ivg (A-C) and slp (D-F) in control (A,D), HSFTZ (B,E) and HSFTZVP16 (C,F)embryos. The tint six stripes of wg are numbered. Arrows (and dots) mark the ptimary stripes of slp, which are expressed at the posterior edges of even-numbered parasegments Ectopic FTZ ~pïessesaltemate smpes of wg (B) and slp (E), whereas ectopic FTZVPI6 activates both genes throughout odd-nurnbered parasegments (C,F). FTZVPI6 also induces wg expression in the and 10Ih even-numbered parasegments (C).

133 In a previous study, four of nine genes tested appeared not to be direct targets of FTZ. These genes, even-skipped (eve), hairy (h), runr (mn)and gooseberry Igsb), showed either no response or a significantly delayed response to pulses of ectopically expressed FTZ (Nasiadka and Krause 1999). In vivo UV-cross-linking studies, however, indicate that FTZ is evenly distnbuted along the promoter of at least one of these genes, me (Walter et al. 1994). It is possible that ectopicaliy expressed FTZ fails to affect these genes due to the absence of available cofactors or the presence of overriding repressors. Hence, we expressed FTZVPl6 to see if such limitations existed and whether they couId be overcome, Figure 3.6B. D, F, H shows that al1 four genes fail to show any response to FTZVPI6 within the 15 minute response time that al1 the other direct responses were observed. This was nue for al1 stages tested, suggesting that FTZ is either not bound to these promoters, or that it is bound to sites chat are somehow non-functional.

The majority of FTZVP16 activity is homeodomain-independent

In previous studies. we have shown that FTZ can regulare the majority of known target genes when the DNA binding activity of its homeodomain is compromised (Fiupatrick et al. 1992; Copeland et al. 1996: Hyduk and Percival-Smith 1996). These results suggest that FTZ can be recruited to response elements on target gene promoters via specific interactions with cofactors bound at those sites. In order to explore the temporal, spatial and promoter-specific properties of these cofacton, we compared the regulatory abiiities of FTZVP16 and FTZAHDVP16 on three FTZ target genes. Figure 3.7 shows the responses offi (A,B), odd (C,D)and en (E,F) in embryos just beginning to cellularize (early stage 5, approx. 210 AEL). For the first two gnes.fr,- and odd. equivalent levels of activation are achieved by both FTZVP16 and FTZAHDVPl6 within the trunk region of the embryo. in contrast, the early response of en, which can be induced by FTZVP16 but not FTZ. fails to occur when the homeodomain of FTZVP16 is removed (compare 3.7E and F). This homeodomain dependence is not observed later dunng the normal temporal window of en expression (compare 3.7G and H). Severai conchsions can be drawn hmthese results. First, FTZ is able to regulate the& and odd genes as eariy as stage 5. This activity is boosted by fusion of FTZ to the VP16 activation domain. Second, binding to the fi and odd promoters can occur independently of the homeodomain, indicating that cofactors necessary for promoter recruitment are present and active at this early stage. Third, in contrast to the other two promoters, binding of FTZ to the en promoter is horneodomain-dependent in stage 5 ernbryos. Howeve. even with the homeodomain present, transcriptional activation at this stage fails to occur unless the bound protein is fused to the VP16 activation domain, 134 Figure 3.6. Eff~tsof FTZVPl6 on non-FTZ target genes. mRNA expression patterns are shown for control ernbryos on the lefl (A,C,E,G)and HSFTZVP16 ernbryos on the right (B,D,F,H). Embryos stained for h (A,B), eve (C,D) and nm (€,FI are shown at the end of cellularization (end of stage j), while those stained for gsb expression (G,H)are shown at gastrulation (stage 6). No effect is observed on any of these genes by ectopic expression of FTZVPL6. HSFTZVP 16 136 Figure 3.7 The FTZ homeodomaio is dispensable for most FïZVPl6 activities. Expression of the& (A,B), odd (CD)and en (E,F,G,H) genes is show in HSFTZVP16 (A,C,E,G) and HSFTZAHDVP16 (B,D,F,H)embryos during early cellularization (A-F) or late cellularization (G,H). With the exception of early activation of en, the FTZ homeodomain is not required for any of the observed FTZVP 16 activities. HSFTZVP 16 HSFTZAHDVP 16 3.5 DISCUSSION

Seleclive binding versus activity regulation

As described earlier, there are two prevalent models that seek to explain how horneodomain proteins achieve their unique developmental roles in vivo (Biggin and McGinnis 1997). The selective binding model argues that homeodomain proteins regdate specific target genes with the aid of cofactors that direct them to a specific subset of their many potentia1 binding sites (Mann 1995; Mann and Chan 1996). The activity regulation model. on the other hand, argues that homeodomain proteins are distributed rather evenly along the promoters of most genes (Walter et al. 1994; Carr and Biggin 1999), and that interacting cofactors act primarily by modulating their transcriptional activity (Biggin and McGinnis 1997; Li and McGinnis 1999: Li et al. 1999a). imponantly, the two models also differ in the number of predicted target genes. The selective binding model predicts a relatively limited number of targets while the activity regulation mode1 predicts a very large number of targets. The data obtained with Our FTZVP16 fusion protein and actual target genes suggest that FTZ binds and regulates a limited number of target genes and that target gene selection occurs primarily by cofactor effects on the aftïnity for specific sites. One of the clearest resuits in favor of selective target gene binding was the limited number of observed FTZVP16-responsive genes. Four of the nine genes moniiored in this study (eve. hairy. mnt and gooseberty) were previously suggested to be non-target genes (Nasiadka and Krause 1999). Consistent with this finding, al1 four failed to respond to FTZVP 16 at any of the developmental stages tested, suggesting that FTZ does not bind to the promoters of these genes. This result is inconsistent, however. with those of previous in vivo cross-linking experiments. Immunoprecipitation of FTZ rnoIecules. cross-linked to DNA by UV irradiation showed that restriction fragments within a 7 kb region of the eve promoter, and within several other promoters includingjiz, could al1 be detected with similar eficiencies (Walter et al. 1994). A possible explanation for this apparent discrepancy is that Our assay measures promoter activity while the cross-linking experiments measured promoter occupancy. In other words, FTZVP16 may be able to bind to the eve, haitv! mrand gooseberry promoters, but the VP16 activation domain may for some reason be incapable of activating them. This explanation, however, is not consistent with the strong activation of me that is induced by the VP16 activation domain when fused to the eve repressor protein Runt (Jimenez et al. 1996)- Similarly, the el6activation dornain is able to activate runr when fused to the runt repressor pmtein Hairy (Jimenez et ai. 1996). Alternative explmations for the inability of FTZVPi6 to activate these promoten are chat there may be prornoter-specific repressors of VFt6 that specifically target the FTZ portion of the fusion pmtein. or that FïZVPI6 binds to regions of these promoters that are somehow non-Cunctional. A third esplanation is that Fi2 only binds these promoters at hi& levels of expression. The cellular concentrations of FTZVPI6 that we induced are approximately 50% or less of endogenous levels of FTZ. We have 139 found in pst studies that these levels are more than suficient to rescueJz-dependent segmentation (Fitzpatrick et al. 1992; Copeland et al. 1996). Also, heterozygous animals with only one wild-type copy of the gene develop normaily (Wakimoto and Kaufinan 1981). Hence, sites detected by cross- Iinking in wild-tye animals may represent a significant number of non-essential sites that are only bound when excess protein is present, A second argument in favor of the selective binding model is that the most muent segmenta1 phenotype generated by ectopically expressed FTZVP16 was the same as that generated by full-length FTZ- If FTZ were widely distributed on the majority of Drosophila promoters, then addition of the VP16 activation domain should have resulted in the spatial and temporal misexpression of rnany of these genes. This in mm would lead to dramatically different segmental phenotypes, perhaps even lethality prior to the secretion of cuticle. Similarly, inappropriate activation of FTZ target genes before and afler the time that FTZ is normally expressed should aIso have led to novet segmental phenotypes. However, FTZVP16 had no obvious effects on segmental patteming when expressed at those stages of development. Although FTZVP16 was able to induce several novel phenotypes within the normal temporal window of& expression, these were relatively low in frequency and could be explained by the observed effects on known target gene expression patterns. The Iimited effects of FTZVP16 on known target genes and segmenta1 phenotypes compare wetl with results obtained with two other QSO homeodomain-VPl6 fusion proteins. Ectopic expression of UBXVP16 (Li and McGinnis 1999) using the ubiquitously expressing armadiIlo promoier generated relatively normal cuticles with mixed Ubx and -4nrp-like segmentai transformations in the thorax and head (Li and McGinnis 1999). DFDVP16 eqressed in the same way generated stronger segmenta1 defects, but the transformed segments were nevertheless DFD-like in identity (Li et al. 1999a). Amongst the suspected target and non-target genes monitored in those studies. only one responded inappropriately and this may have been an indirect effect (Li et al. 1999a). Another result in favor of the selective binding model was the ability of the homeodomain- deleted versions of FTZ and FTZVP16 to regulate FTZ target genes. These proreins have no capacity to bind DNA on their own (Fiapamck et al. 1992), and yet are capable of regulating ai1 of the FTZ target genes investigated (Fitqamck et al. 1992: Copeland et al. 1996). This suggests that pmmoter binding can be mediated by cofacton that are either bound, or are able to bind, to specific sites on FTZ target genes. Indeed, analysis of ajlz reporter gene containhg a portion of the* autoregdatory element has shown that recmitment of FTZ cm be achieved by both homeodomain- dependent and homeodomain-independent mechanisms (Schier and Gehring 1993b; Hyduk and Percival-Smith 1996). Protein-protein interactions such as these have been shown to change equilibrium constants for DNA by nvo to three orders of magnitude (Ptashne 1992). For example, FE-FI can increase the affinil of FTZ for flanking binding sites by 100 foid {Yu et al- 1997). At low protein concentrations, these interactions would favour binding to the small subset of FTZ binding sites that have FTZ-FI (or other cofactor) binding sites nearby. Cofactors have aiso been shown to affect the binding site specificity of other QSO homeodomain proteins. The best characterized of these is a divergent homeodomain protein called Extradenticle (EXD).EXD interacts with HOX proteins via conserved YPWM motifs found N- terminal to their homeodomains (Mann and Chan 1996). HOX-EXD heterodimers bind bipartite response elements, and by changing the spacing and sequence of these elements, binding and responsiveness can be changed From one HOX protein to another (Chan et ai. 1994b; Chan et al. 1997; Ryoo and Mann 1999). Several other critena support Our conclusion that FTZ specificity is determined prirnarily at the level of promoter binding. First, the number of potential FTZ binding sites greatly outnumbers the amount of protein molecules available. In vitro DNA binding studies have show that there are approximately 10-20 FTZ binding sites per kb of genomic DNA. The Drosophila genorne contains about 3 X 10'kb of DNA per diploid genome (Miklos and Rubin 1996). Hence, as many as 6 X 106 rnolecules of FTZ would be required to occupy al1 of these sites. This number is more than two orders of magnitude higher than the 15,000 molecules previously estimated to be expressed in blastoderm stage nuclei (Krause and Gehring 1988). This is the stage when FTZ protein levels are highest and as stated earlier, dosage studies have show that FTZ can regdate target genes effectively at levels that are 24times lower (Wakimoto and Kaufman 198 1; Fitzpamck et al. 1992: Nasiadka and Krause, unpublished observations). indeed, immunolocalization shows bat, when expressed at functional levels in salivary glands. bands of FTZ and other Q50 homeodomain proteins are detected at somewhere between 50 and several hundred discrete chromosomal locations (Serrano et al. 1995: Botas and Auwers 1996: H. Krause. unpublished observations). Taken together: these data suggest that the number of target genes bound and regulated by QSO homeodomain proteins range in the low hundreds, not thousands.

The role of activi@ regdion

Our results show that activity regulation aiso plays an important role in FTZ hnction, but that this role is mainly to rerefe the temporal and spatial windows of target gene regulation and to modulate levels of expression. This conclusion is supported by the following results. First, five of the genes tested @-. odd, sip, en and rvgl could be activated ectopicaily by FTZVP16 in regions and at times that FE! could not induce a response. This shows that FTZ has the ability to bind to these promoters, but that it must be bound in an inactive state. For FTZ to function in these cells. it probably requires the addition of requisite cofactors, the removal of repressors or both. For the five genes listed above, the VP16 activation domain was able to overcome some of these limitations. The regdation by FTZ of en is a good example of this type of temporal and spatial refinement in activity- Our resulrs with FTZVPl6 showed that FTZ can bind to the en promoter 141 during the time thatfi,. autoregulation and odd activation are well under way. However, the ability of FTZ to activate en is normally delayed until cellularization is completed (approx. 45 min). This delay may be necessary to allow other en regulators to resolve into the complex patterns of expression that are required for en to initiate in 14 narrow stripes. Like most homeodomain proteins, FTZ has the ability to function as both a transcriptional activator and repressor. This dual capacity suggests a requirement for distinct activity-regulating cofactors. However, differential activity can also be achieved, at least in part, by binding to different sites on different genes. For example, the response elements required for repression of the Distalless gene by UBX (Vachon et al. 1992; Serrano et al. 1995; Botas and Auwers 1996) and activation by DFD (O'Hara et al. 1993) are different. This also appears to be the case for activation of the dpp gene by UBX and its repression by ABD-A (Capovilla et al. 1994; Sun et al. 1995; Capovilla and Botas 1998). The different cofactors that help recmit the three proteins to these sites rnay also be partly responsible for their differences in transcriptional activity. For example, EXD is thought to generally work as a coactivator. acting in part to alter HOX protein conformation (Li et al. I999a). Other factors bound in the vicinity of these sites are also likely to play a major role in activity regulation (see for example Li et al. 1999b). A better sense of the relative contributions made to functional specificity by binding site selection and activity regulation. and the number of genes regulated by each protein will require more comprehensive methods of identifying and monitoring target gene responses. Methods capable of rnapping binding sites in vivo with greater resolution and with higher levels of sensitivity will also be needed.

Homeudomain protein potential and the need for regulators

In addition to showing that positively acting cofactors are important for FTZ specificity, our data implicate the actions of powerful negative regulators that limit its temporal and spatial domains of activity. The strength and diversity of these negative regulators was emphasized by their ability to suppress the actions of the fused VP16 activation domain despite its previously reported reputation of strength and autonomy. It rnay be the low DNA binding specificity of the homeodomain that has necessitated this need for diverse mechanisms of repression, since low DNA specificity provides the potential to regulate a large number of inappropriate target genes. Indeed, a rapidly growing number of homeodomain proreins have been shown to be capable of functioning as oncogenes or proto- oncogenes (Rabbitts 1994; Look 1997), and oncogenici~can be conferred by fusions to other transcriptional activators (Hunger et al. 1991). Further studies will be required to identi@ many of the cofactors and inhibiton that modulate FTZ activity and to determine how they do so. Chapter 4:

Discussion and future work FTZ, a homeodomain-containing transcriptional replator, plays an important role in Drosophila pattern formation. The& gene is required for segmental subdivision of the embryo and specification of segmenta1 and neuronal identities. These hnctions are accomplished through the regulation of specific subsets of target genes. Despite the fact that FTZ has been the subject of intense studies, the mechanisms by which this transcription factor acts in vivo remain, for the most part, mysterious.

41.1. KINETIC ASSESSMENT OF FTZ GENE-EGULATORY INTERACTIONS

As a first step towards a better understanding of the rnechanisms underlying FTZ-dependent regulation, 1 have assessed, using an in vivo approach, which of the senetically identified downstream genes are directly and which are indirectly regulated by FTZ (chapter 2). This assessment. carried out in HSFTZ embryos, is largely based on the kinetics of FTZ downbmearn gene responses to pulses of ectopic FTZ expression- 1 found that the responses of these downstream genes faIl into two distinct temporal windows. Immediate responses include ectopic activation of endogenous fi,en, prd and odd as well as repression of slp and wg (see Figure 2.8). Based on previous molecular evidence thatfi autoregulation is direct (Schier and Gehting 1992), I suggest that other responses in this temporal window also represent direct gene-regulatol interactions. also detected delayed responses of downstream genes in HSFTZ ernbryos. ïhese responses include ectopic activation of wg and gsb. 1 provide evidence that wg activation requires the induction of an intermedian, factor. suggesting that this and other delayed responses represent indirect gene regdatory interactions. Based on these results, i propose that FTZ carries out multiple direct gene-regulatory functions during its expression throughout even-numbered parasegments. This conclusion differs fiom that of a previous study suggesting thac the only reaI requirement of FTZ is to activate en (Lawrence et al. 1987). Since the kinetic approach was applied to a srnaIl portion of genes, it is IikeIy that target genes identified to date do not represent al1 of the genes regulated by FTZ. To identi@ other direct gene regulatory interactions, a more cornprehensive and systematic approach has to be undertalien (see below). Our results suggest that, in Drosophila embryos, FTZ can act as both a transcriptional activator as well as reptessor. Previous studies based on aanscriptional assays carried out in vitro and in Drosophila tissue culture cells only dernonstrated the abilip of FiZ to function as an activator (Jaynes and O'Farrell 1988; Fitzpatrick and [ngles 1989; Han et ai. 1989; Winslow et al. 1989: Ohkuma et al. 1990; Fitzpanick et al. 1992; Peterson and Herskowitz 1991; Ananthan et al. 1993 Colgan et al. 1993). Future studies wilI have to elucidate the mechanisms that allow FTZ to hction as a repressor (see below). 41.2. THE: MECHANISMS BY WEICH COFACTORS AFIXCT FïZ FUNCTION

Tne abiIity OC FTZ to act as both a transcriptional activatûr and repnssor is most Iikely achieved through specific interactions with various factors. Cofactors have a significant effect on FT2 function. In fact, FTZ hnction could not be executed at al1 without interactions with other factors. How do cofactors affect FTZ hnction? Do they modi@ DNA binding propenies of FTZ, or do they affect FTZ transcriptional potentiaI, or both? L have undertaken an in vivo approach to assess how cofactors affect FTZ hnction (chapter 3). In my approach, [ have andyzed the regulatory efEects of a chimericfishi tarm transcription factor FTZVP 16 on Fi2 target and non-target gene regulation. Since VPI6 activation domain is capable of activating virtually any eukaryotic promoter, FTZVP16 was expected to activate most promoters occupied by FTZ. independently of transcriptional activity regulation. [f FTZ function is strictly determined at the level of DNA binding such that cofactors recmit FTZ to a subset of a limited number of pnes, then the effect of FTZVP16 on segmental patterning should be similar to that of FTZ, as both FTZ and FTZVPI6 would bind and regulate he same set of genes. If, however, FTZ is distributecl on chromatin in a widespread manner, and FTZ hnction predominantty depends on the mnscriptional activity regulation, then FTZVP16 would be expected to activate many more genes han FTZ. This relativeiy unrestricted gene regulation should have a profound effect on segmentai paneming. The results of my studies suggest that the functional specificity of FTZ is rnainly detemined at the Ievcl of DNA binding (chapter 3). This is based on the observation that the effect of FTZVP16 on segmental patterning is limited, resembling that of FTZ. For example, both FTZ and FTZVP 16 are effective in the same narrow temporal window during early embryogenesis. In addition, the only phenotype induced by ectopic FTZVPI6 with high frequency is an anti-fiz phenotype, a phenotype specific to ectopic FTZ (Sûuhl 1985; ish-Homwicz and Gyurkovics 1988). 1 also found that not al1 but only a subset of segmentation genes responds to ectopic FTZVP16. Genes that respond have been previously identified in genetic studies as FTZ downstream genes. Taken together, these results suggest that the distribution of FTZ on chromatin is not widespread but discrete and that FTZ binds and regdates the promoters of a limited number of genes. AIthough activity regulation may not be required for FTZ target gene selection, 1 found that activity regulation phys an important rale in establishing the precise spatial and temporal elspr~ssionof FTZ target genes. It may ais0 detenine which genes are activated by FTZ and which are repressed. Future studies will have to determine how individual factors participating in FTZdependent reptation affect FTZ hnction, Le. which of these factors contrai DNA binding of FTZ and which modulate FTZ transctiptionat potentiai (see below).

43. DISCUSSION AND mrrZiRE WORK 13.1. ARE HOMEOTIC GENES REGULATED BY FTZ?

The kinetic assessment of FTZ gene regulatory interactions has been applied to a Iimited number of genes. It is therefore likely that the direct target genes identified in my study are not the only genes that are directly regulated by FTZ. Other possible candidates include the homeotic genes. Genetic studies demonstrated that the spatial expression patterns of these genes are affected byfi mutations (Ingham and Martinez-Arias 1986; Riley et al. 1987: Boulet and Scott 1988; Jack et al. 1988: Martinez Arias and White 1988; Riley et al. 1991; Core et al. 1997). The finding that their temporal expression overlaps that of FTZ is consistent with the possibility of direct regdation by FTZ. The molecuIar dissection of sorne homeotic gene promoters also suggests that they rnay be directly regulated by FTZ (Qian et al. 1991; Muller and Bienz 1992; Qian et ai. 1993: Core et al- 1997). It would be of interest to determine the kinetics of responses of homeotic genes in HSFTZ embryos and to assess whether they are directly or indirectly regulated by FTZ. In the kinetic studies of homeotic gene regulation, it has to be taken into account that some OC these genes span long regions of DNA. For example. the transcription units of Ubx and Anrp are 77 kb and 100kb, respectively (Schneuwly et al. 1986: O'Connor et al. 1988; Kornfeld et al. 1989). In contrast, the segmentation genes that have been investigated thus far (chapter 2) are relatively short, with an average size of 2 kb. Given that Drosophila polyrnerase II progresses at a rate of about 1.4 kb per minute at 25'~(O'Farrell 1992; Thummel 1992), transcription of the Ubx gene would require 55 minutes, as compared to less than 2 minutes needed to transcribe en or endogenousfi. To assure that al1 direct responses fa11 into the same temporal window, the expression of homeotic genes would have to be visualized with cDNA probes specifically recognizing 5' portions of nascenr transcripts. StiIl. visualization of these transcripts may only occur above a certain threshold concentration. The time to attain this concentration would depend on the rate of transcript accumulation and rnay also depend on the size of a transcription unit. If so, then the ternporaI window of direct gene-regulatory interactions established in my studies (chapter 2) would not be applicable to these genes. To standardize the kinetic responses. fusion transgenes would have to be used in which horneotic regulatory sequences are fused to a reporter gene, the transcript of which is of a similar size and a sirnilar half-life to those previously analyzed. Alternatively. rninigenes missing the majority of homeotic gene introns but containing intact cis-regulatory sequences and emns could be constmcted. It couId be argued that gene-regulatory interactions investigated in HSFTZ embryos occur in ectopic locations and rnay therefore not represent the relevant regulatory relatianships of wild-type embryos. To monitor gene-regdatory interactions in endogenous locations, kinetic analyses couId be carried out in a f~-mutant background (Le. in HSFTZ embryos that are also mutant forfi). This approach. however, could only determine the kinetics of the earliest ETZ interactions as some of the interactions that occur in later stages may depend on the early ones and therefore be non- reproducible in thejz mutant background. This may be circumvented by usingjz ts alleles, which would allow for turning off FTZ function at different deveIopmental time points. A direct regulatory effect of FTZ may not always resuk in immediate gene activation or repression. For example, FTZ function could be delayed by the presence of overriding repressors or the absence of cotàctors whose expression (or activity) is temporally restricted. Altematively, FTZ may play an awiliary dedun'ng gene regulation, acting, for example, as a factor altering chromatio structure. FTZ-dependent modification of chromatin would rnake it possible for other regulators to corne into play and evoke immediate responses from regulated target genes. The action of these regutators and that of FTZ could occur in the same or distinct temporal windows. To test this, the regulation of genes with delayed responses should be analyzed in HSFTZVP16 embryos. If FTZ is capable of occupying the promoters of these genes, they should be activated by the potent and autonomous VPl6 activation domain and produce imrnediate responses to ecropic FTZVP 16.

42.2. HOW MANY FTZ TARGET GENES ARE THERE iN THE DROSOPHU GEMME

How many genes do FTZ and other Q50 homeodomain proteins regulate? To address this, Liang and Biggin (Liang and Biggin 1998) selected at randorn clones from an 8-12 hour Drosophila cDNA library and analyzed the expression pattems of corresponding genes in wild-type and f~.eve and CJh mutant backgrounds. The results of this analysis suggested that at Ieast 87% of Drosophila genes expressed during Iate embryogenesis are downstream of FTZ, EVE, UBX and other Q50 L horneodomain proteins. To address what fiaction of these loci are directly regulated by FTZ, expression patterns of randomly selected clones were analyzed in the fi mutant background at stage 5 (Liang and Biggin 1998)- WiId-type levels of FTZ are the highest at this stage (Krause et al. 1988) and any spatial aiterations in downstream gene patterns detected at that tirne could result from direct regdation. The results of tliis study showed that 25-50% of genes transcribed at stage 5 are regulated by FTZ, suggesting that FTZ rnay directly regutate from 715 to 1430 genes. Does FTZ directly regutaie a thousand Drosophila genes? Regulation of a large number of genes by Fi2 is supponed by the results of in vivo UV-crosslinking studies (Walter et al. 1994). These studies demonstrated that FTZ crosslinks to al1 DNA fragments tested, suggesting thar FTZ distribution on chromatin is widespread. To determine how many genes FTZ reytates, a more systematic and rigorous approach has to be undertaken. One way to proceed is to use a kinetic method, as has ken descrïbed in chapter 2. combined with genome-wide identification of differentially expressed gnes. Since the Drosophila genome has been sequenced (Adams et al. 1000; Myers et al. 7000), cornprehensive analysis of gene expression patterns cm be carried out using an array-bas& approach (Brown and Botstein 1999; Lipshub et d. 1999). In this approach, the cornpiete set of recognized genes is represented as a dense, ordered amy of DNA moIecuIes attached to a solid substrate, with each arrayed DNA elernent reptesenting a specific Drosophila gene. The number and identity of genes regulated by FTZ can be inferred from variations in the giobal 147 transcript profile chat accompanies either ectopic FTZ expression or conditional inactivation of FTZ function by means of, for example, ts alleles. To detect these variations, DNA microarrays would have to be probed with differentially labelled cDNA pools derived from heat-treated embryos carrying the HSFTZ transgene or ts alleles and From identically heat-treated wild-type control embryos. The relative abundance of each mRNA message in a particular type of embryos can be measured by direct comparison of the hybridization signals. For e.xample, if cDNAs derived from HSFTZ embryos are labelled with a red fluorescent dye and cDNAs From wild-type controls are labelled with a green dye, then, after competitive hybridization, the spots within the microarray compiementary to genes spcifically activated by ectopic FTZ will be red (as red-dye labelled cDNAs witl be in relative abundance); whereas the spots complementary to genes repressed in HSFTZ embryos will be green (as green-dye labelled cDNAs will be in relative abundance). in order to distinguish which gene regulatory interactions are direct and which are indirect, a temporal series of genomic expression profiles in HSFTZ embryos that have been heat-treated and allowed to recover for variable periods of time would have to be established. During interpretation of the resultant sets of data, gene-regulatory interactions described in chapter 2 can be used as standard controls for direct and indirect regulation. If the assumptions of my kinetic analysis are correct, there should be no responses that are faster than those of endogenousfi=, en, and wvg. Similar temporal series of genomic expression profiles could be established for embryos in which FTZ function is inactivated by means of ts mutations. Al1 these experiments should be repeated to show reproducibility. Interactions identified in this way can be subsequently veritied by the analysis of expression patterns of candidate target genes in wild-type, HSFTZ, andfi mutant embryos. The expression of genes positively regulated by FTZ should overlap endogenousfi stripes, and should be significantly decreased or abolished in ft- mutant embryos and expanded in HSFTZ embryos. The oppasite alterations in spatial expression should be observed for genes negatively regulated by FTZ. These analyses should also determine temporal windows of uncovered regulatory events, and if they change during the course ofjiz expression (e.g. fiom positive to negative). Since some direct regulatory interactions rnay be delayed, ail responses that fail to exhibit rapid kinetics cannot be assumed to be indirect. DNA microarray technology is not only comprehensive but also very efficient, as an entire genome ciin be surveyed in a single hybridization expriment. This approach would not only determine the minimum of genes that are reguiated by FTZ but it would also reveal the identity of these genes. Sorne of the genes uncovered may have been characterized before, while others may be novel with unknown functions. The DNA microamy-based approach might also identfi target genes whose spatial expression is not aItered in HSFTZ orfe mutant embryos, but whose expression levels are affected. ïhese genes might have been missed in previous genetic or ectopic expression studies. This comprehensive identification of FTZ target genes may allow for clustering of target genes into different groups with respect to the timing, efficiency or sign of their responses. Each of these groups rnay reflect a different rnechanism of FTZ-dependent regulation. 423. MODES OF FE-DEPENDENT REGULATION

Identification of direct target genes should be followed by elucidation of the mechanisms by which these genes are regulated. This would require identification of factors that together with FTZ regulate these genes as well as the characterizaiion of response elements mediating this regulation. Kinetic analysis has identified six direct target genes of FTZ: four positively and two negatively regulated (see Figure 2.8). This suggests that there are at lest hvo different modes of FTZ-dependent regulation: one required for activation and another for repression. It is not clear what mechanisms allow FTZ to carry out opposite transcriptional functions. This rnay depend on differential interactions with other factors andfor the structure of FTZ target gene promoters. It has been found, for example, that the autoregdatory element (AE) from the& promoter activatcs TATA-containing promoters but not those coniaining the initiator element (Inr) and downstream promoter element (Dpe) (Ohtsuki et aI. 1998). tt would be of interest to establish if ail FTZ target gene promoters contain a TATA box or if some contain lnr element. These differences in the core promoter structure could determine whether a gene is regulated by FTZ or not, or whether regulation is positive or negative. The mechanism of positive gene regulation by FTZ does not seem to be the same for all its direct targets. This is based on the observation that theare differences in the spatial expression domains of positive target genes in HSFTZ embryos (chapter 2). For exarnple. endogenous fi= and en are ectopically activated in only one cell in odd-numbered parasegments (Figures 2.1 B and 2.28) (Ish-Horowicz et al. 1989; Fitzpatrick et al. l992), whereas prd is induced throuehout odd-numbered parasegments (Figure 2.4B). There are also differences in the temporal windows of these inductions. In particular, ectopic activation of odd and endogenousfi occurs during early cellularization (Figures 2-68 and 3.3B), whereas en is not induced until the end of cellularization (Figure 3.43. These differences suggest that FTZ interacts wih different sets of cofactors to regulate each of these genes. To further explore differences in mechanisms underlying the regulation of individual FTZ [arget genes, it should be established what domains of FTZ are required for their regulation. It has already been demonstrated, for exiunpie, that the FTZ homeodomain is dispensable for ectopic activation offi, en, and odd as welt as repression of wg (chapter 3) (Copeland et al. 1996; Hyduk and Percival-Smith 1996). Could the homeodornain also be dispensable for the ectopic activation of prd or the repression of slp? What requirements are there for other domains of FTZ? Similady, it should be determined which FTZ target genes are regulated by the known FTZ cofactors FTZ-FI and PRD. For exarnple, previous studies demonstrated that, although required for en and wg regulation, FTZ-FI is dispensable for endogenousjïz regulation (Guichet et al. 1997; Yu et al. 1997). 1s FTZ-FI required for the induction of ode/ and prd and the repression of J;lp? Are the same dornains of FTZ-FI involved in the regulation of different FTZ target genes? Similarly, is PRD, which has been shown to be necessaq for the repression of wg by FTZ (Copeiand et al. I996), also required for sip repression? i 49

If SU, the same mechanism could be used for the repression of wg and slp, and FTZ-dependent regulation of these genes may be mediated by conserved cis-acting regulatory elements. One approach to identiQ these elernents would be to compare wg and slp prornoter sequences that respond to FTZ (as determined by transgenic deletional and mutational analysis) and to search for sequence conservation. Similarly, the promoters of positive target genes regulated by the same mechanisms could be compared to identifi cis-regulatory elements mediaung FTZ-dependent regulation. Such sequence alignments could be further enended to include the homologues of FTZ target genes from various Orosophila species. The response elements mediating FTZ-dependent regulation could be also identified based on biochernical approaches such as EMSA LElectrophoretic -Mobility ShiR Assay) and DNase 1 footprinting. These studies could be hrther extended to include CO-transfectionexperiments in cultured Orosophila cells.

4.2.4.jIz AUTORECULATION

At present, the best-characterized target of FTZ-dependent regulation is the autoregulatory element (AE) fiom thefi upstrearn prornoter (see Figure 1.8) (Schier and Gehring 1992; Han et al. 1993; Schier and Gehring 1993a; Han et al, 1998). Deletional and mutational anaiysis of the A€ (Schier and Gehring L993a) as well as in vitro binding studies with ernbryonic nuclear extracts (Han et al. 1993) revealed that, in addition to FTZ binding sites, a nurnber of non-FTZ binding sites are required for autoregulation. It is not clear, at present, what factors act through these sites. Some candidates have been suggested based on in vitro biochemical purification (Han et al. 1993; Han et al. 1998), others based on the sequence sirnilarity to consensus binding sites (Schier and Gehring 1993a). Analysis of the activities of mutant versions of AE or 323 fPE in HSFTZVP16 embryos could determine the mechanisms by which some of these factors act. If FTZVPL6 can activate these mutant constructs, it would suggest that factors nomally acting through these sites do not affect the DNA binding of FTZ but rather its transcriptional activity. The locatized distribution of factors involved infi autoregulation accounts for the spatially restricted expression of the A€ element in HSFTZ embryos (Schier and Gehring 1993a; Hyduk and Percival-Smith 1996). Genetic and ectopic expression studies demonstrated that several pair-rule genes act as negative regulators of)= expression (Carroll and Scon 1986; Howard and Ingham 1986; Ish-Horowicz and Pinchin 1987; Frasch et al. 1988; Ingham and Gergen 1988; Carroll and Vavra 1989; Lawrence and Johnston 1989a; Manoukian and Krause 1992; Cadigan et al. 1994b; Cadigan et ai, I994a; Mullen and DiBardo 1995; Tsai and Gergen 1995; Yu and Pick 1995; Saulier-Le Drean et al. 1998). However, the response elements for these proteins have not been precisely mapped. Do these pair-rule gene products directly affect FTZ-dependent regulation via the AE? It has been demonstrated that SLP binds to the AE element in vitro and in yeast cells (Yu et al. 1999), suggesting that SLP may repressjk by acting through the AE. It would be of interest to compare the spatial expression of the A€ to that of endogenousfi in HSFTZ embryos. If AE stripes expand 1su posteriorly into even-numbered parasepents, it would indicate that ODD, which normally represses jiz in the middle of even-numbered parasegments (Mullen and DNardo 1995), does not regulate the AE and therefore acts through other regulatory regions in theh promoter. A similar conclusion would be reached for EVE andor SLP if expression of AE expanded anteriorIy or were induced in an additional set of stripes in the posterior portions of even-numbered pwegments. If, however, the spatial limitations of the AE activation by ectopic FTZ were similar to those of endogenous fi,then this would suggest that the effects of ail negative regulators on fk rnay be mediated, at least in part, by AE. If so, then the regulatory effect of FTZVPL6 on this element should be analyzed. Any ectopic induction of AE in HSFTZVPI6 but not HSFTZ embryos would suggest that some Iimitations result frorn transcriptional activity regulation of FTZ. The effect of ectopic expression of ODD, SLP and EVE on AE should also be investigated to determine whether or not these factors act through the AE. jz autoregulation does not appear to take place in the developing nervous system (Hiromi and Gehring 1987) and neither endogenous nor ectopic FTZ is capable of activating the AE in this tissue (Schier and Gehring 1993a). This could be due to the fact that some cofactors required for FTZ-dependent regulation are not expressed or active in the nervous system. It would be of interest to analyze the efiect of FTZVPI6 on the AE in the developing nervous system. If FTZVP16 activates this element in this tissue, it would mean that FTZ is recruited to the AE but requires augmentation of tfanscriptional activity to function. if FTZVP16 fails to activate the AE, it may be due to the fact that FTZ is no longer being recniited to this etement. FTZ recruitment may depend on FTZ-FI, which binds cooperatively with FTZ tu DNA fragments derived frorn the AE (Yu et al. 1997) but is not expressed in the developing riervous system. It should be determined whether ectopic expression of FTZ-F 1 in the deveioping nervous systern is suficient for endogenous and ectopic FTZ to activate the A€. Since activation of the AE may also depend on other cofactors that regulate FTZ transcriptional activity, FTZ-F I should also be co-expressed with FTZVP16. If FTZ-FI is indeed required for fTZ recruitment to the AE, then FTZ-FI coexpression with the horneodomain-deleted versions of FTZ and FTZVP16 should also resuit in activation of the AE in the developing nervous system. Expression of a reporter gene directed by the zebra eiement is nor affected by mutations in thefi coding region, suggesting that this eiement does not participate in f~ autoregulation (Hiromi and Gehrïng 1987). Based on the observation that transcriptional activity regdation plays an important role in FTZ target gene regulation (chapter 3), it is possible that FTZ binds in vivo to the zebra element but that its repulatory effeçt only takes place in the context of an intactfi gene. ïhis would be the case if factors required for the aanscrïptional activity of MZ did not bind to the zebra elernent but elsewhere in the32 promoter. Without these factors, FTZ wouId remain in a functionally neutral state. In vitro binding srudies have shown that FTZ does bind to the zebra element (Henry Krause, unpublished results). Moreover, the zebra element also contains FTZ-FI binding sites (Ueda et ai. 1990). To hrther uivestigte whether FTZ occupies the zebra element in vivo, the activity of 151 this elernent should be analyzed in HSFîZVP16 embryos. If FTZ indeed binds to this element, FTZVP16 should activate its expression. The ability of FTZ to act through the zebra element in the context of the full-length promoter may account for the functional interaction reported between the zebra and upstream element (Yu and Pick 1995). In particular, it has been observed that the expression pattern conferred by both elements combined differs quantitatively and qualitatively frorn patterns directed by the separate elernents (Yu and Pick 1995). In addition to the zebra element, the activity of the neurogenic regulatory unit should also be analyzed in HSFTZVP16 embryos to determine whether this element participates in autoregulation. Although previous midies suggested thatjiz autoregulation does not begin until the later stages of cellularization (Yu and Pick 1995; Hyduk and Fercival-Smith 1996), we have shown that FTZ is capable of autoregulating its own expression in pre-cellularized embryos, which is when fc expression norrnally begins (Figure 3.38). This autocegulatory activity is weak to begin with but gradually strengthens. These weak early effects on the fi= gene are in contrast to the strong transcriptional activation of): by ODD at the same time (Saulier-Le Drean et al. 1998). We also know that at this early tirne, FTZ is a strong transcriptional activator of odd expression (Figure 3.6B). This suggests that, during the early stages offi expression,fiz autoregulation is predominantly indirect via odd. This mutual induction mode1 is consistent with the precise coincidence of the& and odd expression patterns in cellularking ernbryos (see Figure 1.5) (Manoukian and Krause 1992; Fujioka et al. 1993). In addition, it is also consistent with the observation that the early expression of odd is significantly reduced inJz mutant embryos (Figure ZAC), and bat an analogous reduction in expression occurs for fi: in odd mutant embryos (Saulier-Le Drean et al. 1998). Precedents for indirect autoregutation have been previously reponed for other homeodomain-containing regulaton (e-g. EN, UBX, DFD. LM) (Heemskerk et al. 1991; Gonzalez-Reyes et al. 1992: Tremml and Bienz 1992; Thuringer and Bienz 1993; Thuringer et al. 1993). However, these indirect pathways typically require signaling to and from nearby cells. The indirect& autoregulation by ODD could be required to assure an overlap beween the spatial expression of#z and odd during early cellularization. This, in turn, may be crucial for the precise temporal FTZ-dependent activation of en. en induction does not take place until the end of cellular blastoderm, despite the fact that FTZ is expressed at the end of syncytial blastoderm. This delay could be controlled by ODD which is a well-estabiished negative regulator of en (DiNard0 and O'Farrell 1987: Mullen and DiNardo 1995; SauIier-Le Drean et al. 1998) and whose expression overlapsfit expression during early cellularization (see Figure 1.5) (Manoukian and Krause 1992; Fujioka et al. 1995). At the end of ceIlularization, ODD is cleared out from the anterior-rnost FTZ expressing cells by activiues of me, opa and rpd3 (see Figure 1.5) (Manoukian and Krause 1992; Benedyk et al. 1994; Fujioka et al. 1995: Mamewik and Levine 1999). This allows FTZ to activate en in these cells.

43.5. MODULATORS OF FTZ ACTMTY 152 Ectopic expression of FTZ and FTZVPl6 does not lead to indiscriminate regulation of FTZ target genes (chapter 2, 3). This demonstrates that the activities of both factors are restricted. [t is possible to sumise the identity of some of the limiting factors based on the spatial and temporal domains of resmction. The following is a surnmary of these modulators and an assessrnent of their modes of action.

4.2.5.1. Factors aîfectiog the temporal wiodow of FTZ transcriptional activity

Both FTZ and FTZVP16 were e.utraordinarily ineffective pnor to 1:30 and after 4:30 AEL (Figure 3.28). This inactivity suggests the actions of global negative regulators such as those that modulate chromatin structure. Some of the factors that organize chromatin into transcriptionally inaccessible andor inert States have been demonstrated to be important regulators of FTZ target gnes. For example, the Polycomb (Pc) goup gene produco have been shown to regulate Ubx (Wedeen et al. 1986) and en (Moazed and OtFarreII I992), while the histone deacetylase RPD3 helps mediate repression of odd (Mannewik and Levine 1999). To test whether chromatin-modiQing complexes indeed affect the temporal window of FTZ-dependent regulation, the regulatory effects of FTZ and FTZVP16 could be analyzed in embryos carrying mutations in these factors. The ability of FTZ and FTZVP16 to prematurely activate FTZ target genes in a Pc gene mutant background, for example, would demonstrate chat Pc proteins affect FTZ activity. Importantly, premacure target gene induction by ectopic FTZ or FTZVP16 in embryos mutant for Pc group genes would redefine the role of the Pc products in gene regulation. Currently, Pc proteins are believed to control the maintenance rather than the initiation af gene expression (Bienz and Muller 1995: Simon 1995: Pirrotta 1997b; Pirrotta 1997a). The regulation of FTZ activiry by chromatin-modibing complexes may not be the only explanation for the inability of FTZ and FTZVPI6 to act beyond the narrow temporal window during early embryogenesis. Many FTZ target genes continue to be expressed in certain regions during later embryogenesis demonstrating that these genes are dlaccessible to various cranscription factors within those cells. Although FTZVP16 rnay not be capable of inducing FTZ target genes at these Iater stages due to. for example. the activity of chromatin-rnodiGing complexes, this chimeric transcription factor should be able to "super-activate" target genes in cells where they are nonnally being transcribed. However, activation above normal leveis of expression has only been observed during the times that FTZ is nonnally active. This argues for an additional level of regulation at these times that may be more FTZ-specific. This regulation could be exerted by repressors, or by the lack of essential CO-activator(s).A positively acting cofactor that rnay be Iirniting in older embryos is FTZ-FI. FTZ-FI is present during eariy embryogenesis due to matemally provided mRNA (Lavorgna et al. 1991; Ayer et al. 1993; Yu et al. 1997). Zygotic expression of FTZ-F1 is both temporaily and spatially restncted and daes not begin until late embryogenesis (Ueda et al- 1990; Lavorgna et al. 1991). To test whether a hnctional interaction with FTZ-FI could temporally [imit 153 FTZ activity, FTZ-FI could be ectopically expressed outside the temporal window of its endogenous expression and the effects of such expression on FTZ-dependent regulation could be analyzed. An additional issue to be addressed is how FTZ-FI affects FTZ function. In vitro studies demonstrated that FTZ-F1 physicatly interacts with FTZ (Guichet et al. 1997) and that both proteins bind cooperatively to cis-regulatory elernents derived fiom FTZ target genes such as en (Florence et al. 1997) and endogenous fir (Yu et al. 1997). This suggests that FTZ-FI probably helps recruit FTZ to target gene promoters. It would be of interest to investigate whether FTZ-FI activity is sufficient for FTZ-dependent regulation. If it is, then an enhancer containing only FTZ and FTZ- F1 sites would be expected to activate reporter gene expression in Drosophila embryos in a FTZ- and FTZ-FI-dependent manner. Biggin and McGinnis argue that FTZ-Fl does not affect recmitment of FTZ to target gene promoters but is required to modulate its transcriptional activity (Biggin and McGinnis 1997). This is based on the results of in vivo W-crosslinking studies which suggested that the DNA-binding specificity of FTZ is similar to that of EVE, another QSO homeodomain protein (Walter et al. 1994). This suggests that the in vivo binding of these proteins is predominantly determined by their highly conserved homeodomains rather than their specific interactions with cofactors. Cooperative binding with different cofactors woufd introduce significant differences in their DNA-binding specificities. Since MZ-FI is not required for EVE function and since the possibility that a yet unidentified cofactor binds cooperatively with EVE in the same way that FTZ- F1 and FTZ do is remote, it has becn proposed chat cofacton do not affect in vivo bindine of Q50 horneodomain proteins and that FTZ-FI is exchsively required for the transcriptional activity regulation of FTZ. To test this furthet. it would be of interest to analyze the activity of FTZVP t6 in embryos carryingfi-fl mutations. [f FïZVP16 can regulate FTZ target genes in the absence of FTZ- FI. this would dernonstrate that the requirement of FTZ-FI could be substituted by the heterologous activation domain, suggesting that FTZ-FI is not required for DNA binding but activity regulation of FTZ. The effects of FTZVP16 on 323 fPE (see Figure 1.8) could also be analyzed similarly. The activity of this element depends on severid heteroIogous binding sites. one of which is a repeated FANT module, which is bound by FTZ-FI in vitro (Han et al. 1993; Han et al. 1998). If indeed FTZ- F1 is required for FTZ transcriptional activity regulation rather than DNA binding, deletion of the repeated FANT module should have no effect on the activity of 323 fPE in HSFTZVP16 embryos- One of the earliest known represson offi gene expression (prior to cellularization), and hence a potential negative regulator of FTZ activity, is the zinc finger protein Tramtrack (TTK) (Harrison and Travers 1990; Brown et al. 1991; Read and ManIey 1992). TTK is uniformly expressed in syncytial stage embryos prior to the normal onset offi gene expression and begins to disappear as@ espression initiates (Hamson and Travers 1990: Brown et al. 1991; Read et al. 1992; Read and Manley 1992; Brown and Wu 1993)- [ts active may be dkted towards Fi7target gene promoten, FTZ itself or both. It wodd be of interest to establish the extent of the TTK effect on FTZ-dependent regulation by detennining whether FTZ target genes other then endogenousfi= are 154 also repressed by TTK. One way to proceed is to analyze the effect of ectopic expression of TTK on FTZ target gene regulation.

4.2.5.2. Functional interactions of FTZ with aperiodically distributed factors

At the beginning of cellularization. FTZVP16, and to a lesser degree FTZ, is able to activate target gene expression in the trunk of the embryo, but not at either of the termini (chapters 2, 3). This could reflect negative regulation of FTZ activity by members of the terminal group proteins or by terminally expressed gap proteins (Jurgens and Hartenstein 1993; Sprenger and Nusslein-Volhard 1993). Again, these may act on FTZ target gene promoters, FTZ itself or both. Since the terminal group of genes acts via a protein kinase cascade, and phosphorylation of FTZ has been shown to be important for its function (Krause and Gehring 1988; buse and Gehring 1989; Dong et al. 1998; Dong and Krause 1999), FTZ activity could be regulated by the terminal group of genes at the lever of posttranslational modifications. On the other hand, it is possible that the restriction of FTZ activity to the genn band does not result from negative regulation, but rather reflects positive functional interactions benveen FTZ and factors whose expression is restricted to the germ band. Potential candidates for such factors are pmducts encoded by the feashirr (rsh) (Fasano et al. 1991) and odd-paired (opa) genes (Benedyk et al. 1994). To explore Further what factors and mechanisms underlie restriction of FTZ activity to the germ band, FTZ activity should be analyzed in embryos canying mutations in the terminal group genes as well as mutations in opa and rsh. Analysis of the activities of FTZ polypeptides with point mutations in the hek-turn-helix motif of the homeodornain demonstrated that. although some mutant venions of FTZ could not comptetely rescue the fe phenotype, they stil! autoregulate eficiently in parasegment 8. even rescuing cuticle structures derived from this parasegment (Furukubo-Tokunaga et al. 1992). This suggests that FTZ activity ma): be somehow enhanced in this region of the ernbryo. My preliminary experirnents testing the regulatory effects of ectopic FTZ and FTZVP16 on target genes at midceIluIarization provided further support for the local enhancement of FTZ activity in the posterior portion of the ernbryo. In particular. 1 found that, althoughfiz autoregulation becomes greatly limited in odd-nurnbered parasegrnents, ectopic FTZ can still induce endogenous fc expression with high frequency in parasegments 9 and I I (Figure 4IB). At the same tirne, a similar induction of odd in these parasegments is detected in HSFTZVPI6 embryos (Figure 4.1F). In addition, ectopic FTZVP 16 can also induce wg in even-numbered parasegments 8 rind 10 (Figure 4.11). Figure 4.1. Enhancement of FTZ and FTZVPl6 activity in the posterior portion of the embryo. The effects of ectopic FTZ (B,E,H) and FTZVPI6 (C,F,I) are show on expression patterns of endogenousjiz (A-C), odd (D-F)and ivg (G-1) at rnidcellularization. Wild-type expression patterns are shown in the lefi-rnost column (A,D,G). Although at rnidcellularization fi- autoregulation becomes greatly Iirnited in odd-numbettd parasegments, ectopic FTZ can still induce endogenousft- expression with high penetrance in parasegments 9 and II (B) (a region marked with a bracket). Sirniiar localized induction is detected for odd (F) and wg (1) in HSFTZVP16 embryos.

157 These results demonstrate that the activity of FTZ and FTZVPI6 is enhanced in parasegments 8-1 1. Given the position of this effect and the timing of its occurrence. it is possible that this enhancement is due to positive tùnctional interactions between FTZ and one or more gap proteins expressed in parasegments 8- 1 1. A good candidate is knirps (hi)as it is expressed in these parase_ments (Rothe et al, 1989). Additional candidates are Kruppel (Kr) and giant (gr)whose expression patterns partially overlap the region demarcated by parasegments 8-1 1 (Mohler et aI. 1989; Hulskamp et al. 1990). Interestingly, a genetic interaction betweenfi and Kr has been reported previously (Duncan 1986). To establish whether gap proteins rnay be responsible for the enhancement of FTZ hnction in this posterior region of the embryo, FTZ activity could be analyzed in embryos mutant for hi,Kr and gr. In addition, the effects of ectopic coexpression of these genes together with FTZ could also be examined. Since this regionaI increase in activity was observed for both FTZ and FTZVP16, factors responsible for this enhancement most likely affect DNA binding rather than augmentation of FTZ transcriptional activity. A mechanism based on transcriptional activity rcgulation would be expected to be more FTZ-specific with a lesser effect on FTZVP16. The enhancement of FTZ Function in parasegments 8-1 1 may not result from a positive, but a negative interaction with factors that are expressed outside these parasegments One candidate is hunchback (hb), which is expressed in the anterior portion of the embryo (Schroder et al. 1988: Taue 1988). HB and FTZ have been reponed to exen opposing effects on Libr gene regulation (Muller and Bienz 1991: Qian et al. 1991; Muller and Bienz 1992: Uiang and Bienz 1992: Qian et al, 1993). It has been suggested that FTZ and HB regulate Ubx via the BRE and PBX enhancers of the U6.r promoter (Muller and Bienz 199 1: Qian et al. 199 1: Muller and Bienz 1992; Zhang and Bienz 1997; Qian et a1.1993)- In vitro studies identitïed several binding sites for FTZ and HB within these enhancers (Qian et al. 1991: Muller and Bienz 1992; Zhang and Bienz 1992: Qian et al. 1993). These sites are closely clustered, suggesting that HB abolishes FTZ-dependent activation by interfering with FTZ binding to the Ubx promoter. This mechanism is supported by in vitro binding studies demonstrating that HB bIocks the binding of FTZ and also displaces FTZ protein that is pre-bound to overlapping sites within the BEenhancer (Qian et al, 1993). It would be of interest to examine the regulatory effect of FTZVPI6 on Ubx. lf indeed FTZ cannot occupy the Ubx promoter in the presence of HB, then ectopic FTZVPI6 should be unable to activate übx in the anrerior portion of the ernbryo. However, if FTmP16 can activate Ubx anteriorlyl this would suggest that Hl3 blocks FTZ activity. not DNA binding. In addition to Ubx. the spatial expression of other FTZ-regulated homeotic genes should be analyzed in HSFTZVPl6 embryos to determine the rnechanisrns underlying the spatial restrictions of their activation.

1.2.5.3. Functional interactions of ET2 with periodicalty distribated factors

By mid-cellularïzation FTZ activity begins to be modulated with a pair-nile penodicity within the tmnk of the embryo. This negative effect, which is exerted on FTZ but not FTZVPI6 (compare 158 Figures 3.3E and 3.3F). is likely to be mediated by the pair-rule protein Hairy (H). In hairy mutant embryos. endogenous fir gene expression expands to fil1 the mnk of the embryo (Carroll and Scott 1986; Howard and Ingham 1986: Ingham and Gergen 1988; Carroll and Vavra 1989: Lawrence and Johnston 1989a; Tsai and Gergen t995; Yu and Pick 1995). Conversely. when H is expressed throughout the embryo,fr= stripes are rapidly repressed (Ish-Horowicz and Pinchin 1987). In HSFTZ embryos, there is a decrease in the response off^ in odd-numbered parasegments toward the end of cellularization (Figure 3.3E). This is when and where endogenous fc stripes begin to expand in hairy mutant embryos (Yu and Pick 1995). The ability of FTZVP16 to overcome repression in these regions (Figure 33F) suggests that H does not affect promoter occupancy of FTZ. H is a heliu-loop- helix protein (Rushlow et al. 1989) that requires the widely utilized co-repressor protein Groucho (GRO) for some of its tùnctions as a repressor (Paroush et al. 1994: Fisher et al. 1996: Jimenez et al. 1997). Repression offi= ha been dernonstrateci to be GRO-dependent (Jimenez et al. 1997). The actions of VPl6 can clearly overcome the effects of these nvo proteins. The detailed mechanisrn by which H and GR0 regulate gene expression is not known. Interestingly. it has been demonstratecl that H physically interacts with FTZ (Copeland 1997). It would be of interest to address the relevance of this in vitro interaction. One way to proceed is to map the interacting domains and analyze whether deletion or mutation of these domains allows FTZ to overcome H-dependent repression. lt would also be of interest to determine whether FTS interacts physically with GRO. Although ectopic FTZVPI6 can overcome the early effects of H. it cannot overcome negative effects occumng in the same regions (odd-numbered parasegments) during gastrulation (Figure 331). This suggests the activity of a different repressor and a different mode of repression. A candidate for this negative regulator is EVE. EVE is expressed in odd-numbered parasegments (Frasch and Levine 1987) and genetic anaiysis suggests that EVE is a repressor offi expression (Fmsch et al. 1988). However. EVE begins to be expressed weIl before the start of gastrulation. Indeed. experiments with ectopically expressed EVE showed that, during cellularization, EVE actuaIIy acts as an activator offi= expression (Manoukian and Krause 1992). Only at the onset of gastrulation does it become a repressor. In vitro and tissue culture studies have suggested that, when EVE acts as a repressor. it may be doing so by interferhg with either TBP function or TBP binding to TATA elements (Han and Manley 19936: TenHiumsel et al. 1993; Austin and Biggin 1995; LJm et al. 1995: Li and Manley 1998)- This possibility would be consistent with the inability of FTZVPi6 to ovemde the effects of EVE: TBP is one of the basa1 mscnption factors required for RNA polymerase ii binding and activity (Harnpsey 1998). Interestingly, uniike ft:. the en gene can be activated by FTZVP16 in regions where EVE is present (compare Figures j3I and 3.40. This may be explaincd by previous studies which showed that en does not appear to be directly regulated by EVE (Manoukian and Krause 1993: Fujioka et al. 1995). During germ band extension, two other negative regdators of& expression begin to exert their effects; the pair-rule proteins SLP and ODD. Both normally affectfi expression in even- numbered parasegments (Cadigan et ai. 1994b; Mullen and DiNardo 1993'). In slp mutant embryos, an 159 additional set ofjz stripes appem in the posterior regions of even-numbered parasegments during late stages of germ band extension (Cadigan et al. 1994b). This suggested that SLP normally represses fi: at these late stages offi expression. 1 have carried out some preliminary rxperiments to test whether SLP can act as a negative regularor of FTZ activity during eady stages of ernbryogenesis. In particular. 1 have analyzed the effect of ectopic FTZ on endogenous fc expression in a slp mutant background. Figure 43B shows that& stripes have not yet expanded in stage 7 dp mutant embryos, suggesting that SLP rnay not be the sought fier negative regulator. However. when FTZ is expressed ectopically in this same genetic background (Figure 3.7D), endogenous fi: stripes expand not only anteriorly, as obsewed in wild-type embryos (Figure 4.X) but posteriorly as well. This pattern is identical to that generated by FTZVP16 in wild-type embqos (compare Figures 4.2D and 3-31), suggesting that SLP-mediated repression is overcome by the actions of the -16 activation dornain. The ability of FTZVP16 to ovemde SLP activity would suggest that. as with H, the action of SLP is normally exened at steps subsequent to promoter binding by FTZ. Like H, SLP has been demonstrated to physically interact with FTZ in vitro (Copland 1997). It would be of interest to determine whether this interaction is required for antagonizing FTZ-dependent regulation. For this purpose. the interaction dornains would have to be mapped and the effects of deleting or rnutating of these domains examined. As with EVE. ODD appears to provide an effective block of both FTZ and FTZVP 16 activities. These effects are cornplex, vaqing in a prornoter- and nage-specific fashion. For example, previous studies have shown that ODD is a potent repressor of en in pre-cellularized ernbryos, and at the same time. an acrivator of its own expression and that off?: (Saulier-Le Drean et al. 1998). At gastrulation, repression of en continues (Mullen and DiNudo 1995: Saulier-Le Drean et al. 1998), but ODD then switches frorn an activator to a repressor of& (see Figure 1.6A) (Mullen and DiNardo 1995: Saulier-Le Drean et al. 1998). This leads to the resolution offi mipes towards the anterior margins of even-nurnbered parasegments. Since FTZ positively regulates odd (chapter 7), this also results in odd mipe resolution. The activities of ODD are reflected by the actions of ectopic FTZ and FïZVP16 on the en.& and odd promoters. In cellularizing embryos. at which tirne ODD is expressed throughout the tmnk of the embqio. FTZ is unable to activate rn (Figure 3.4B) but at the same tirne can weakly activate endogenous @ expression (Figure 33B) and eficiently activate odd (Figure 2.6B). FTmP16 induces simiiar responses: modest activation offi (Figure 33C). strong activation of odd (Figure 3.7C) and loss of its prior ability to activate en (Figure 3.4F) (prior to the initiation of odd expression, FTZVP16 can activate en in the tmnk (Figure 3.4C)). Upon eastnilation. FTZ is able to activate neither en nor ft in ODD-expressing cells. FTZW 16 also b exhibits similar limitations at this stage. As has been suggested for EVE, this suggests that ODD may 160 Figure 13 Negative regulation of FTZ activity by SLP. The effect of SLP onfi= regulation was analyzed in slp- (B) and slp-;HSFTZ (D) embryos during early gennband extension. Expression of& in wild-type (A) and HSFTZ (C) embryos is show for cornparison. In slp embryos (B),fr,- expression is reinitiated during early germband extension in the posterior regions of even-numbered parasegrnents. This ectopic expression of the fi=gene (marked by arrows in D) can be significantly enhanced by ectopic induction of FTZ, but is not detected in HSFTZ embryos that cary a wild-type slp gene (C). The effects of FTZ in a slp mutant background (D) are the sme as those of FTZVP16 in a wild-type background (Figure 3 3 1).

162 repress transcription by interfenng with basal transcription factor activities, Aitematively, ODD may interfere with FTZ function by preventing it fiom binding to target gene promoters. It would be of interest to determine what regulatory units of thefi= prornoter rnediate regulation by ODD. Are the opposing effects of ODD mediated by the same regulatory elements? This could be addressed by anaiyzing the effects of ectopic ODD on individual fi= promoter elements. Funire studies will focus on elucidating the mechanisrns by which already identified factors such as H, EVE, SLP and ODD affect FTZ-dependent regulation. In addition. novel factors that rnodiQ FTZ function should be isolated and the rnechanisrns of their action probed- These factors could be identified using biochemical rnethods. For exarnple. a two-hybrid screen (Fields and Song 1989; Fields and Sternglanz 1994) could be applied to identiQ proteins interacting with FTZ. Gel retardation experiments with ernbryonic protein extracts and the autonomous FTZ-dependenc regulatory units frorn different target genes could be used to identi@ protein complexes occupying these cis-acting elements, as was applied to thefi= autoregulatory element (Han et al. 1993: Han et al. 1998). Factors binding to the autoregdaton, element were also identified using a one-hybnd screen in yeast (Yu et al, 1999). This method could be expanded to inctude other regulatory units mediating FTZ-dependent regulation. Genetic selection in yeast provides sorne distinct advantages over a strictly in vitro approach. In particular. unlike in vitro DNA-binding techniques. this method has the potential to identiQ factors that also affect aspects of transcriptional regulation. tn addition. since it is a clonal method. it rnay be able to detect proteins present in ernbryonic emcts at Iow concentrations. Novel genes affecting FTZ-dependent regulation could be atso isolated based on genetic rnethods camed out in Drosophila. One way to proceed is to screen for fc modifiers that enhance the reduced viability of partial loss-of-function alleles. The rationate is to have& hnction at such a low level that the elimination of one of the two doses of gene pniduct From a modi@ing locus would result in Iethality. Analogous genetic screens for haplo-insuficient enhancers have ken carried out and have provided insights into a variety of developmental processes (GenIer et a[. 1989: Simon et al. 1991: Doyle and Bishop 1993; Harding et al. L995: Raflery et al. 1993: Gellon et al, 1997). In the screen forfi,- modifiers,fi= function can be reduced by usinefi,- temperature sensitive aIleIes. Loci identified in such a screen rnay not only include factors that act in paralle1 with FTZ but also those that act upstrearn or dowsueam of FTZ. Sincejk autoreplates. some of the factors conhoiiingfi expression may also directly affect FTZ function.

12.6. MECHAMSMS CONTROLLING RECULATORY .QM) DEVELOPMENTAL SPECiFICITIES OF Q50 HOMEODOMNN PROTEINS

Analysis of genomic expression profiles in HSFTZ andfi mutant embryos will detemine whether FTZ regulates a Iimited number or the majority of Drosophila genes. If Fi2 reglates a Iimited nurnber of genes, this would be consistent with the results presented in chapter 3 that argue against widespread distniution of FTZ on chrornatin. This would support a notion that specificity of 163 QSO homeodomain proteins is determined at the Ievel of DNA binding (Garcia-Bellido 1977; Andrew and Scott 1992; Botas 1993). As a further [est, it would be of interest to determine, using genome- wide analysis, how rnany and what genes are regulated by other Q50 horneodornain proteins. for example, what genes are regulated by EVE? EVE is ternporally coexpressed with FTZ dunng early ernbryogenesis (Harding et al. 1986; Macdonald et al. 1986). In vivo UV-crosslinking studies suggest that EVE and FTZ have very similar DNA-binding specificities (Walter et al. 1994). If this werr indeed the case, then FTZ and EVE would be expected to regulate the same set of target genes. The analysis of the effects of FTZVP16 on segmentation gene regulation (chapter 3) revealed that only a subset of these genes respond in HSFTZVP 16 ernbryos. Al1 the genes that responded to ectopic FTZVP16 also responded to ectopic FTZ suggesting that. normally, FTZ does not occupy the prornoters of non-target genes. tt could be argued that a lirnited number of genes were investigated and that a more cornprehensive study is required to detennine whether in vivo binding of FTZ is limited. It would be of interest to analyze variations in the global transcript profile in HSFTZVP16 embryos to determine whether FTZVP16 regulates the sarne or larger set of genes than FTZ. If FTZVPI6 regulated a larger set of genes than FTZ, this would indicate that sorne promoten are bound by FTZ in vivo but not regulated, suggesting that not only selective binding but also transcriptional activity regulation determines what genes becorne targets of FTZ-dependent regulation. The levels of ectopic FTZ or FTZVf 16 that have been induced in our experiments are lower than those of endogenous FTZ (chapter 3). It is possible that, when protein levels are low, the functional specificity of FTZ depends predominantly on selective DNA binding. In panicular. FTZ at Low levels would only occupy the highest affinity binding sites. Cooperative binding with cofactors would assure preferential occupation and regdation of FTZ target gene prornoters. At higher protein concentrations, however, FTZ could bind in a more widespread fashion, occupying sorne of its binding sites independently of cofactors. Since potential FTZ binding sites are dism3uted abundantly throughout the Drosophila genorne, higher levels of FTZ rnay bind to many additional potential target gene prornoters. If this were the case, then FTZ target gene selection at high protein concentrations rnay predominantly depend on transcriptional activity regulation. Since endogenous fi= is detected at different levels throughout its phases of expression, selection of FTZ target genes could be controlled by both DNA binding and transcriptional activity regulation. To test whether higher levels of FTZ do indeed tn'gger more widespread binding, it should be established whether the global transcript profile in HSFTZVP16 embryos depends on the level of FTZVP16. The level of FTZVP16 can be elevated, for example, by increasing the durations of heat-treatments. To tiirther investigate the in vivo binding of FTZ, it would be of interest to determine whether FTZ recognizes and occupies hi@ a£Eîîity sites in vivo, as identified in vitro, without the aid of cooperative interactions with cofactors. Previous studies demonstrated that, when rnultirnerized, high afinity sites of FTZ alone couid not confer FTZ-dependent regdation on the reporter genes (Vincent et al. 1990). Sirnilariy, it was found that sequences sunoundhg the BRE moduIe within the 164 Ubx promoter contain very strong FTZ binding sites that do not direct a FTZ-dependent expression pattern (Qian et al. 1991). This could result fiom an inability of FTZ to bind to these sites in vivo. Alternatively, FTZ could recognize and occupy these sites, but due to the lack of additional binding sites for cofactors regulating FTZ transcriptional activity, FTZ would reside in a functionally neutral state. The latter possibility is consistent with the observation that transcriptional activity regulation plays an important role in FTZ function (chapter 3) and with the finding that another QSO homeodomain protein DFD can occupy in vitro binding sites in vivo without additional sites for cofactors, but remains in a functionaily neutral state (Li et al. 1999a). To test whether FTZ occupies high affinity binding sites in vivo without the aid of cofactors, the regulatory effects of FTZVPI6 on reporter genes controlled by these sites should be investigated. If FTZVP16 can activate these reporter genes, this would indicate that FTZ is capable of recognizing and occupying binding sites without the aid of cofactors. It would funher confirrn the role of transcriptional activity regulation in FTZ target gene selection and FTZ functional specificity. it would be of interest to deterrnine how this in vivo occupancy of FTZ binding sites depends on FTZ levels. 1s the binding of FTZ to its in vitro high affinity sites relevant to FTZ function in vivo? The close match between FTZ target sequences in the autoregulatory element of thefi2 upstream enhancer (Pick et al. 1990; Schier and Gehring 1992; Schier and Gehring 1993a) and the FTZ consensus sequences derived from binding site selection and mutational studies (Florence et al. 1991) argues that fùnctional FTZ sites in vivo correspond to the consensus sequences recognized in vitro. Further support cornes kom experiments demonstrating that in vitro measured affinity differences between FTZ binding sites show a good correlation to the transcriptional activity of complex regulatory eiements containing these sites as measured in transgenic flies (Schier and Gehring 1992; Schier and Gehring 1993b). A similar correlation between the in vitro affinity and the in vivo activity of binding sites has also been reported for another homeodomain protein BCD (Driever and Nusslein-VoIhard 1988a; Driever et al, 1989). Despite these results, functional analysis of thefi,- autoregulatory element revealed that deletion of the single high afinity binding site does not detectably affect its function and that multiple medium affînity binding sites are suficient to mediate DNA binding and transcriptional regulation by FTZ (Pick et al. 1990: Schier and Gehring 1992; Schier and Gehring 1993a). Strikingly, most of the medium affinity sites are conserved in Drosophila virilis and Drosophila hydei (Maier et al. 1990), which have presumably diverged fiom Drosophila melanogasfer some 60 million years ago. This suggests that in vivo DNA binding specificity of FTZ rnay be somewhat relaxed and that the high affhity sequences, as selected in vitro, may not be important target sequences occupied in vivo. The in vivo recognition of medium affinity sites might help to discriminate against closely related horneodomain proteins expressed in the sarne cells at the same the, This could be possible if these proteins differed in their binding behaviour towards non- optimal sites, despite the fact that al1 of them recognize with similar hi& affimity the sarne consensus binding site. Altematively, the occupancy of medium fliity sites could require 165 cooperative binding with specific cofactors, which would recognize adjacent sequences. This would ensure selective binding of different homeodomain proteins to their regulatory elements.

1.3. CONCLUSION

The first step in the direction of understanding the molecular aspects of FTZ hnction in vivo is to identiS, genes that are directly regulated by FTZ. This has been approached using a combination of genetic and molecular techniques. As a ~sult.several FTZ target genes have been identified. A more comprehensive analysis such as a genome-wide DNA microarray-based approach is now required to determine whether these genes represent ail or only a subset of genes directly tegulated by FTZ. Future studies will elucidate molecular aspects of FTZ-dependent regulation. This should reveal the mechanisms allowing FTZ to act as both a transcriptional activator and a tepressor. Characterization of response elements mediating FTZ-dependent regulation as well as the identification of cofactors participating in this regulation could provide valuable insights into these mechanisms. As with the activities of other homeodomain proteins, FTZ function in vivo is expected to depend on a variety of factors. The specificity code generated by these factors is likely to Vary depending on the cell type, the presence or absence of cenain cofactors in the cell, the stage of development, etc. The challenge in the future will be to apply a variety of in vivo approaches to reveal the molecular aspects of the mechanisms controlling FTZ fünction. Refereoces

Acampora, D., M. D'Esposiro, A. Faiella, M. Pannese, E, Migliaccio, F. Morelli, A. Stomaiuolo, V. Nigro, A. Sirneone, and E. Boncinelli. 1989. The human HOX gene famiIy. Xucleic Acids Res If: 10385-403.

Ace, CI.,MA DaIrymple, F.H. Ramsay, V.G. Preston, and C.M. Preston. 1988. Mutational analysis of the herpes simplex virus type 1 ms-inducing factor Vmw62. J Gen Virol69: 2595-605.

Adams, M.D. et aI. 2000. The genome sequence of Drosophila mslanogaster. Science 287: 7185-95.

Affoiter, M., A. Percival-Smith. M. Muller, W. Leupin, and WJ. Getsring. 1990. DNA binding pmperties of the purified Antennapedia homeodomain. Proc Nati Acad Sci U S .4 87: 4093- 7.

Akam. M.E. 1983, The Iocation of Ultrabitharav transcrïpts in Drosophila tissue sections. Embo J 2: 2075-84.

Akam, M.E. and A. Martiner Arias. 1985. The distribution of iiitrabithorm vanscripts in Drosophila embryos. Embo J4: 1689-1700,

Akam, M.E., A. Martinez-Arias, R. WeinzierI, and C.D. Wilde. 1985. Function and expression of ultrabithorax in the Drosophila embryo. Cold Spring Hurb Svmp Quant Biol 50: I!E-1OO.

Ananthan, J., R. Baler, D. Morrissey, J. Zuo, Y. Lm, M. Weir, and R. Voellmy. 1993. Synergistic activation of transcription is rnediated by the N-terminal domain of Drosophila fishi tarazu homeoprotein and can occur without DNA binding by the protein. Md Ce11 Biol 13: 1599- 609,

Andrew, D.J. and M.P. Scott. 1992. Downstream of the homeotic genes. ~VavBiol 4: 5-1 5.

Appel, B. and S. Sakonju. 1993. Cetl-type-specific mechanisms of transcriptional repression by the homeotic gent products UBX and ABD-A in Drosophita embryos. Embo J12: 1099-109. apRhys, C.M., D.M. Ciufo. EA. ONeill, T.J. Kelly, and G.S. Hayard. 1989. Overiapping octamer and TAATGARAT motifs in the VF65-response ekments in herpes sirnplex virus immediate- early promoters represent independent binding sites for cellular nuclear factor III. J Viro163: -798-8 11.

Arnold, H.H. and B. Winter. 1998. Muscle differenciation: more cornplexity to the nenvork of rnyogenic regdators, CmOpin Genet Dev 8: 539-4.

Arnosti, D.N., S. Bamb. M. Levine, and S. Small. 1996- The eve stripe 2 enhancer employs multiple modes of transcriptional synergy. Development 122: 205-14.

Ashburner, M.! C. Chihara, P. Meltzer, and G. Richards. 1974. Temporal conml of puffing activity in polytene chromosomes. Coid Spring Harb Svmp Qum Bio138: 655-67.

Austin, R.J. and M.D. Biggin. 1995. A domain of the even-skipped protein represses transcription by preventing TFIID binding to a promoter, reptession by cooperative blocking ~.lolCe11 Biul 15: 4683-93.

Ayer, S., N. WaIker, M. Mosammaparasq J.P. Neison, B.2- Shilo, and C. Benyajati. 1993. Activation and repression of Drosophila alcohol dehydrogenast distal transcription by two steroid hormone receptor superfamily members binding to a comman response element. Nucleic Acids Rus 21: 1619-27. Baker, N.E. 1987. Molecular cloning of wingless, a segment polarity gene in Drosophila: the spatial distribution of a transcript in embryos. Embo J6: 1765-1773.

Baker, N.E. 1988. Localization of transcripts frorn the wingless gene in whole Drosophila embryos. Development 103: 289-98.

Bateson, W. 1894. Materials for the study of variation with especial regard to discontinuity in the origin of the species. In. McMillan, London.

8anerson, W. and B. Rohan. 1983. Characterization of the herpes simplex virion-associated factor responsible for the induction of alpha genes. J Viral 46: 37 1-7.

Baumgartner, S., D. Bopp, M. Burri, and M. Noll. 1987. Structure of two genes at the gooseberry locus related to the paired gene and their spatial expression during Drosophila ernbryogenesis. Genes Dev 1: 1247-67. Baumgartner, S. and M. Noll. 1990. Network of interactions arnong pair-mle genes regulating paired expression during primordial segmentation of Drosophila. hfech Dev 33: 1- 18.

Baumruker, T., R. Sturm, and W. Hem. 1988. OBPlOO binds rernarkably degenerate octamer motifs through specific interactions with flanking sequences. Genes Dev 2: 1400-13. Beachy, P.A., S.L. Helfand, and D.S. Hogness. 1985. Segmental distribution of bith0ra.u complex proteins during Drosophila development. Nature 313: 545-51.

Beachy, P.A., M.A. Krasnow, E.R. Gavis, and D.S. Hogness. 1988. An Ultrabithorav protein binds sequences near its own and the Antennapedia P 1 promoters. Cell55: 1069-8 1.

Bejsovec, A. and A. Martinez Arias. 1991. Roles of wingless in panerning the Iarval epidermis of Drosophila. Developmenr 113: 471-85.

Bender, W., M.E. Akam, F. Karch, P.A. Beachy, M. Peifer, P. Spierer. and E.B. Lewis. 1983. Molecular genetics of the bithorax cornplex in Drosophila melanogaster. Science 22 1 : 23-9.

Benedyk, M.J., J.R. ~Mullen,and S. DiNardo. 1994. odd-paired: a zinc finger pair-rule protein required for the timely activation of engrailed and wingless in Drosophila embryos. Genes Dev 8: 105- 17.

Berthelsen, J., V. Zappavigna, F. Mavilio, and F. Blasi. 1998. PrepI, a noveI tbnctional partner of Pbx proteins. Embo J 17: 1423-33.

Bhat, K.M.,G. Farkas, F. Karch, H. Gyurkovics, J. Gausz, and P. Schedi. 1996. The GAGA factor is required in the early Drosophila embryo not only for transcriptionai regdation but also for nuclear division. Development 122: 11 13-24.

Bienz, M. and J. Muller. 1995. Transcriptional silencing of horneotic genes in Drosophila. Bioessays If: 771-84.

Bienz, M. and G. Trernml, 1988. Domain of Ultrabithorav expression in Drosophila visceral mesoderm from autoregulation and exclusion. Nature 333: 576-8.

Biggin, M.D. and W. McGinnis. 1997. Regdation of segmentation and segmenta1 idenhty by Drosophila homeoproteins: the role of DNA binding in hnctional activity and specificity. Development 124: 4425-33.

Biggin, M.D. and R. Tjian. 1988. Transcription factors that activate rhe Ultrabithom promoter in devetopmentally staged extracts. Ce11 53: 699-7 11. Biggin, M.D. and R. Tjian. 1989. A purified Drosophila homeodomain protein represses transcription in vitro. Cell58: 43340.

Billeter, M., Y.Q. Qian, G. Otting, M. Muller, W. Gehring, and K. Wuthrich. 1993. Determination of the nucIear magnetic resonance solution structure of an Antennapedia homeodomain-DNA cornplex. J Mol Biol 234: 1084-93.

Binari, R, and N. Pemmon. 1994. Stripe-specific regulation of pair-rule genes by hopscotch, a putative Jak family tyrosine kinase in Drosophila. Genes Dev 8: 300-12.

Blair, S.S. 1993. Mechanisms of comparîment formation: evidence that non-proliferating ceils do not play a cntical role in defining the DN lineage testriction in the developing wing of Drosophila. Dt.velopment 119: 339-5 1.

Bopp, D,, M. Bum, S. Baumgarîner, G. Frigerio, and M. Noll. 1986. Conservation of a Iarge protein domain in the segmentation gene paired and in functionally related genes of Drosophila. Cell 47: 1033-40.

Botas, 1. 1993. ControI of morphogenesis and differentiation by HOMMox genes. Curr Opin Cd Bi01 5: 10 15-23.

Botas, 3. and L. Auwers. 1996. Chromosomal binding sites of Ultrabithom homeotic proteins. me ch Dev 56: 129-38.

Boulet, A.M. and M.P. Scott. 1988. Control elements of the P2 promoter of the Antennapedia gene. Genes Dev 2: 1600-14.

Bourouis, M., P. Moore, L. Ruel, Y. Grau, P. Heitzler, and P. Simpson. 1990. An early embqonic product of the gene shaggy encodes a serinelthreonine protein kinase related to the CDC28/cdc2+ subfamily. Embo J9: 7877-84.

Breen, T.R. and P.J. Harte. 1993. Trithorax regulates multiple homeotic genes in the bithorax and Antennapedia complexes and exerts different tissue-specific, parasegment-specific and promoter-specific effects on each. Developmenr 117: 119-34.

Bridges, C. and T.H. Morgan. 1923. The third-chromosome group of mutant characters of Drosophila melanogaster. Carnegie Insritution of Washington Publications 327: 1-25 1.

Britten, RJ. and E.H. Davidson. 1969. Gene regulation for higher cells: a theory. Science 165: 349- 57.

Bronner, G. and H. Jackle. 1991. Control and function of terminal gap gene activity in the posterior pole region of the Drosophila embryo. Mech Dev 35: 205-1 1.

Brown, J.L., S. Sonoda, H. Ueda, M.P. Scott, and C, Wu. 1991. Repression of the DrosophiIa fushi tarazu (ftz) segmentation gene. Embo J 10: 665-74.

Brown, J.L. and C. Wu. 1993. Repression of Drosophila pair-rule segmentation genes by ectopic expression of tramtrack, Development 117: 45-58.

Brown, P.O. and D. Botstein. 1999. Exploring the new world of the genome with DNA microanays. Xat Genet 2 1: 33-7.

Brun, RP., J.B. Kim, E. Hu, S. Altiok, and B.M. Spiegelman. 1996. Adipocyte differentiation: a transcriptional regdatory cascade. CmOpin Cell Biol 8: 826-32. Burglin, T.R. 1994. A comprehensive classification of horneobox genes. In Guidebook to the Homeobox Genes (ed. D. Duboule), pp. 26-71. Oxford University Press, OIrford, England.

Burglin, TR. L997. Analysis of TALE superclass horneobox genes (MEIS, PBC, KNOX, Iroquois, TGIF) reveals a novel domain conserved between plants and animals. Nucleic Acids Res 25: 4 173-80.

Burglin, T.R. and G. Ruvkun. 1992. New motif in PBX genes [letter]. Nat Genet 1: 3 19-20. Burz, D.S., R, Rivera-Pomar, H. Jackle, and S.D. Hanes. 1998. Cooperative DNA-binding by Bicoid provides a mechanism for threshold- dependent gene activation in the Drosophila embryo. Embo J 17: 5998-6009.

Cadigan, K.M., U. Grossniklaus, and WJ. Gehring. 1994a. Functional redundancy: the respective roles of the two sloppy paired genes in Drosophila segmentation. Proc Nad Acad Sci USA 91: 6324-8.

Cadigan, KM., U. Grossniklaus, and WJ. Gehring. 1994b. Localized expression of sloppy paired protein maintains the polarity of Drosophila parasegments. Genes Dev 8: 899-913.

Cai, J., Y. Lm, L.F. Appel, and M. Weir. 1994. Dissection of the Drosophila paired protein: functional requirements for conserved motifs. Mech Dev 47: 139-50.

Campbell, M.E., J.W. Palfreyman, and C.M. Preston. 1984. Identification of herpes simplex virus DNA sequences which encode a trans-acting polypeptide responsible for stimulation of immediate early transcription. J Mol Bi01 180: 1-19.

Capovilla, M. and J. Botas. 1998. FunctionaI dominance among Hox genes: repression dominates activation in the regulation of Dpp. Devrlopment 125: 4949-57.

Capovilla, M., M. Brandt, and J. Botas. 1994. Direct regulation of decapentaplegic by UItrabithom and its role in Drosophila midgut morphogenesis. Ce11 76: 361-75.

Capovilla, M., E.D. Eldon, and V. Pimotta. 1992. The giant gene of Drosophila encodes a b-ZIP DNA-binding protein that regulates the expression of other segmentation gap senes. Development 1lJ: 99- 1 12.

Carr, A. and M.D. Biggin. 1999. A cornparison of in vivo and in vitro DNA-binding specificities suggests a new mode1 for homeoprotein DNA binding in Drosophila embryos. Embo J 18: 1598-608.

Carroll, S.B. 1990. Zebra patterns in fly embryos: activation of smpes or repression of interstripes? CeIl 60: 9- 16.

Carroll, S.B., A. Laughon, and B.S. Thalley. 1988. Expression, function, and regulation of the hairy segrnentation protein in the Drosophila embryo. Genes Dev 2: 883-90.

Carroll, S.B., RA. Laymon, MA. McCutcheon, P.D. Riley, and M.P. Scott. 1986a. The localization and regulation of Antennapedia protein expression in Drosophila embryos. Ce11 47: 113-22.

Carroll, S.B. and M.P. Scott. 1985, Localization of the hhi tarani protein during DrosophiIa embryogenesis. Ce11 13: 47-57.

Carroll, S.B. and M.P. Scott. 1986. Zygoticaliy active genes that affect the spatial expression of the fùshi tarani segmentation gene du~gearly Drosophila embryogenesis. Ce11 45: 113-26. Carroll, S.B. and S.H. Vavra. 1989. The zygotic control of Dmsophila pair-mle gene expression. 11. Spatial repression by gap and pair-mle gene products. Developmenr 107: 673-83.

Carroll, S.B., G.M. Winsiow, T. Schupbach, and M.F. Scott. L986b. Materna1 connol of Drosophila segmentation gene expression. Nature 323: 278-80.

Casanova, J. and RA. White. 1987. Tram-regulatory functions in the Abdominal-B gene of the bithora.. complex. Development 101: 117-22.

CasteIli-Gair, J. and M. Akam. 1995. How the Hox gene Ultrabithomv specifies two differenr segments: the significance of spatial and temporal regulation within metameres. Developmenr 121: 2973-82.

Celniker, S.E., D.J. Keelan, and E.B. Lewis. 1989, The molecular genetics of the bithorax complex of Drosophila: characteriziition of the products of the Abdominal-B domain. Genes Dev 3: 1424-36.

Chadwick, R. and W. McGinnis. 1987. Temporal and spatial distribution of transcripts from the Defurmed gene of Drosophila. Embo J6: 779-789.

Chan, C.S., L. Rastelli, and V. Pirrotta. 1994a. A Polycomb response erement in the Ubx gene that determines an epigenetically inhented state of repression. Embo J 13: 3553-64.

Chan, S.K., L. Jaffe, M. Capoviila, J. Botas, and R.S. Mann. 1994b. The DNA binding specificity of Ultrabithorav is moduiated by cooperative interactions with extradenticle, another homeoprotein. Cd78: 603- 15.

Chan, S.K. and R.S. Mann. 1993. The segment identity functions of Ultrabithorm are contained within its homeo domain and carboxy-terminal sequences. Grnes Dev 7: 796-8 1 1.

Chan, S.K. and R.S. Mann. 1996. A structural mode1 for a homeotic protein-extradenticle-DNA complex accounts for the choice of HOX protein in the heterodimer. Proc Nat1 Acad Sei US A 93: 5223-8.

Chan, S.K.. H. Popperl. R. Krumlauf, and R.S. Mann. 1996. An extradenticle-induced conformational change in a HOX protein overcomes an inhibitory function of the conserved hexapeptide motif. Embo J 15: 2476-87.

Chan, S.K.. H.D. Ryoo, A. Gould, R. Krumlauf, and R.S. Mann. 1997. Switching the in vivo specificity of a minimal Hox-responsive element, Developmenc 121: 2007-14.

Chang, C.P., L. Brocchieri. W.F. Shen, C. Largman, and M.L. Cleary. 1996. Pbx modulation of Hox homeodomain amino-terminal arms establishes different DNA-binding specificities across the Hox locus. Mol Ce11 Bi01 16: 1734-45.

Chang, C.P., W.F. Shen, S. Rozenfeld, H.J. Lawrence, C. Larpman, and M.L. Cleary. 1995. Pbx proteins display hempeptide-dependent cooperative DNA binding with a subset of Hox proteins. Genes Dev 9: 663-74.

Cho, K.W., J. Goea; C.V. Wright, A. Fritz, I. Hardwicke, and EMDe Robertïs. 1988. Differential utilization of the same reading Frame in a Xenopus homeobox gene encodes two related proteins sharing the same DNA-binding specificity. Embo J7: 713949.

Chouinard, S. and T.C. Kaufman. 1991. Control of expression of the homeotic labial (lab) locus of Drosophila melanogaster: evidence for both positive and negative autogenous regulation. Developmenr 113: 1267-80. 17 1 Clark, KA. and D.M. McKearin. 1996. The Drosophita stonewall gene encodes a putative transcription factor essential for genn ce11 development. Developmenr 122: 937-50.

Clark, S.G., A.D. Chisholm, and H.R. Horvitz, 1993. Conml of ce11 fates in the central body region of C. elegans by the homeobox gene lin-39. CeIl 74: 43-53'.

Clevers. H.C. and R. Grosschedl. 1996, Transcriptional conuol of lymphoid development: lessons from gene targeting. Immun01 Toduy 17: 336-43.

Cohen, B., E.A. Wimmer, and S.M. Cohen. L991. Eady development of leg and wing primordia in the Drosophila embryo. Mech Dev 33: 239-40.

Colgan, J., S. Wampler, and J.L. Manley. 1993. Interaction between a transcriptional activator and transcription factor [IB in vivo. Nafure 362: 549-53.

Copeland, J.W. 1997. The role of protein-protein interactions and phosphorylation in the function of the homeodomain protein Fushi tarazu, In, University of Toronto.

Copeland, J.W., A. Nasiadka, B.H. Diemch, and H-M. Krause. 1996. Patterning of the Drosophila embryo by a homeodomain-deieted Fh polypeptide. !Varure 379: 162-5.

Core, N., B. Charrowc, A. McCormick, C. Vola, L. Fasano, M.P. Scon, and S. Kerridge. 1997. Transcriptional regulation of the Drosophila homeotic gene teashin by the homeodomain protein Fushi tarazu. :khDa, 68: 157-72.

Coulter. DE.. €.A. Swaykus, M.A. Beran-Kwhn, D. Goldberg, E. Wieschaus. and P. Schedl. 1990. Molecular analysis of odd-skipped, a zinc finger encoding segmentation gene with a novel pair-rule expression pattern. Embo J9: 3795-801.

Cousens, DA., R, Greaves. CA. Goding, and P, O'Hare. 1989. The C-terminal 79 amino acids of the herpes simplex virus regulatory protein, Vmw65, efficiently activate transcription in yeast and mammalian cells in chimeric DNA-binding proteins. Embo J8: 2337-42.

Couso, J.P., M. Bate. and A. Martinez-Ais. 1993. A wingless-dependent polar coordinate system in Drosophila imagina1 discs [see comments]. Science 259: 184-9.

Couso, J.P. and M. Gonzalez-Gaitan. 1993. Embryonic limb development in Drosophila. Trends Gener 9: 371-3.

Crick, F.H. and P.A. Lawrence. 1975. Compartments and polyc1ones in insect development. Science 189: 340-7.

Cseqesi, P. and E.N. Olson. 1991. Myogenin induces the myocyte-specific enhancer binding factor MEF-2 independently of other muscle-specific gne products. hlCeII Bi01 11: 4854-62.

Cume, R.A. and R.G. Roeder. 1989. Identification of an octamer-binding site in the mouse kappa light- chah immunoglobuiin enhancer. Mol Cell BioZ9: 4239-47.

Dahanukar, A. and R.P. Wharton. 1996, The Nanos gradient in Drosophila embryos is gencrated by translational regulation. Genes Dev 10: 16 10-20.

Dalton, D., R. Chadwick, and W. McGinnis. 1989. Expression and embryonic function of ernpty spiracles: a Drosophila homeo box gene with two patteming functions on the anterior- postenor axis of the embryo. Genes Dev 3: 1940-56. Darimont, B.D., R.L. Wagner, J.W. Aprïletti, M.R. Stallcup, P.J. Kushner, JD. Baxter, R.J. Fletterick, and K.R. Yamamoto. 1998. Structure and specificity of nuclear receptor- coactivator interactions. Genes Dev 12: 3343-56.

Damell, S.E., Jr., I.M. Kerr, and GR. Stark. 1994. Jak-STAT pathways and transcriptional activation in response to IFNs and other e.mceilular signaling proteins. Science 264: 1415-21-

Davidson, I., C. Fromental, P. Augereau, A. Wildeman, M. Zenke, and P. Chambon. 1986. Cell-type specific protein binding to the enhancer of sirnian virus 40 in nuclear extracts. Nature 323: 544-8-

Davis, 1. and D. ish-Horowitz, 199 1. Apical localization of pair-mle transcnpts requires 3' sequences and lirnits pmtein diffusion in the Drosophila blastoderm embiyo. Ce11 67: 927-40.

Dearolf, C.R., J. Topol, and C.S. Parker. 1989a. The caudal gene product is a direct activator of tiishi mutranscription during Drosophila embryogenesis. Nature 341: 340-3.

Dearolf, C.R., J. Topol, and C.S. Parker. I989b. Transcriptional control of Drosophila füshi tarazu zebra stxipe expression. Genes Dev 3: 384-98.

Dearolf, C.R., J. Topol, and C.S. Parker. 1990. Transcriptional regulation of the Drosophila segmentation gene hshi tarazu (fh)[published erratum appears in Bioessays 1990 Apr; 12(4):198]. Bioessays 12: 109-13.

Deguchi, Y., J.F. Moroney, G.L. Wilson, C.H. Fox, H.S. Winter, and J.H. Kehrl. 1991. Cloning of a human homeobox gene that resembles a diverged Drosophila homeobox gene and is expressed in activated lymphocytes. New Bi01 3: 353-63.

Delorenzi, M. and M. Bienz. 1990. Expression of Abdominal-B homeoproteins in Drosophita embryos. Development 108: 323-9.

DeRisi, J.L., V-R. [yer, and P.O. Brown. 1997. Evploring the metabolic and genetic control of gene expression on a genomic scale. Science 278: 680-6.

Desplan, C., J. Theis, and P.H. O'Farrell. 1985. The Drosophila developmental gene, engrailed, encodes a sequence- specific DNA binding activity. Nature 318: 630-5.

Desplan, C.. J. Theis, and P.H. O'Farrell. 1988, The sequence specificity of homeodomain-DNA interaction. Cell54: 1081-90.

Dessain, S.. C.T. Gross, M.A. Kuziora, and W. McGinnis. 1992. Antp-type homeodomains have distinct DNA binding specificities that correlate with their differeot ngulatory fünctions in embryos. Embo J 11: 991-1002.

Di Nocera, P.P. and I.B. Dawid. 1983. Transient expression of genes introduced into cultured cells of Drosophila. Proc Nail Acad Sci USA 80: 7095-8.

Diaz-Benjumea, F.J. and S.M. Cohen. 1993. Interaction benveen dorsal and ventral cells in the imaginai disc directs wing development in Drosophila. Ce11 75: 741-52.

DiNardo, S., J. Heemskerk, S. Dougan, and P.H. O'Farrell. 1994. The making of a maggot: patterning the Drosophila embryonic epidermis. Curr Opin Gmer Dev 4: 529-33.

DiNado, S., J.M. Kuner, J. Theis, and P.H. O'Farrell, 1985. Development of embryonic pattern in D. melanogaster as revealed by accumulation of the nuclear engrailed protein. Ce11 43: 59-69. DiNardo, S. and P.H. O'Farrell. 1987. Establishment and refinement of segmental pattem in the Drosophila embryo: spatial control of engrailed expression by pair-Pmle genes. Genes Dev 1: 1212-25.

DiNardo, S., E. Sher, J. Heemskerk-longens, J.A. Kassis, and P.H. O'FarreIl. 1988. Two-tiered regulation of spatially pattemed engrailed gene expression during Drosophila embryogenesis. Nature 332: 601-9.

Doe, C.Q., Y. Hiromi, W.J. Gehring, and C.S. Goodman. 1988. Expression and function of the segmentation gene hshi tarani during Drosophila neurogenesis. Science 239: 170-5.

Dolan, J.W. and S. Fields. 1991. Cell-type-specific transcription in yeast Biochim Biophys Acta 1088: 155-69.

Dong, J.. L.H. Hung, R, Strome, and H.M. Krause. 1998. A phosphorylation site in the ftz homeodomain is required for activity. Embo J 17: 2308-18.

Dong, J. and H.M. Krause. 1999. Tissue-specifîc requirements for a phosphorylation site in the Fushi tarazu homeodomain. Dev Genes Evol209: 427-3 1.

Doyle, H.J. and J.M. Bishop. 1993. Torso. a receptor tyrosine kinase required for embryonic panem formation, shares substrates with the sevenless and EGF-R pathways in Drosophila. Genes Dev 7: 633-46.

Dranginis, A.M. 1990. Binding of yeast al and alpha 2 as a hetemdimer to the operator DNA of a haploid-specific gene. Nature 317: 682-5.

Drees, B., Z. Ali, W.C. Soeller, KG. Coleman, S.J. Poole, and T. Komberg. 1987. The transcription unit of the Drosophila engrailed locus: an unusually small portion of a 70,000 bp gene. Embo J6: 7803-9.

Driever, W. 1993. Matemal control of anterior development in the Drosophila embryo. In The Deveiopmenr of Drosophila melanogasrer (ed. M. Bate and A. Martinez-Arias). Cold Spting Harbor Laboratory Press.

Driever. W. and C. Nusslein-Volhard 1988a The bicoid protein detemines position in the Drosophila embryo in a concentration-dependent manner. Ceil 54: 95-104.

Driever. W. and C. Nusslein-Volhard. 1988b. A gradient of bicoid protein in Drosophila embryos. Cell54: 83-93.

Driever, W. and C. Nusslein-Volhard. 1989. The bicoid protein is a positive replator of hunchback transcription in the early Drosophila embryo. Nature 337: 138-43.

Driever, W., G, Thoma, and C. Nusslein-Volhard. 1989. Determination of spatial domains of zvgotic nene expression in the Drosophila ernbryo by the affiniq of binding sites for the bicoid b morphogen. Nature 340: 363-7.

Dubnau, J. and G. Struhl. 1996. RNA recognition and translational regulation by a homeodomain protein [see comments] [published erratum appears in Nature 1997 Aug 14388(6643):697]. Nature 379: 694-9.

Duboule, D. 1992. The vertebrate limb: a mode1 system to study the HoflOM gene network during development and evolution. Bioessays 14: 375-84.

Duboule, D. and P. Dolle. 1989. The structura[ and functional organization of the murine HOX gene family resernbles that of Drosophiia homeotic senes- Embo J8: 1397-505. Duncan, 1. 1986. Control of bithorax compleiu fünctions by the segmentation gene fushi tarazu of D. melanogaster. CeIl 47: 297-309.

Durand, B., M. Saunders, C. Gaudon, B. Roy, R. Losson, and P. Chambon. 1994. Activation Function 1 (AF-2) of retinoic acid receptor and 94s retinoic acid receptor: presence of a conserved autonomous constitutive activating domain and influence of the nature of the response elernent on AF-2 activity. Embo J13: 5370-82.

Eaton, S. and T.B. Kornberg. 1990. Repression of ci-D in posterior compartments of Drosophila by engrailed. Genes Dev 4: 1068-77.

Edgar, B.A., G.M. Odell, and G. Schubiger. 1987. Cytoarchitecture and the patteming of tus hi tarazu expression in the Drosophila blastoderm. Genes Dev 1: 1326-37.

Edgar, B.A., M.F. Weir, G. Schubiger, and T. Kornberg. 1986. Repression and turnover pattern fushi tarazu RNA in the early Drosophila ernbryo. CeIl 47: 747-54.

Edmondson, D.G., T.C. Cheng, P. Csejesi, T. Chakraborty, and E.N. Oison. 1992. Analysis of the rnyogenin promoter reveals an indirect pathway for positive autoregulation mediated by the muscle-specific enhancer factor MEF-2. Mol Ce11 Bi01 12: 3665-77.

Ekker, SC.. D.G. Jackson, D.P. von Kessler, B.I. Sun, K.E. Young, and PA. Beachy. 1994. The degree of variation in DNA sequence recognition arnong four Drosophila homeotic proteins. Embo J13: 3551-60.

Ekker. SC, D.F. von Kessler, and PA. Beachy. 1992. Differential DNA sequence recognition is a determinant of specificity in homeotic gene action. Embo J 11: -1059-71.

Ekker, S.C., K.E. Young, D.P. von Kessler, and P.A. Beachy. 1991. Optimal DNA sequence recognition by the Ultrabithom. homeodomain of Drosophila. Embo J 10: 1179-86.

Eldon, E.D. and V. Pirrotta. 1991. Interactions of the Drosophila gap gene giant with materna1 and zygotic pattern-forming genes. Developmenr 111: 367-78.

England, B.P.. A. Admon. and R. Tjian. 1992. Cloning of Dmsophila transcription factor Adf-1 reveals hornology to Myb oncoproteins. Proc Narl Acad Sci USA 89: 683-7.

England. B.P.. U. Heberlein. and R. Tjian. 1990. Puriîïed Drosophila transcription factor. Adh distal factor-1 (Adf-1), binds to sites in several Drosophila promoters and activates transcription. J Bi01 Chem 265: 5086-94.

Fainsod. A.? L.D. Bogarad, T. Ruusala, M. Lubin, DM. Crothers, and F.H. Ruddle. 1986. The homeo domain of a murine protein binds 5' to its own homeo box. Proc Nail Acad Sci USA 83: 9532-6.

Fajas, L.. J.C. Fmchart. and J. Auwerx 1998. Transcriptional conml of adipogenesis. Curr Opin Ce11 Biot 10: 165-73.

FaIkner. F.G. and H.G. Zachau. 1984. Correct transcription of an irnmunoglobulin kappa gene requires an upstream fragment containhg conserved sequence elements. Xature 310: 714.

Farkas, G., J. Gausz, M. Galloni, G. Reuter, H. Gyurkovics, and F. Karch. 1994. The Trithorax-Iike gene encodes the Drosophila GAGA factor. Nature 371: 806-8.

Fasano, L, L, Rode- N. Cor% E. Alexandre, C, Vola, B. Jacq, and S. Kemdge. 1991. The gene teashirt is required for the development of Drosophila embryonic trunk segments and encodes a protein with widely spaced zinc fmger motifs. Cell6J: 63-79. Fields, S. and 0. Song. 1989. A novel genetic system to detect protein-protein interactions. Nature 340: 245-6.

Fields, S. and R. Stemglanz. 1994. The NO-hybrid system: an assay for protein-protein interactions. Trends Genet 10: 286-92.

Finkelstein, R. and N. Pemmon. 1990. The orthodenticle gene is regulated by bicoid and torso and specifies Drosophila head development [se comments]. Nature 316: 485-8.

Fisher, A.L., S. Ohsako, and M. Caudy. 1996. The WRPW motif of the haiq-related basic helk- loop-helix repressor proteins acts as a bamino-acid transcription repression and protein- protein interaction domain. Mol Celi Biol16: 2670-7.

Fitzpatrick, V.D. and C J. 1ngIes. 1989. The Drosophila fushi tarazu polypeptide is a DNA-binding transcriptional activator in yeast cells. Nature 337: 666-8.

Fitzpamck, V.D.. A. Percival-Smith, CJ. Ingles. and H.M. Krause. 1992. Homeodomain-independent activity of the fushi truani polypeptide in Drosophila embryos. Nature 356: 610-2.

Fjose, A.. W.J. McGinnis, and W.J. Gehring. 1985. Isolation of a hornoeo box-containing gene from the engrailed region of Drosophila and the spatial distribution of its transcripts. Nature 313: 284-9.

Flegel. W.A.. A.W. Singson. J.S. MargoIis. A.G. Bang, J.W. Posakony, and C. Mune. 1993. Dpbx. a new homeobox gene closely related to the human proto-oncogene pbxl molecular structure and developrnental expression- A4ech Dev 41: 155-6 1.

Florence, B., A. Guichet, A. Ephrussi, and A, Laughon. 1997. Ftz-FI is a cofactor in Fa activation of the Drosophila engrailed gene. Developmenr 121: 839-47.

Florence, B.. R. Handrow, and A. Laughon. 1991. DNA-binding specificity of the fushi tarazu horneodornain. Mol Ceil Bi01 11: 36 13-23.

Fortini, ME, Z.C. Lai. and G.M. Rubin. 1991. The Drosophila zfh-1 and zfh-7 genes encode novel proteins containing both zinc-finger and homeodomain motifs. Mech Dev 34: 113-12,

Francis-Lang, H.. 1. Davis. and D. Ish-Horowitz. 1996. Asyrnrnetric localization of Drosophila pair- mIe vanscripts hmdisplaced nuclei: evidence for directional nuclear export. Embo J 15: 640-9.

Frasch, M.. T. Hoey. C. Rushlow, H, Doyle, and M, Levine. 1987. Characterimion and localization of the even-skipped protein of Drosophila. Embo J6: 749-59.

Frasch, M. and M. Levine. 1987. Complementary patterns of even-skipped and fushi tatazu expression involve theu differential regdation by a common set of segmentation genes in Drosophila. Genes Dev 1: 98 1-95.

Frasch, M.. R. Warrior, J. Tugwood, and M. Levine. 1988. Molecular anatysis of even-skipped mutants in Drosophila development. Genes Dev 2: 1824-38.

Frayne, E.G. and T. Sato, 1991. The Ultrabithom~gene of Dmsophila and the specification of abdominal histoblasts. Dm BioI 146: 26577.

Freyd, Ci.. S.K. Kim. and H.R Horvin. 1990. Novel cysteine-nch motif and homeodomain in the product of the Caenorhabdiris eIegans ce11 Iineage gene Lin-1 1. Nature 344: 876-9. Frigerio, G., M. Buk, D. Bopp, S. Baurngartner, and M. Noll. 1986. Sûucture of the segmentation gene paired and the Drosophila PRD gene set as part of a gene network. CellJ7: 735-46.

Frohnho fer, H.G. and C. Nusslein-Volhard. 1986. Organization of antenor pattern in the Drosophila embryo by the matemal gene bicoid. Narure 324: 110-115.

Frohnhofer, H.G. and C. Nusslein-Volhard. 1987. Materna1 genes required for the anterior localization of bicoid activity in the embryo of Drosophila. Gmes Dev 1: 880-890.

Fujioka, M., J.B. Jaynes, and T. Goto. 1995. Early even-skipped stripes rict as morphogenetic mdients at the single ceIl level to establish engrailed expression. Development 121: 4371-82. C

Fumkubo-Tokunaga, K., S. Flister, and W.J. Gehring. 1993. Functional specificity of the Antennapedia homeodomain. Proc ~VatlAcad Sci USA 90: 63604.

Furukubo-Tokunaga, K.. M. Muller, M. Affolter, L. Pick, U. Kloter, and W.J. Gehring. 1992. [n vivo analysis of the helk-tum-helix motif of the fushi tamu homeo domain of Drosophila melanogaster. Genes Dev 6: 1082-96.

Garber, R.L., A. Kuroiwa, and WJ. Gehring. 1983. Genomic and cDNA clones of the homeotic locus Antennapedia in Drosophila. Embo J2: 2027-36.

Garcia-Bellido. A. 1975. Genetic conuol of wing disc development in Drosophila. CIBA Foundarion Symp. 29: 161-183.

Garcia-Bellido, A. 1977. Homoeotic and atavic mutations in insects. dmer. Zool. 17: 613-629.

Garcia-Bellido. A.? PA. Lawrence. and G. Morata. 1979. Compamnents in animal development. Sci. Amer. 241: 102-1 10.

Garcia-Bellido. A., P. Ripoll, and G. Morata. 1973. Developmental compartmentalisation of the wing disk of Drosophila. Nat New Biol 245: 251-3.

Gad, U. and H. Jackle. 1987. Pole region-dependent repression of the Drosophila gap gene Knippel by matemal gene products. Cell51: 549-55.

Gaul, U. and H. Jackle. 1989. Analysis of matemal effect mutant combinations elucidates regulation and function of the overlap of hunchback and Kruppel gene expression in the Drosophila blastodem embryo. Developmenr 107: 65 1-62.

Gaul. U. and H. Jackle. 1991. Role of gap genes in early Drosophila development. ddv. Genet. 27: 339-772.

Gehring, WJ.. M. Affolter, and T. Burglin. 1994a. Homeodomain proteins. Annu Rev Biochem 63: 487-516,

Gehring, WJ., Y.Q. Qian, M. Billetei K. Furukubo-Tokunaga, AS. Schier, D. Resendez-Perez, M. Affolter. G. Otting and K. Wuthrich. 1994b. Homeodomain-DNA recognition. Cell78: 21 1- 23.

Gellon, G., K.W. Harding, N. McGinnis, M.M. Martin. and W. McGinnis. 1997. A genetic screen for modifiers of Defonned homeotic function identifies novel genes required for head development Developmenr 124: 3311-3 1.

Gergen, J-P- and BA. Butler. 1988. Isolation of the Drosophiia segmentation gene runt and analysis of its expression during embryogenesis. Genes Dm 2: 1 179-93. Gerster, T., C.G. Balmaceda, and R.G. Roeder. 1990. The cell type-specific octarner transcription factor OTF-2 has two domains required for the activation of transcription. Embo J9: 1635- 43.

Gertler, F.B., R.L. Bennett, MJ. Clark, and F.M. Hoffmann. 1989, Drosophila ab1 tyrosine kinase in embryonic CNS auons: a mle in auonogenesis is revealed through dosage-sensitive interactions with disabled. CeII 58 103-13.

Gibson. G., A. Schier, P. LeMotte, and WJ. Gehring. 1990. The specificities of Sex combs reduced and Antennapedia are defined by a distinct portion of each protein that includes the homeodomain. Ce11 62: 1087-103,

Gilmour, D.S. and J.T. Lis. 1986. RNA polymerase II interacts with the promoter region of the noninduced hsp70 gene in Drosophila meIanogmter cells. Mol Ce11 Bi01 6: 3984-9.

Goding, C.R. and P. O'Hare. 1989. Herpes simplex virus Vmw65-octamer binding protein interaction: a paradigm for cornbinatoriat control of transcription. Virology 173: 363-7.

Gonzalez-Reyes, A., A. Macias, and G. Morata. 1993. Autocatalysis and phenotypic expression of Drosophila homeotic gene Deformed: its dependence on polarity and homeotic gene tirnction. Developmenr 116: 1059-68.

Gonzalez-Reyes, A. and G. Morata. 1990. The developmental effect of overexpressing a Ubx product in Drosophila embryos is dependent on its interactions with other homeotic products. CeIl 61: 515-2.

Gonzalez-Reyes, A.. N. Urquia, W.J. Gehring, G. Struhl, and G. Morata. 1990. Are cross-regulatory interactions benveen homoeotic genes functionally significant? Nature 344: 78-80.

Gossett, L.A.. D.J. Kelvin, EA. Sternberg and E.N. Olson. 1989, A new myocyte-specific enhancer- binding factor that recognizes a conse~edeiement associated with multiple muscle-specific genes. Mol Ce11 Bi01 9: 5012-33.

Goto. T.. P. Macdonald, and T. Maniatis, 1989. Early and Iate periodic patterns of even skipped expression are controlled by distinct regdatory eIernents that respond to different spatial cues. Ce11 57: 4 13-32-

Gould, A.P., J.J. Brookman, DL Stmtt, and R.A. White. 1990. Tarnets of homeotic gene control in Drosophila [pubiished erratum appears in Nature 1990 ~ec-6:348(6301):560c Xature 348: 308-12.

Gould, A.P. and RA. White. 1992. Connectin, a target of homeotic gene control in Drosophila. Drvelopment 116: 1 163-74.

Goutte. C. and A.D. Johnson. 1988. al protein ahers the DNA binding specificity of alpha ? repressor. Ce11 52: 875-82.

Goutte, C. and A.D. Johnson. 1993. Yeast al and alpha 2 homeodomain proteins fonn a DNA- binding activity with propenies distinct from those of either protein, J Mol Bi01 233: 359- 7 1.

Graba, Y., D. Aragnol, P. Laurenti, V. Garzino, D. Chmot, H. Berenger, and J. Pradel. 1992. Homeotic control in Drosophila; the scabrous gene is an in vivo target of Ultrabithorax proteins. Embo J 11: 3375-84.

Graham, A., N, PapaIopulu, and R Knimtauf. 1989. The murine and Drosophila homeobox gene compIexes have common features of organization and expression. Ce11 57: 367-78. Greaves, R. and P. O'Hare. 1989. Separation of requirements for pmtein-DNA cornplex assembly from those for functional activity in the herpes simplex virus regulatory protein Vmw65. J Viroi63: 1641-50.

Grieder, N.C., T. Marty, H.D. Ryoo, RS. Mann, and M. Affolter. 1997. Synergistic activation of a Drosophila enhancer by HOMIEXD and DPP signaling. Embo J 16: 7402-IO.

Gross, C.T. and W. McGinnis. 1996. DEAF-1, a novel protein that binds an essential region in a Defonned response element. Embo J 15: 196 1-70.

Grossniklaus, U., R.K. Pearson, and W.J. Gehring. 1992. The Drosophila sloppy paired locus encodes two proteins involved in segmentation that show homolog to rnammalian transcription factors. Genes Dev 6: 1030-5 1.

Gubb, D. 1983'. Domains, compartments and determinative switches in Drosophila developrnent. BioEssays 2: 27-3 1.

Guichet, A., J.W. Copeland, M. Erdelyi, D. Hlousek, P. Zavorszky. J. Ho, S. Brown, A. Percival- Smith, H.M. Krause, and A. Ephrussi. 1997. The nuclear receptor homologue FU-FI and the horneodornain protein Fa are mutually dependent cofacton. Narure 385: 548-52.

Gutjahr. T., E. Frei, and M. Noll. 1993. Complex regulation of early paired expression: initial activation by gap genes and pattern modulation by pair-rule genes. Developmenr 117: 609- 23.

Gutjahr, T., C.E. Vanario-Alonso, L. Pick, and M. Noll. 1994. Multiple regulatory elernents direct the cornplex expression pattern of the Drosophila segmentation gene paired. Mech Dev 18: 119-28.

Hacker, U., U. Grossniklaus, W.J. Gehring, and H. Jackle. 1992, Developmentally regulated Drosophila gene farnily encoding the fork head domain. Proc NatI Acad Sci USA 89: 8754- 8 *

Hader, T., A. La Rosee, U. Ziebold. M. Busch. H. Taubert. H. Jackle. and R. Rivera-Pornar. 1998. Activation of posterior pair-rule stripe expression in response to materna1 caudal and zygotic knirps activities. Mech Dev 71: 177-86.

Hafen, E.. A. Kuroiwa and W.J. Gehring. 1984a. Spatial distribution of transcripts fiom the segmentation gene fishi tarazu during Drosophila embryonic development. Cd37: 833-4 1.

Hafen, E., M. Levine, and W.J. Gehring. I984b. Regulation of Antennapedia transcript distribution by the bithom complex in Drosophila. Nature 307: 287-9.

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

Harnpsey, M. 1998. Molecular genetics of the RNA polymerase II generd transcriptional rnachinery. Microbiol Moi Bi01 Rev 62: 365-503.

Han, K., M.S. Levine, and J.L. Manley. 1989. Synergistic activation and repression of transcription by Drosophila homeobox proteins. Cell56: 573-83.

Han. K. and J.L. Manley. 1993a. Functional domains of the Drosophila Engraiied protein. Embo J 12: 2723-33.

Han, K. and JL. Manley. 1993b. Transcriptional repression by the DrosophiIa even-skipped protein: definition of a minimal repression domain. Genes Dm 7: 49 1-503. Han, W., Y. Yu, N. Altan, and L. Pick. 1993. Multiple proteins interact with the fushi tarazu proximal enhancer. Mol CeIl Biol13: 5549-59. Han, W., Y. Yu, K. Su, R.A. Kohanski, and L. Pick. 1998. A binding site for multiple transcriptional activators in the fushi tarazu proximal enhancer is essential for gene expression in vivo. Mol Ce11 Biol18: 3384-94.

Hanes, S.D. and R. Brent, 1989. DNA specificity of the bicoid activator protein is determined by homeodomain recognition heliv residue 9. Ce11 57: 127583.

Hanes, S.D. and R. Brent. 1991. A genetic mode1 for interaction of the homeodomain recognition heli. with DNA, Science 251: 426-30,

Harding, K., T. Hoey, R Wanior, and M. Levine. 1989. Autoregulatory and gap gene response elements of the even-skipped promoter of Drosophila. Embo J 8: 1205- 12. Harding, K. and M. Levine. 1988. Gap genes define the limits of antennapedia and bithorax gene expression during early development in Drosophila. Embo J7: 205-14.

Harding, K., C. Rushlow, H.J. Doyle, T. Hoey, and M. Levine. 1986. Cross-regulatory interactions among pair-mle genes in Drosophila. Science 233: 953-9.

Harding, K., C. Wedeen, W. McGinnis, and M. Levine. 1985, Spatially regulated expression of homeotic genes in Drosophila. Science 229: 123642.

Harding, K.W..G. Gellon, N. McGimis, and W. McGinnis. 1995. A screen for modifiers of Deformed function in Drosophila. Generics 110: 1339-52.

Harrison, D.A.. P.E. McCoon, R. Binari, M. Gilman. and N. Pemmon. 1998. Drosophila unpaired encodes a secreted protein that activates the JAK signaling pathway. Genes Dev 12: 3252-63.

Hamson. S.D. and A.A. Travers. 1988. Identification of the binding sites for potential regulatory proteins in the upstream enhancer element of the Drosophila fushi rarazu gene [published erratum appears in Nucleic Acids Res 1989 Jan 25; 17(2):830]. ~Vucleicdcids Res 16: 1 1403- 16.

Harrison, S.D. and A.A. Travers. 1990. ïhe tramtrack gene encodes a Drosophila finger protein that interacts with the Az transcriptional regulatory region and shows a noveI embryonic expression pattern, Embo J 9: 207- 16.

Hartmann. C.. H. Taubert. H. Jackle, and M.J. Pankratz. 1994. A two-step mode of sniue formation in the Drosophila bIastodem requires interactions among primary pair rule genei- Mech Dev 15: 3-13.

Heberlein, LI., B. England, and R. Tjian. 1985. Characterimion of Drosophila transcription factors that activate the tandem promoters of the alcohol dehydrogenase gene. Cell11: 955-77.

Heberlein, U. and R. Tjian. 1988. Temporal pattern of alcohol dehydrogenase gene transcription reproduced by Drosophila stage-specific embryonic extracts. Nature 331: 410-5.

Heemskerk, J., S. DiNado, R, Kostriken, and P.H. O'FarrelI. 1991. Multiple modes of engrailed regdation m the progression towards cell fate determination. Nart- 352: 404-10.

Heery, D.M., E. Kalkhoven, S. Hoare, and MG. Parker. 1997. A signature motif in transcriptional CO-activatorsmediates binding to nuclear receptors [see comments]. Nature 387: 733-6. Heina, N., M. Zernik, and R.G. Roeder. 1981. The structure of the human histone genes: clustered but not tandemly repeated. Cell24: 661-8.

Hem, W., RA. Sturm, R.G. Clerc, L.M. Corcoran, D. Baltimore, PA. Sharp, HA. Ingraham, M.G. Rosenfeld. M. Finney, G. Ruvkun, and et al. 1988. The POU domain: a large conserved region in the mammalian pit- 1, oct- 1, oct-2, and Caenorhabditis elegans unc-86 gene products. Genes Dev 2: 15 13-6.

Herskowitz, 1, 1989. A regulatory hierarchy for ceII specialization in yeast. Nature 342: 749-57.

Hiromi, Y. and W.J. Gehring. 1987. Regdation and hnction of the Drosophila segmentation gene fushi tarazu. Ce11 50: 963-74.

Hiromi, Y., A. Kuroiwa, and W.J. Gehring. 1985, Conml elements of the Drosophila segmentation gene hshi tarazu. Cell13: 603-L3.

Hoch, M., N. Gerwin, H. Taubert, and H. Jackle. 1992. Competition for overlapping sites in the regulatory region of the Drosophila gene Kruppel. Science 256: 94-7.

Hoch, M., E. Seifert, and H. Jackle. 199 1. Gene expression mediated by cis-acting sequences of the Kruppel gene in response to the Drosophila morphogens bicoid and hunchback. Embo J 10: 2267-78.

Hoey, T. and M. Levine. 1988. Divergent homeo box proteins recognize similar DNA sequences in Drosophila. Nature 332: 858-6 1.

Hoey, T., R. Wûrrior, J. Manak, and M. Levine. 1988. DNA-binding activities of the Drosophila melanogaster even-skipped protein are mediated by its homeo domain and influenced by protein context. Mol CdBi01 8: 4598-607.

Holstege, F.C.. E.G. lennings. JJ, Wyrick. T.I. Lee, C.J. Hengartner. M.R. Green. T.R. Golub, ES. Lander, and R.A. Young, 1998. Dissecting the regulatory circuitry of a eukaryotic genome. Ce11 95: 7 17-28.

Hooper. J.E. and M.P. Scott. [989. The Drosophila patched gene encodes a putative membrane protein required for segmental patteming. Cd59: 75 1-65.

Hooper. J.E. and M.P. Scott. 1992. The Molecuiar Genetic Basis of Positional Information in insect Segnents, in Early Embryanic Development of Animals (ed. W. Hennig). Spnnger-Verlag, Berlin Heidelberg.

Hooper, K.L., S.M. Parkhurçt, and D. 1sh-Horowitz. 1989. Spatial control of hairy protein expression during ernbryogenesis, Developmenr 107: 489-504.

Hou, X.S.. M.B. Melnick, and N. Perrimon- 1996. Mareite acts downsrream of the Drosophila HOPIJAK kinase and encodes a protein sirnilar to the mamrnalian STATs [published erratum appears in Cell 1996 Apr 19;85(2):following 2901. Cell84: 41 1-9.

Howard, K. 1988. The generation of periodic pattern during early Drosophila ernbryogenesis. Developmenf (suppl.): 35-50.

Howard, K. and P. [ngham. 1986. Regulatory interactions between the segmentation genes hshi taraz& haky, and engrailed in the DrosophiIa blastoderm. Cell U: 949-57.

Howard, K., P. Ingham, and C. Rushlow. 1988, Region-specific alleles of the Drosophila segmentation gene hairy. Genes Dev 2: 1037-46. Howard, K.R. and G. Struhl. 1990. Decoding positional information: regdation of the pair-rule gene hairy. Development 110: 1223-3 1.

Hughes, S. and H.M. Krause. 200 1. Establishment and maintenance of parasegmental compamnents. Development in press.

Huibregtse, J.M. and D.R. Engelke. 1989. Genomic footprinting of a yeast tRNA gene reveals stable complexes over the 5'-flanking region. Mol Ce11 Bi01 9: 3744-52-

Hulskamp, M., C. Pfeifle, and D. Tautz. 1990. A morphogenetic gradient of hunchback protein organites the expcession of the gap genes Knippel and knirps in the early Drosophila embryo. Nature 346: 577-80.

Hulskamp, M., C. Schroder, C. Pfeifle, H. Jackle. and D. Tautz. 1989. Posterior segmentation of the Drosophila embryo in the absence of a materna1 posterior organizer gene. Nature 338: 679- 32.

Hunger, S.P.. N. Gaiili, A.J. Carroll, W.M. Crist, M.P. Link, and M.L. Cleary. 1991. The t(1;19)(q23;p13) results in consistent fusion of E2A and PBXl coding sequences in acute lymphoblastic leukemias. Blood 77: 687-93.

Hyduk, D. and A, Percival-Smith. 1996. Genetic characterizarion of the homeodomain-independent activity of the Drosophila fushi tarazu gene product. Generics 142: 481-92.

Ikeda, K., D.J. Stege- A. Eberharte. and J.L. Workman. 1999. Activation domain-specific and genenl transcription stimulation by native histone acetyltransferase complexes. Mol Ce11 Bi01 19: 855-63.

Ingham. P. and J.P. Gergen. 1988. Interactions between the pair-nile genes runt. hairy, even-skipped andfurhi tara and the establishment of periodic pattern in the Drosophila embryo. Developmenr (suppl.) 104: 5 1-60.

Ingham, P.W. 1985. A clona1 analysis of the requirement for the nithorav gene in the diversification of segments in Drosophila. J Embryol Eip ~Morphol89:339-65.

Ingham, P.W. 1988. The molecular genetics of embryonic pattern formation in Drosophila [published erratum appears in Nature 1988 Oct 20:335(6 192):744]. Nature 335: 25-34.

Ingham, P.W. 1991, Segment polarity genes and cell patterning within the Drosophila body segment [published erratum appears in Cun Opin Genet Dev 199 1 Oct; 1(3):417]. Curr Opin Grner Dev 1: 261-7.

Ingham. P.W., N.E. Baker, and A. Martinez-Arias, 1988. Regulation of segment polarity genes in the Drosophila blastoderm by fushi tarani and even skipped. Nature 331: 73-5.

Ingham, P.W., K. Howard, and D. Ish-Horowicz 1985. Transcription pattern of the Drosophila segmentation gene hairy. Nature 318: 443-445.

Ingham, P.W., D. Ish-Horowitz, and K.R. Howard. 1986. Correlative changes in homeotic and segmentation gene expression in Kiuppel embryos of Drosophila. Embo J5: 1527-1537.

ingham, P.W. and A. Martinez Arias. 1992. Boundaries and fields in early embryos. Cell68: 221-35.

Ingham, P.W. and A. Martinez-Arias. 1986. The correct activation of Antennapedia and bithorax cornplex genes requires the fushi tarani gene. Namre 324: 592-7. ingham, P.W., A.M. Taylor, and Y. Nakano. 1991. Role of the Drosophila patched gene in positional signalling. Nature 353: [84-7.

Ingraham, H.A., S.E. Flynn, J.W. Voss, V.R. Alber&M.S. Kapiloff, L. Wilson, and M.G. Rosenfeld. 1990. The POU-specific domain of Pit-1 is essential for sequence-specific, high afinity DNA binding and DNA-dependent Pit-1-Pit-1 interactions. Ce11 61: 1021-33. iristi, V.F., A. Martinez-Arias, and M. Akam, 1989. Spatial regulation of the Antennapedia and Ultrabithorav homeotic genes during Drosophila early development. Embo J8: 1527-37.

Ish-Horowitz, D. and H.G. Gyurkovics. 1988. Ectopic segmentation gene expression and metameRc regulation in Drosophila. Development (suppl.): 67-73.

Ish-Horowitz, D.. KR. Howard, S.M. Pinchin, and P.W. Ingham. 1985. Molecular and genetic analysis of the hairy locus in Drosophila. Cold Spring Harb Symp Quanr Bi01 50: 135-44. ish-Horowitz, D. and S.M. Pinchin. 1987. Pattern abnormalities induced by ectopic expression of the Drosophila gene hairy are associated with repression of f& transcription. Cell51: 405-15.

Ish-Horowitz, D., S.M. Pinchin, P.W. tngham, and H.G. Gyurkovics. 1989. Autocatalytic Az activation and metarneric instability induced by ectopic fa expression. CeII 57: 123-32.

Jack, T., M. Regulski. and W. McGinnis. 1988. Pair-mle sementation genes regulate the expression of the homoeotic selector gene Deformed. Genes Dev 2: 645-65 1.

Jackle, H., D. Tautz, R. Schuh, E. Seifert, and R. Lehmann. 1986. Cross-regulatory interactions arnong the gap genes of Drosophila. Nature 321: 668-670.

Jacob, F. and J. Monod. 196 1. Genetic regulatory mechanisms in the synthesis of proteins. J. Mol. Biol. 3: 3 18-356.

Jaynes. J.B. and P.H. OfFrneIl. 1988. Activation and repression of transcription by homoeodomain- containing pmteins that bind a common site. Nature 336: 744-9.

Jaynes, J.B. and P.H. O'Farrell. 1991. Active repression of transcription by the engrailed homeodomain protein. Enibo J 10: 1427-33.

Jiang, J., T. Hoey. and M. Levine. 1991. Autoregulation of a segmentation gene in Drosophila: combinatorial interaction of the even-skipped homeo box protein with a distal enhancer eIement. Genes Dev 5: 165-77.

Jimenez, G., 2. Paroush, and D. ish-Horowitz. 1997. Groucho acts as a corepressor for a subset of negative regulators, including Hairy and Engrailed. Genes Dev 11: 3072-82.

Jimenez, G., S.M. Pinchin, and D. Ish-Horowitz. 1996. In vivo interactions of the Drosophila Hairy and Runt transcriptional tepressors with target promoters. Embo J 15: 7088-98.

John, A., S.T. Smith, and J.B. Jaynes. 1995. Inserting the Fa homeodomain into engrailed creates a dominant transcriptional repressor that specifically turns off Ftz target genes in vivo, Development 12 1: 180 1-13.

Johnson. A.D. 1992. A combinatorial regutatoy circuit in budding yeast. In Trmcriptional regulation (ed. S.L. McKnight and K. Yamamoto), pp. 975-1006. Cold Soring Harbor Laboratory Press, Cold Soring Harbor.

johnson, A.D. and 1. Herskowitz. 1985. A repressor (MAT alpha 2 Product) and its opentor control expression of a set of cell type specific genes in yeast. Cell12: 237-47. Johnson' F.B. and MAKrasnow. 1992. Differential regulation of transcription preinitiation complex assembly by activator and repressor homeo domain proteins. Genes Dev 6: 2 177- 89.

Johnson, W.A. and J. Hirsh. 1990. Binding of a Drosophila POU-domain protein to a sequence element regulating gene expression in specific dopaminergic neurons. Narure 343: 467-70.

Jones, B. and W. McGinnis. 1993. The regulation of empty spiracles by AbdominaI-B mediates an abdominal segment identity hnction. Genes Dev 7: 229-40.

Jorgensen, E.M. and R.L. Garber. 1987. Function and misfinction of the hvo promoters of the Drosophila Antennapedia gene. Genes Dev 1: 54-55.

Jurgens, G. 1985. A group of genes controlling spatial expression of the bithorux complex of Drosophila Nature 316: 1533- 1535.

Jurgens, G. and V. Hartenstein. 1993. The terminal regions of the body paaem. In The developmenr of Drosophilu melanogasrer (ed. M. Bate and A. Martinez Arias). pp. 687-746. Cold Spring Harbor Laboratory Press.

Jurgens, G., E. Wieschaus, C. Nusslein-Volhard, and H. Kluding. 1984. Mutations affecting the pattern pf the larval cuticle of Drosophila melanogasrer. II Zygotic loci on the third chromosome. Rom's Archives of Developmental Biology 193: 283-395.

Kagoshima, H., K. Shigesada, M. Satake, Y. Ito, H. Miyoshi, M. Ohki, M. Pepling, and P. Gergen. 1993. The Runt domain identifies a new family of heteromenc transcnptional regulators [letter]. Trends Genet 9: 338-4 1.

Kalionis, B. and P.H. O'Farrell. 1993. A universal target sequence is bound in vitro by diverse homeodomains. Mech Dev 43: 57-70.

Karnps, M.P., C. Murre, X.H. Sun, and D. Baltimore. 1990. A new homeobox gene contributes the DNA binding domain of the t(1;19) translocation protein in pre-B ALL, Ceil 60: 547-35.

Kania, M.A., AS. Bonner, J.B. Duffy, and J.P. Gergen. 1990. The Drosophila segmentation gene runt encodes a novel nuclear regulatory protein that is also expressed in the developing nervous system. Genrs Dev 4: 170 1- 13.

Karch, F, W. Bender, and B. Weiffenbach. 1990. abdA expression in Drosophila embryos. Genes Dev 4: 1573-87.

Karch, F., B. Weiffenbach, M. Peifer, W. Bender, 1. Duncan, S. Celniker, M. Crosby, and €.B. Lewis. 1985, The abdominal region of the bithorax cornplex. CeIl 43: 8 1-96.

Karlsson, J. 1981. The distribution of regenerative potential in the wing disc of Drosophila. J Embryol Erp ~Clorphol61:303-16.

Karlsson. 1. 1984, Morphogenesis and cornpartments in Drosophila. In Pattern Formation: a Primer in Developmental Biolog~(ed. G.M. Malacinski and S.V. Bryant), pp. 323-337. Macmillan and Co, New York London.

KarIsson, O.. S. Thor, T. Norberg, H. Ohlsson, and T. Edlund. 1990. Insulin gene enhancer binding protein Isl-1 is a member of a novel class of proteins conrainine both a homeo- and a Cys- His domain. Nature 344: 879-82.

Karr, T.L. and TB. Komberg. 1989. fushi tarani protein expression in the cellular blastoderm of Drosophila detected usîng a novel imaging technique. Developrnent 106: 95-103. Kassis, JA. 1990, Spatial and temporal control elements of the Drosophila engrailed gene. Grnes Dev 4: 433-43.

Katan, M., A. Haigh, C.P. Venijzer, P.C. van der Vliet, and P. O'Hare. 1990. Characterimion ofa cellular factor which interacts functionally with Oct-l in the assembly of a multicompnent transcription complex. ~VucleicAcids Res 18: 6871-80.

Kaufman, T.C., R. Lewis, and B.T. Wakimoto. 1980. Cytogenetic ana1ysis of chromosome 3 in Drosophila melanogaster: The homeotic gene complex in polytene chromosome 84A-B. Genefics 94: 1 15-133.

Kaufinan, T.C., MA. Seeger, and G. Olsen. 1990. Molecular and genetic organization of the antennapedia gene cornplex of Drosophila rnelanogaster. il& Genet 27: 309-62.

Kauhann, E. and W. Knochel. 1996. Five years on the wings of fork head. Mech Dev 57: 3-20.

Keleher, C.A., C. Goutte, and A.D. Johnson. 1988. The yeast cell-type-specific repressor alpha 2 acts cooperatively with a non-cell-type-specific protein. Ce11 53: 927-36.

Keleher, C.A., S. Passmore, and A.D. Johnson. 1989. Yeast repressor alpha 2 binds to its operator cooperativeiy with yeast protein Mcml. Mol Ce11 Bi01 9: 5228-30.

Keleher, C.A., M.J. Redd, J. Schultr, M. Carlson, and A.D. Johnson. 1992. Ssn6-Tup 1 is a general repressor of transcription in yeast, Ce11 68: 709-19.

KeiIerman, K.A., D.M. Mattson, and 1. Duncan. 1990. Mutations affecting the stability of the fushi tamu protein of Drosophila. Genes Dev 1: 1936-50.

Kennison, I.A. and J.W. Tamkun. 1988. Dosage-dependent modifiers of polycomb and antennapedia mutations in Drosophila. Proc Nat1 Acad Sci USA 85: 8 13640.

Kenyon, C. 1994. If birds can fly, why can't we? Homeotic genes and evolution. Cd1 78: 175-80.

Kilchherr, F.. S. Baumgartner. D. Bopp, E. Frei, and M. Noll. 1986. Isolation of the paired gene of Drosophila and its spatial expression during early embryogenesis. Nature 321: 493-199.

Kissinger. C.R., B.S. Liu, E. Martin-Blanco, T.B. Kornberg, and C.O. Pabo. 1990. Crystal mcture of an engrailed homeodomain-DNA complex at 2.8 A resolutian: a framework for understanding homeodomain-DNA interactions. Ce11 63: 579-90.

Klemm, J.D., MA. Rould, R. Aurora, W. Hem, and C.O. Pabo. 1994. Crystal structure of the Oct-l POU domain bound to an octamer site: DNA recognition with tethered DNA-binding modules. Cell77: 21-32.

Klingensmith, J. and N. Penïmon. 1991. Segment Polarity Genes and Intercellular Communication in Drosophila. In Genetic Approches to Understanding Cell dctivarion (ed. J. Mond, A. Weiss, and J. Gambier).

Klingler, M- and J.P. Gergen. 1993. Regulation of runt transcription by Drosophiia segmentation genes. Mech Dm 13: 3-19.

Klingler, M., J. Soong, B. ButIer. and J.P. Gergen. 1996. Disperse versus compact elements for the regulation of runt smpes in Drosophila, Dev Bi01 177: 73-84.

Knipple, D.C., E. Seifert, U.B. Rosenberg, A. Preiss, and H. Jackle. 1985. Spatial and temporai patterns of KNppel gene expression in early Drosophila embryos. Nature 317: 40-4. Knoepfler, P.S. and M.F. Kamps. 1995. The pentapeptide motif of Hox proteins is required for cooperative DNA binding with Pbxl, physicalIy contacts Pbxl, and enhances DNA binding by Pbxi. Mol Cell BiollS: 58 11-9.

Komachi, K., MJ. Redd, and A.D. Johnson. 1994. The WD repeats of Tup1 interact with the homeo domain protein alpha 2. Genes Dev 8: 2857-67.

Karnberg, T. 1981. Engrailed: a gene controlling cornpartment and segment formation in Drosophila. Proc Narl Acad Sci USd 78: 1095-9.

Karnberg, T., 1. Siden, P. O'Farrell, and M. Simon. 1985. The engraiied locus of Drosophila: in situ localization of transcripts reveals compartment-specific expression. Ce11 10: 45-53.

Kornfeld, K., R.B. Saint, P.A. Beachy, PJ. Haste, D.A. Peanie, and D.S. Hogness. 1989. Structure and expression of a farnily of U1trabithora.x mRNAs generated by alternative splicing and polyadenylation in Drosophiia, Genes Dev 3: 243-58.

Kosman, D, and S. Small. 1997. Concentration-dependent paneming by an ectopic expression dornain of the Drosophila gap gene knirps. Developrnent 124: 1343-54.

Krarner, S.G., T.M. links, P. Schedl, and J.P. Gergen. 1999. Direct activation of Sex-lethal transcription by the Drosophila runt protein. Developmenr 126: 191-200.

Krasnow, M.A.. E.E. Saffman, K. Kornfeld, and D.S. Hogness. 1989. Transcriptional activation and repression by Ultrabithomu proteins in cultured Drosophila cells. Cri1 57: 103 1-43.

Krause, H.M. and W. Gehring. 1988, The location , modification, and furtction of the fushi tara,- protein during Drosophila em bryogenesis. In Ce//larFactors in Developrnenr and D~ferentiarion:Embryos. Terafucarcinomas.and Diflerrnriared Tissues. pp. 1 05- 123. Alan R. Liss, Inc,

Krause, H.M. and W.J. Gehring. 1989. Stage-specific phosphorylation of the füshi tarazu protein during Drosophila development. Emho J8: 1 197-204.

Krause. H.M., R. Klemenz, and W.J. Gehring. 1988. Expression, modification. and localization of the fushi tarazu protein in Drosophila embryos. Gmes Dev 2: 1021-36.

Kraut. R. and M. Levine, 1991a. Mutually repressive interactions beween the gap genes gianr and ffippel define middle body regions of the Drosophila embryo, Developmenr 111: 6 11-622.

Kraut, R. and M. Levine. 1991b. Spatial regulation of the gap gene giant during Drosophila devebpment. Developmenr 111: 60 1-9.

Kristie, T.M., J.H. LeBowitz, and P.A. Sharp. 1989, The octamer-binding proteins form multi- protein-DNA cornpiexes with the HSV alpha TIF regdatory protein. Edo J8: 4229-38.

Kristie. T.M. and P.A. Sharp. 1990. Interactions of the Oct-l POU subdomains with specific DNA sequences and with the HSV alpha-ms-activator protein- Genes Dev 1: 2383-96.

KrumIauf, R. 1994. Hox genes in vertebrate development. Cd78: 19 1-20 1.

Kurant, E., C.Y. Pai, R Shatf, N. Halachmi, Y.H. Sun, and A. SaIzberg. 1998. Dorsotonals/homothora.. the Drosophila homoiogue of meisl, interacts with e.vtradenticle in patterning of the embryonic PNS, Development 125: 1037-48.

Kuroiwa, A.. E, Hafen, and W.J. Gehring. 1984. Cloning and transcriptional analysis of the segmentation gene fushi tarazu of Dmsophila. Cell37: 825-3 1, 186 Kuroiwa, A., U. Kloter, P. Baurngartner, and W. Gehring. 1985. Cloning of the homoeotic Sex combs reduced gene in Drosophila and in situ localization of iis transcripis. Embo J1.

Kuziora, M.A. and W. McGinnis. 1988. Autoregulation of a Drosophila homeotic selector gene. Ceil 55: 477-85.

Kuziora, MA. and W. McGinnis. 1989. A homeodomain substitution changes the regulatory specificity of the deformed protein in Drosophila embryos. CeIl 59: 563-71.

Kuziora, M.A. and W. McGinnis. 1990. Altering the regulatory targets of the Deformed protein in Drosophila embryos by substituting the Abdominal-B homeodomain. Mech Dev 33: 83-93.

La Rosee, A., T. Hader, H, Taubert, R. Rivera-Pomar, and H. Jackle. 1997. Mechanism and Bicoid- dependent control of hairy stnpe 7 expression in the posterior region of the Drosophiia embryo. Embo J 16: 4403-1 1.

LaBella, F., H.L. Sive, R.G. Roeder, and N. Heina. 1988, Cell-cycle regulation of a hurnan histone H2b gene is mediated by the H2b subtype-specific consensus element. Genes Dev 2: 32-9.

Lacasse, E.C. and YA. Lefebvre. 1995. Nuclear localization signals overlap DNA- or RNA-binding dornains in nucleic acid-binding proteins. Nucleic Acids Res 23: 1647-56.

Lagenaur, C. and N. Agabian. 1978. Caulobacter flagellar organelle: synthesis. compamnentation. and assembly. J Bacteriol 135: 1062-9.

Lai, J.S., M.A. Cleary, and W. Hem. 1992. A single arnino acid exchange transfers VP ldinduced positive control from the Oct-1 to the Oct-2 home0 domain [published erratum appears in Genes Dev 1992 Dec;6(12B):2663]. Genrs Dev 6: 2058-65.

LaII. S., H. Francis-Lang, A. Flament, A. Norvell, T. Schupbach, and D. Ish-Horowicz. 1999. Squid hnRNP protein promotes apical cytoplasrnic transport and localization of Drosophila pair- rule transcripts. Cell98: 17 1-80.

Larnka, M.L.. A.M. Boulet, and S. Sakonju. 1992. Ectopic expression of LJBX and ABD-6 proteins during Drosophila embryogenesis: cornpetition, not a functional hierarchy, explains phenotypic suppression. Developmenr 116: 841-54.

Langeland LA., S.F. Attai, K. Vorwerk, and SB. Carroll. 1994. Positioning adjacent pair-rule smpes in the posterior Drosophila embryo. Development 120: 3945-55.

Langland, I.A. and S.B. Carroll. 1993. Conservation of regulatory elements controlling hairy pair- rule stnpe formation. Development 117: 585-96.

LardeIli, M. and D. Ish-Horowitz. 1993. Drosophila hairy pair-rule gene regulates embryonic patteming outside its apparent smpe domains. Developmenr 118: 355-66.

Lassar, A.B., J.N. Buskin, D. Lockshon, R.L. Davis, S. Apone, S.D. Hauschka, and H. Weintraub. 1989. MyoD is a sequence-specific DNA binding protein requinng a region of myc homology to bind to the muscle creatine kinase enhancer. Ce11 58: 823-3 1.

Laughon, A., A.M, Boulet, I.R. Bermingham, Jr, RA. Lapon, and M.P. Scott. 1986. Stnictwe of uanscripts from the horneotic Antennapedia gene of Drosophila melanogaster: two promoters control the major protein-coding region. Mol Ce11 Bi01 6: 4676-89.

Laughon, A., W. Howell, and M.P. Scott. 1988. The interaction of proteins encoded by Drosophila horneotic and segmentation genes with specific DNA sequences. Development 104: 75-83. Laughon, A. and M.P. Scott. 1984. Sequence of the Drosophila segmentation gene: protein structure hornology with DNA-binding pmteins. Narure 310: 25-3 1.

Lavorgna, G., H. Ueda, J. Clos, and C. Wu. 1991. FTZ-FI, a steroid hormone receptor-like protein irnplicated in the activation of fushi tau.Science 252: 848-51. Lawrence, PA. 1992. The ibfaking ofa Flv: The Genetics of Animal Design. Blackwell Scientific Publications, Oxford.

Lawrence, P.A. and P. Johnston. 1989a. Analysis of firnction of the pair-rule genes hairy, even- skipped and fushi tarazu in rnosaic Drosophila ernbryos. Developmenr 107: 847-53.

Lawrence, P.A. and P. Johnston. 1989b. Pattern formation in the Drosophila ernbryo: allocation of cells to parasegments by even-skipped and khi tarani. Development 105: 76 1-7.

Lawrence, P.A., P. Johnston, P. Macdonald, and G. Stnihl. 1987. Borders of parasegments in Drosophila embryos are dehited by the khitarazu and even-skipped genes. Narure 328: 440-2.

Lawrence, P.A. and G. Morata. 1976. The comparanent hypotbesis. In Insecr Developmenr (ed. P.A. Lawrence). Blackwell Scientitic Publishing, Oxford.

Lawrence, PA. and L. Pick. 1998. How does the hshi tarazu gene activate engrailed in the Drosophila embryo? Dev Gener 23: 18-34.

Lawrence. PA. and J. Sampedro. 1993. Drosophila segmentation: afier the first three hours. Development 119: 97 1-6.

Lawrence. PA. and G. Struhl. 1996. Morphogens. cornpartments, and pattern: lessons from drosophila? CeIl 85: 93'1-61.

Lehmann, R. 1988. Phenotypic cornparison between maternai and zygotic genes controlling the segmental pattern of the Drosophila ernbryo. Developmenr fsuppl.) 104: 17-27.

Levine, M., E. Hafen. R.L. Garber. and W.J. Gehring. 1983. Spatial distribution of Antennapedia transcripts durinç Drosophila development. Embo J2: 203746.

Levine, M. and T. Hoey. 1988. Homeobox proteins as sequence-specific transcription factors. Cell 55: 53730.

Lewis, €.B. 1963. Genes and developrnental pathways. Amer. Zool. 3: 33-56.

Lewis, E.B. 1978. A gene complex conuolling segmentation in Drosophila. Nature 276: 565-70.

Lewis, R.A., T, Kaufman, R.E. Deneil, and P. Tallenco. 1980a- Genetic analysis of the Antennapedia Gene Cornplex (ANT-C) and adjacent chromosomal regions of Drosophila melanogasrer. 1. Polytene chrornosorne segments 846-D. Generics 95: 367-381.

Lewis, RA., B.T. Wakirnoto, R.E. Denell, and T. Kaufinan. 1980b. Genetic analysis of the Antennapedia Gene Cornplex (ANT-C) and adjacent chrornosomal regions of Drosophila melanogasrer. 11. Polytene chromosome segments 84A-84B 1,2. Generics 95: 383-397.

Li, C. and J.L. Manley. 1998. Even-skipped represses transcription by binding TATA binding protein and blocking the TFIID-TATA boa interaction. Mol CelI Bi01 18: 3771-81.

Li, F.Q., K. Takernaru, M. Goto, H, Ueda, H. Handa, and S. Hirose. 1997. Transcriptional activation thmugh interaction of MBF2 with TFiiA. Genes Celis 2: 143-13. Li, F.Q., H, Ueda, and S. Hirose. 1994. Mediators of activation of fushi tarani gene transcription by BmFTZ-F1. Mol Ce11 Biol14: 30i3-21.

Li, X. and W. McGinnis. 1999. Activity regulation of Hox proteins, a mechanism for altering hnctional specificity in developrnent and evolution, Proc Narl Acad Sei USA 96: 6802-7.

Li, X., C. Mune, and W. McGinnis. 1999a. Activity regulation of a Hox protein and a role for the homeodornain in inhibiting mscriptionaI activation. Embo J 18: 198-21 1.

Li, X. and M. Noll. 1993. Role of the gooseberry gene in Drosophila embryos: maintenance of wingless expression by a wingless--goosebeny autoregulatory loop. Embo J 12: 4499-509.

Li, X., A. Veraksa, and W. McGinnis. I999b. A sequence motif distinct fiom Hox binding sites controls the specificity of a Hox response element. Developmenr 126: 5581-9.

Liang, 2. and M.D. Biggin. 1998. Eve and ftz regulate a wide anay of genes in blastoderm embryos: the selector homeoproteins directly or indirectIy regulate rnost genes in Drosophila [published erratum appears in Development 1999 Feb;176(5):following table of contents]. Developmenr 125: 437 1-82.

Limbourg-Bouchon, B., D. Busson. and C. Lamour-Isnard, 1991. Interactions between fused. a segment-polarity gene in Drosophila, and other segmentation genes. Development 112: 417- 29.

Lin, L. and W. McGinnis. 1992. Mapping functional specificity in the Dfd and Ubx homeo domains. Genes Dev 6: 1 O7 1-8 1.

Lipshutz R.J.. S.P. Fodor. T.R. Gingeras, and DJ. Lockhart. 1999. Wigh density synthetic oligonucleotide arrays. Nat Gener 21: 20-4.

Loewy, Z.G., RA. Bryan, S.H. Reuter. and L. Shapiro. 1987. Contml of synthesis and positioning of a Caulobacter crescentus flagella protein. Genes Dev 1: 626-35.

Logan, C.. M.C. Hanks. S. Noble-Topham. DLNallainathan, N.J. Provart. and A.L. Joyner. 1992. Cloning and sequence cornparison of the mouse. human. and chicken engrailed genes reveal potential functional domains and regulatory regions. Dev Gener 13: 335-58.

Look, A.T. 1997. Oncogenic transcription factors in the human acute ieukemias. Science 278: 1059- 64.

Lu, Q. and M.P. Kamps. 1996. Structural deterrninants within Pbxl that mediate cooperative DNA binding with pentapeptide-containing Hox proteins: proposal for a model of a Pbxl-Hox- DNA cornplex. Mol Ce11 Bi01 16: 163240.

Lu, Q.. P.S. Knoepfler, 1. Scheele, D.D. Wright, and M.P. Kamps. 1995. Both Pbxl and UA-Pbxl bind the DNA motif ATCAATCAA cooperatively with the products of multiple murine Hox genes, some of which are themsehes oncogenes. Mol Ce11 Bi02 15: 3786-95.

Lutkin? T., A. Dierich, M. LeMeur, M. Mark, and P. Chambon. 1991. Disruption of the Hox-1.6 homeobox gerie results in defects in a region comsponding to its rom1 domain of expression. Ce11 66: 1 105-19.

Ma, X., D. Yuan, K. Diepold. T. Scarborough, and 5. Ma 1996, The Dmsophila morphogenetic pmtein Bicoid binds DNA cooperativeiy. Developmenr 122: 1195-206-

Macdonald, P.M., P- Ingham, and G. Süuhl. 1986. Isolation, structure, and expression of even- skipped: a second pair- mle gene of Drosophila containing a homeo box. Ce11 17: 721-34. Macdonald, P.M. and G. Stmhl. 1986. A rnolecular gradient in early Drosophila ernbryos and its mie in specifying the body pattern. Nature 324: 537-45.

Macias, A., J. Casanova, and G. Morata. 1990. Expression and regulation of the abd-A gene of Drosophila. Developmenr 110: 1197-207.

Mackem, S. and B. Roiman. 1982. Structural features of the herpes simplex virus alpha gene 4, O, and 27 promotet-regulatory sequences which confer alpha regulation on chirneric thymidine kinase genes. J Virol44: 939-49.

Mahoney, PA. and J.A. Lengyel. 1987. The zygotic segmentation mutant tailles alters the blastoderm fate map of the Drosophila embryo. Dev Bi01 122: 464-70.

Maier, D., A. Preiss, and J.R. Powell. 1990. Regulation of the segmentation gene fushi t;uant has been functionalIy conserved in Drosophila. Embo J9: 3957-66.

Mak, A. and A.D. Johnson. 1993. The carboxy-terminal tail of the homeo domain protein alpha 2 is required for function with a second horneo domain protein. Genes Dev 7: 1862-70.

Malicki, J., K. Schughart, and W. McGinnis. 1990. Mouse Hox-23 specifies thoracic segmental identity in Drosophila embryos and larvae. Ce11 63: 96 1-7.

Manak, J.R. and M.P. Scott. 1994. A class act: conservation of homeodomain protein finctions. Dev Suppl: 6 1-77.

Mangin. M.. M. Ares. Jr., and A.M. Weiner. 1986. Human U2 small nuclear RNA genes contain an upstream enhancet. Embo J5: 987-95.

Mann. R.S. 1995. The specificity of horneotic gene function. Bioessays 17: 855-63.

Mann, R.S. and M. Affolter. 1998. Hox proteins meet more partners. Curr Opin Grner Dev 8: 423-9.

Mann, R.S. and S.K. Chan. 1996, Extra specificity from e.madenticle: the partnership benveen HOX and PBXEXD horneodomain proteins [published erratum appears in Trends Genet 1996 Auç: 12(8):328]. Trends Genet 12: 258-62.

Mann. R.S. and D.S. Hogness. 1990. Functional dissection of Ultrabithora. proteins in D. melanogaster. Cr11 60: 597-6 IO.

Mannervik, M. and M. Levine. 1999. The Rpd3 histone deaceqlase is required for segmentation of the Drosophila ernbryo. Proc ~VatlArad Sci USA 96: 6797-801.

Manoukian, AS. and H.M. Krause. 1992. Concentration-dependent activities of the even-skipped protein in Drosophila embryos. Genes Dev 6: 1740-51.

Manoukian, AS. and H.M. Krause. 1993. Control of segmentai asprnetry in DrosophiIa embryos. Development 118: 785-96.

Marsden H.S.. M.E. Campbell, L. Ham. M.C. Frame, D.S. Parris, M. Murphy, R.G. Hope. MT. Muller, and C.M. Preston. 1987. The 65,000-Mr DNA-binding and virion trans-inducing pmteins of herpes simplex virus type 1. J Virol61: 2428-37.

Martinez Arias, A. 1986- The Rntennapedia gene is required and expressed in parsegments 4 and 5 of the Drosophila embryo. Embo J1: 135-141.

Martinez Anas, A, and R.A. White. 1988. ültrubithorm and engrailed expression in Drosophila embryos mutant for segmentation genes of the pai~nileclass. Devefopmenr 102: 325-338. Martinez-Arias, A,, P.W. Ingham, M.P. Scotf and M.€-Akam. 1987. The spatial and temporal deployment of Dfd and Scr ttanscripts throughout development of Drosophila. Developrnent 100: 673-83.

Martinez-Arias, A. and PA. Lawrence. 1985. Parasegments and compartments in the Drosophila embryo. Nature 313: 63942.

Martizez Arias, A., N.E. Baker, and P.W.Ingharn. 1988. Role of segment polarity genes in the definition and maintenance of cell states in the Drosophila embryo. Development 103: 157- 70. Mattaj, LW., S. Lienhard, J. Jiricny, adE.M. De Robertis. 1985. An enhancer-like sequence within the Xenopus U1 gene promoter facilitates the formation of stable transcription complexes. Nature 3 16: 163-7,

McGinnis. N.. M.A. Kuziora, and W. McGinnis. 1990. Human Hox-4.2 and Drosophila defomed encode similar regulatory specificities in Drosophila embryos and larvae. Cd63: 969-76.

McGinnis, W.. R.L. Garber, J. Wirz A. Kuroiwa, and W.J. Gehring. 1984a- A homologous pmtein- coding sequence in Dmsophila homeotic genes and its conservation in other metazoans. Cell 37: 403-8.

McGinnis, W. and R. Krumlauf. 1992. Homeobox genes and axial patteming. Cell63: 283-302.

McGinnis, W.. M.S. Levine, E. Hafen, A. Kuroiwa, and W.J. Gehring. 1984b. A conserved DNA sequence in homoeotic genes of the Drosophila Antennapedia and bithora.. complexes. Narure 308: 428-3 3.

Meinhardt. H. 1983. CeIl determination boundaries as organizing regions for secondary embryonic fields. Dev Biol96: 375-85.

Meinhardt, H. 1986a. Hierarchical inductions of cell states: a mode1 for segmentation in Drosophila. J Cell Sci Suppl1: 357-81.

Meinhardt H. 1986b. The threefold subdivision of segments and the initiation of legs and wings in insects. Trends Genet. 2: 36-11.

MerriIl, V.K., F.R. Tumer, and T.C. Kauhan. 1987. A genetic and developmental analysis of mutations in t9e Defonned locus in Drosophila meianogaster, Dev Biol122: 379-95.

Meyer. K.B. and M.S. Neuberger. 1989. The immunoglobulin kappa locus contains a second, strongr 6-cell- specific enhancer which is located downstrearn of the constant region. Embo J8: 193'9-64.

Mihm H. and E.T. Kaiser. 1988. A chemically synthesized Antennapedia homeo domain binds to a specific DNA sequence. Science 212: 925-7.

Miklos. G.L. and G.M. Rubin. 1996. The role of the genome project in deteminhg gene function: insights hmmode1 organisms. CeIl86: 521-9.

Minnich, SA. and A. Newton. 1987. Promoter mapping and ce11 cycle regulation of fiagellin gene transcription in Caulobacter crescentus, Proc Nat[ Acad Sci US-4 81: 1142-6.

Mirkovitch?J. and JE. Damell, Jr. 1991. Rapid in vivo footpnnting rechnique identifies proteins bound to the TTR gene in the mouse liver. Genes Dev 5: 83-93. Mlodzik, M., C.M. De Montrion, Y. Hiromi, H.M. Krause, and W. Gehring. 1987. The influence on the blastoderm fate map of matemal-effeçt genes that affect the antero-postecior pattern in Drosophila. Genes Dev 1: 603-6 14.

MIodzik, M., A. Fjose, and W. Gehring. 1985. Isolation of caudal' a Drosophila homeobox containing gene with matemal expression, whose transcripts fonn a concentration gradient at the pre-blastoderm stage. Embo J 4: 2961-2969.

Mlodzik, M. and W.J. Gehring. 1987. Expression of the caudal gene in the genn Iine of Drosophila: formation of an RNA and protein gradient during early embryogenesis. Ce11 18: 465-78.

Mlodzik, M.. G. Gibson, and W.J. Gehring. 1990. Effects of ectopic expression of caudal during Drosophila development. Developmenr 109: 271-7.

Moazed, D. and P.H. O'Faml!. 1992. Maintenance of the enpiled expression pattern by Polycomb group genes in Drosophila. Developmenr 116: 805-10.

Mohler, J., E.D. Eldon, and V. Pirrotta. 1989. A novel spatial transcription pattern associated with the segmentation gene? gianr, of Drosophila. Embo J8: ljB-ljj8.

Mohler. J. and K. Vani. 1992. Molecular organization and embryonic expression of the hedgehog gene involved in celi-cell communication in segmental patterning of Drosophila. Developmenr 115: 957-7 1.

Monica, K., N. Galili, J. Nourse, D. Saltman. and ML. Cleary. 1991. PBX2 and PBX3. new homeobox genes with extensive homology to the human proto-oncogene PBXl. Mol Crll Biol11: 6149-57.

Morata. G. and PA. Lawrence. 1975. Control of cornpartment development by the engrailed gene in Drosophila. Narure 255: 614-7.

Morinaga, T.. H. Yasuda T. Hashimoto. K. Higashio, and T. Tamaoki. 1991. A human alpha- fetoprotein enhancer-binding protein, ATBF 1, contains four homeodomains and seventeen zinc fingers, Mol CeIl Bi01 11: 6041-9.

Momssey, D., D. Askew. L. Raj, and M-Weir. 199 1. Functional dissection of the paired segmentation gene in Drosophila embryos. Genes Dev 5: 1684-96.

Moskow, J.J.. F. Bullrich, K. Huebner, 1.0. Dam, and A.M. Buchberg. 1995. Meisl, a PBX1-related homeobox gene involved in myeloid leukernia in BXH- 2 mice. hl01 Ce11 Bi01 15: 5434-43.

Mullen, J.R. and S. DiNardo. 1995. EstabIishing parasegments in Drosophila embryos: roles of the odd- skipped and naked genes. Dev Biot 169: 295-308.

Muller. J. and M. Bienz. I991. Long range repression conferring boundaries of Ultrabithorar expression in the Drosophiia embryo. Embo J10: 3 147-55.

Muller, J. and M. Bienz. 1992. Sharp anterior boundary of homeotic gene expression conferred by the !khi tarazu protein. Embo J11: 3653-61.

Muller, M.: M. Affolter, W. Leupin, G. ûtting, K. Wuthrich, and WJ. Gehrïng. 1988. isolation and sequence-specific DNA binding of the Antennapedia homeodornain. Embo JI: 4299-304.

Murata, Y. and R.P. Wharton. 1995. Binding of pumilio to materna1 hunchback mRNA is requued for posterior patteming in Drosophila embryos. Cell80: 747-56.

Myers, E.W- et aI. 2000. A whole-genome assembiy of Drosophila. Science 287: 2196-204. Nakamura, T., NA. Jenkins, and N.G. Copeland. 1996. Identification of a new farnily of Pbx-related homeobox genes. Oncogene 13: 2235-42.

Nakano, Y., 1. Guerrero, A. Hidalgo, A. Taylor, J.R Whittle, and P.W. Ingharn. 1989. A pmtein with several possible membrane-spanning domains encoded by the Drosophila segment polarity gene patched. Nature 311: 508-13.

Nasiadka, A. and H.M. Krause. 1999. Kinetic analysis of segmentation gene interactions in Drosophila embryos. Developmeni 126: 1515-26.

Nauber, U., M.J. Pankratz, A. Kienlin, E. Seifert, U. Klemm, and H. Jackle. 1988. Abdominal segmentation of the Drosophila embryo requires a hormone receptor-like protein encoded by the gap gene knirps. Nature 336: 489-92.

Neely, KI., A.H. Hassan, A.E. Wdlberg, D.J. Steger, B-R Cairns, A.P. Wright, and J.L. Workrnan. 1999. Activation dornain-mediated targeting of the SWSNF complex to promoters stimulates transcription fiom nucleosome arrays. MOI Ce11 4: 649-55.

Neuteboorn. S.T., L.T. Peltenburg, MA. van Dijk, and C. Murre. 1995. The hexapeptide LFPWMR in Hoxb-8 is required for cooperative DNA binding with PbxI and Pb2 proteins. Proc Nail Acad Sci U S A 92: 9 166-70.

Nibu. Y., H. Zhang, and M. Levine. 1998- Interaction of short-range repressors with Drosophila CtBP in the embryo. Science 280: 1014.

Nolte. R.T., G.B. Wisely, S. Westin, I.E. Cobb, M.H. Lambert. R. Kurokawa M.G. Rosenfeld, T.M. Willson, C.K. Glass, and M.V. Milbum. 1998. Ligand binding and CO-activatorassembly of the peroxisome proliferator- activated receptor-gamma. Nanrre 395: 13743.

Nourse, J., J.D. Mellentin, N. Galili, J. Wilkinson, E. Stanbridge, S.D. Smith, and M.L. Cleary. 1990. Chromosomal translocation t(l;l9) results in synthesis of a homeobox fusion mRNA that codes for a potential chimeric transcription factor. Cell60: 535-45.

Nusslein-Volhard, C., H.G. Frohnhofer, and R. Lehmann. 1987. Determination of anteroposterior polarity in Drosophila. Science 238: 1675-8 1.

Nusslein-Volhard, C. and E. Wieschaus. 1980. Mutations affecting segment number and polarity in Drosophila. Nattire 287: 795-80 1.

Nusslein-Volhard, C., E, Wieschaus. and H. Kluding. 1984. Mutations affecting the pattern of the larval cuticle of Drosophila melanoguster. 1 Zygotic loci on the second chromosome. Roux's Archives of Developmenral Biology 193: 267-282.

O'Connor, M.B.. R. Binari, LA. Perkins, and W. Bender. 1988. Alternative RNA products from the Ultrabithom. domain of the bithomv complex. Embo J7: 435-45.

Odenwald, W.F., J. Garbern, H. Amheiter, E. Tournier-Lasserve, and R.A. Lanarini. 1989. The Hox- 13 homeo box pmtein is a sequence-specific DNA-binding phosphoprotein [published erratum appears in Genes Dev 1989 Aug;3(8): 126q. Genes Dev 3: 158-72.

O'Far~ell,P.H. 1992. Developmental biolow. Big genes and little genes and deadlines for transcription [news]. ~Varure359: 366-7,

O'Hara, E., B. Cohen, S.M. Cohen, and W. McGinnis. 1993. Distal-less is a downstream gene of Deformed required for ventral mavillary identity. Developmenr 117: 847-56. 193 Ohkuma, Y., M. Horikoshi, R.G. Rwder, and C. Desplan. 1990. Binding site-dependent direct activation and repression of in vitro transcription by Drosophila homeodomain pmteins. Ce11 61: 475-84.

Ohno, C.K. and M. Petkovich. 1993, FTZ-FI beta, a novel member of the Drosophila nuclear receptor family. Mech Dev 10: 13-24.

Ohno, C.K., H. Ueda, and M. Petkovich. 1994. The Drosophila nuclear receptors FTZ-FI alpha and FTZ-FI beta cornpete as monomers for binding to a site in the fushi tarani gene. Mol Ce11 Bi01 11: 3 166-75.

Ohta, N., L.S. Chen, DA. Mullin, and A. Newton. 1991. Timing of flagellar gene expression in the Caulobacter ceII cycle is detennined by a transcriptional cascade of positive regu1atot-y genes. J Bacferiol173: 1514-27.

Ohtsuki, S., M. Levine, and H.N. Cai. 1998. Different core promoters possess distinct regulatory activities in the Drosophila embryo. Genes Dev 12: 547-56.

Orenic, T.V., D.C. Slusarski, ILL. KroII. and R.A. Holmgren. 1990. Cloning and characterization of the segment polarity gene cubitus intempnis Dominant of Drosophila. Genes Dev 1: 1053- 67.

Otting, G., Y.Q. Qian, M. Billeter, M. Muller, M. Affolter, W.J. Gehring, and K. Wuthrich. 1990. Protein--DNA contacts in the structure of a homeodomain-DNA cornplex detennined by nuclear magnetic resonance spectroscopy in solution. Embo J9: 3085-92.

Otting, Ci., Y.Q. Qian, M. Muller, M. Affolter, W. Gehring, and K. Wuthrich. 1988. Secondary structure determination for the Antennapedia homeodomain by nuclear magnetic resonance and evidence for a helix-turn-helk motif. Embo J7: 4305-9.

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

Pai. C.Y.. T.S. Kuo, T.J. Jaw, E. h-a~t,C.T. Chen. D.A. Bessarab, A. Salzberg, and Y.H. Sun. 1998. The homo thora,^ homeoprotein activates the nuclear localization of another homeoprotein, extradenticle, and suppresses eye development in Drosophila. Genes Dev 12: 43546.

Pankratz, MJ. and H. Jackle. 1993. Blatoderm segmentation. In The Development of Drosophila melanoguster (ed. hl. Bate and A. Martinez-Arias), pp. 467-516. Cold Spring Harbor Labotatory Press.

Pankratz, MJ., E. Seifett, N. Gerwin, B. BiIIi, LI. Nauber. and H. Jackle. 1990. Gradients of Kruppei and knirps gene products direct pair-mIe gene stripe patterning in the postenor region of the Drosophila embryo. Cell61: 309-1 7.

Paro, R. 1990. imprinting a determined state into the chromatin of Drosophila. Trends Genet 6: 416-21.

Paroush, Z., RL. Finiey, Jr., T. Kidd, S.M. Wainwrighh P.W. Ingharn, R. Brenh and D. Ish- Horowicz 1994. Groucho is required for Drosophila neurogenesis, segmentation, and sex determination and interacts directly with ha*-related bHLH proteins. Ce11 79: 805- [ 5.

Parslow, T.GI D.L. Blair, W.J. Murphy, and D.K. Gramer. 1984. Structure of the 5' ends of immunoglobulin genes: a novel conserved sequence. Proc Nat1 Acad Sci USR 81: 2650-4.

Passmore, S., R Elble, and B.K. Tye. 1989. A protein involved in minichromosome maintenance in yeast binds a transcriptional enhancer conserved in eukaryotes. Genes Dev 3: 921-35. Passner, J.M., H.D. Ryoo, L. Shen, RS. Mann, and A.K. Agganval. 1999. Structure of a DNA-bound Ultrabithomu-Extradenticle homeodomain complex [see comments]. Nature 397: 714-9.

Peifer, M., F. Karch, and W. Bender. 1987. The bithomu cornplex: Control of segmentai identity. Genes Dev 1: 89 1-898.

Peifer, M., C. Rauskolb, M. Williams, B. Riggleman, and E. Wieschaus. 1991. The segment polarity gene armadillo interacts with the wingless signaling pathway in both embryonic and adult pattern formation. Developmenr 111 : 1029-43.

Peifer, M. and E. Wieschaus. 1990. Mutations in the Drosophila gene extradenticle affect the way specific homeo domain proteins reguiate segmenta1 identity. Grnes Dev 4: 1209-23.

Percival-Smith, A., M. Muller, M. AffoIter, and W.J. Gehring. 1990. The interaction with DNA of wild-type and mutant fushi tarazu homeodomains [published erratum appears in EMBO J 1997 Jan: l i( 1 ):382]. Embo J9: 3967-74.

Perrimon, N.. L. Engstrom, and A.P. Mahowald. 1989. Zygotic lethals with specific materna1 effect phenotypes in DrosophiIa melanogaster. 1. Loci on the X chromosome. Genetics 121: 333- 52.

Pemmon, N, and A.P. Mahowald. 1987. Multiple functions of segment polariry genes in Drosophila. Dev Bi01 119: 587-600.

Pemmon, N. and D. Smouse. 1989. MuItipie hnctions of a Drosophila homeotic gene. zeste-white 3, during segmentation and neurogenesis. Dev Bi01 135: 287-305.

Peterson, C.L. and 1. Herskowitz, 1992, Charactenzation of the yeast SWI1, SW12. and SWI3 genes. which encode a global activator of transcription. Crll68: 573-83.

Phelan. M.L.. 1. Rambaldi. and MS. Featherstone. 1995. Cooperative interactions between HOX and PBX proteins mediated by a conserved peptide motif. .Mol Cr11 Bi01 15: 3989-97.

Pick, L.. A. Schier. M. Affolter. T. Schmidt-Glenewinkel, and W.J. Gehring, 1990. Analysis of the ftz upstream element: germ layer-specific enhancers are independently autoregulated. Genes Dev 4: l??4-B.

Pignoni. F.. R.M. Baldarelli. E. Steingrimsson, RJ. Diaz A. Patapoutian. J.R. Merriam. and JA. Lengel. 1990. The Drosophila gene tailIess is expressed at the embryonic termini and is a member of the stemid receptor superfamiiy. Ceil 62: 151-63.

Pignoni. F., E. Steingrimsson. and I.A. Leneel. 1992. bicoid and the terminal synem activate tailless expression in the early Drosophila embryo. Development 115: 239-5 1.

Pinsonneault, J.. B. Florence, H. Vaessin, and W. McGinnis. 1997. A mode1 for emdenticle function as a switch that changes HOX proceins from repressors to activators. Embo J16: 2032-42.

Piper, D.E., A.H. Batchelor, CI-Chang, M.L. Cleary, and C. Wolberger. 1999. Structure of a Hoxi3 1-PbxI heterodimer bound to DNA: deof the hexapeptide and s fourth homeodomain helix in cornplex formation. Cell96: 587-97.

Pimotta. V. 1997a. Chromatin-silencing mechanisms în Drosophila maintain patterns of gene e.qression. Trends Genet 13: 3 14-8.

Pimitta. V, 199%. PcG complexes and chmatin silencing. Cuir Opm Genet Dev 7: 29-58. Pomerantz, J-L., TM, Kristie, and PA. Sharp. 1992. Recognition of the surface of a homeo domain protein. Genes Dev 6: 2047-57.

Poole, S.J., L.M Kauvar, B. Drees, and T. Kornberg. 1985, The engrailed locus of Drosophila: structural analysis of an embryonic transcript. CellilO: 37-43.

Poortinga, G., M. Watanabe, and S.M. Parkhurst. 1998. Drosophiia CtBP: a Hail-interacting protein required for embryonic segmentation and haiy-mediated transcriptional repression. Embo J17: 2067-78.

Popperl, H., M. Bienz, M. Studer, S.K. Chan, S. Aparicio, S. Brenner, R.S. Mann, and R. Kmlauf. 1995. Segmental expression of Hoxb-1 is controiled by a highly conserved autoregulatory loop dependent upon exd/pb~Cell 81: 103142.

Preat, T., P. Therond. C. Lamour-Isnard, B. Limbourg-Bouchon, H. Tricoire, 1. Erk, M.C. Mariol, and D. Busson. 1990. A putative serindthreonine protein kinase encoded by the segment- polarity iüsed gene of Drosophiia, Nature 347: 87-9.

Preston. CM., M.C. Frame, and M.E. Campbell. 1988. A complex formed between cell components and an HSV structural polypeptide binds to a viral immediate early gene regulatory DNA sequence. Ceil 52: 53-34.

Price. M.. D. Lazzaro, T. Pohl, M.G. Mattei, U. Ruther? J.C. Olivo, D. Duboule, and R. Di Lauro. 1992. Regional expression of the homeobox gene Nk-2.2 in the developing mammalian forebrain. Neuron 8: 24 1-55.

Pritchard, D.K. and G. Schubiger. 1996. Activation of transcription in Drosophila embryos is a gradua1 process mediated by the nucIeocytoplasmic ratio. Genes Dev 10: 1 13 1-42.

Ptashne, M. 1992. A genecic switch: phage lambda and higher organisms. Cell Press and Blackwelt Scientific Publications.

Qian, S.. M. Capovilla and V. Pirrotta. 199 1. The bx region enhance- a distant cis-control element of the Drosophila Ubx gene and its regulation by hunchback and other segmentation genes. Embo J 10: 1415-23.

Qian, S.. M. Capovilla and V. Pirrotta. 1993. Molecular mechanisms of pattern formation by the BRE enhancer of the Ubs gene. Embo 3 12: 3865-77.

Qian, Y.Q., M. Billeter, G. Otting, M. Muller, W.S. Gehring, and K. Wuthrich. 1989. The structure of the Antennapedia homeodomain detennined by NMR spectroscopy in solution: comparison with prokajotic repressors [published erratum appears in Cell 1990 May 4:61(3):548]. Ce11 59: 573-80.

Qian, Y.Q., K. Furukubo-Tokunaga, D. Resendez-Perez, M, Muller, WJ. Gehring, and K. Wuthrich. 1994. Nuclear magnetic resooance solution structure of the fushi tarazu homeodomain from Drosophila and cornparison with the Antennapedia homeodomain. JMol Bi01 238: 333-15.

Rabbitts, T.H. 1994. Chromosomal transIocations in human cancer. Nature 372: 143-9.

Rafiery, L.A., V. Twombly, K. Wharton. and W-M, Gelbart. 1993'. Genetic screens to identiQ elements of the decapentapiegic signaling pathway in Drosophila. Generics 139: 241-54-

Rauskolb, C., M. Peifer, and E. Wieschaus. 1993. extradenticle, a regulator of homeotic gene activity, is a homolog of the homeobox-containing human proto-oncogene pbxi. Cell74: 1101-12- Rauskolb, C. and E. Wieschaus. 1994. Coordinate regulation of downstrearn genes by extradenticle and the homeotic selector proteins. Embo J 13: 3561-9.

Read, D., M. Levine, and J.L. Manley. 1992. Ectopic expression of the Drosophila aamuack gene resuits in multiple embryonic defects, including cepression of even-skipped and fushi tarazu. Mech Dev 38: 183-95.

Read, D. and J.L. Manley. 1992. Alternatively spliced transcripts of the Drosophila trammck gene encode zinc finger proteins with distinct DNA binding specificities. Embo J 11: 1035-44.

Rechsteiner, M. 1988. Regulation of enzyme levels by proteolysis: the role of pest regions. ddv Enzyme Regul27: 135-5 1.

Rechsteiner, M. and S.W. Rogers. 1996. PEST sequences and regulation by proteolysis. Trends Biochem Sci 2 1: 267-7 1.

Reichardt. H.M., K.H. Kaesmer, J. Tuckermann, O. Kretz, O. Wessely, R. Bock, P. Gass. W. Schmid, P. Herrlich, P. Angel, and G. Schua. 1998. DNA binding of the glucoconicoid receptor is not essential for survival [see comments]. CeIl 93: 53 1-41.

Riddihough, G, and D. Ish-Horowitz. 199 1. individual stripe regulatory elements in the Drosophila hairy promoter respond to maternal, gap, and pair-rule genes. Genes Dev 5: 840-54.

Rieckhof, G.E., F. Casares. H.D. Ryoo, M. Abu-Shaar, and R.S. Mann. 1997. Nuclear translocation of extradenticle requires homothom. which encodes an extradenticle-related homeodomain protein. Cell91: 17 1-83.

Riedl, A. and M. Jacobs-Lorena. 1996. Determinants of Drosophila fûshi tarazu mRNA instability. :Vol Ce11 Bi01 16: 3047-53.

Riggleman, B., P. Schedl, and E. Wieschaus. 1990. Spatial expression of the Drosophila segment polarity gene annadillo is posttranscnptionally regulated by wingless. Ce11 63: 549-60,

Riggleman, B., E. Wieschaus, and P. Schedl, 1989. Molecular analysis of the armadillo locus: unifonnly distributed uanscripts and a protein with novel intemal repeats are associated with a Drosophila segment polarity gene. Genes Dev 3: 96-1 13.

Rijsewijk. F.. M. Schuermann, E. Wagenaar. P. Parren, D. Weigel, and R. Nusse. 1987. The Drosophila homolog of the mouse mammary oncogene int-1 is identical to the segment polarity gene wingless. Ce11 50: 649-57.

Riley, G.R., €.M. Jorgensen, RK. Baker, and R.L. Garber. 1991. Positive and negative control of the Antennapedia promoter P2. Dev Suppl 1: 177-85.

Riley, P.D., S.B. Carroll, and M.P. Scott. 1987. The expression and regulation of Sex combs reduced protein in Drosophila embryos. Genes Dev L: 716-730.

Rivera-Pomar. R. and H. Jackle. 1996. From gradients to stripes in Drosophila embryogenesis: fiIling in the gaps. Trends Genet 12: 478-83.

Rivera-Pomar, R., D. Niessing, U. Schmidt-Ott, W.J. Gehring, and H. Jackle. 1996. RNA binding and translational suppression by bicoid [see comments]. Natwe 379: 746-9.

Rosenberg, U., C. Schroder, A. Preiss, A. Kienlin, S. Cote, 1. Riede, and H. Jackle. 1986. Structwal homolog of the pmduct of the Drosophila Kruppel sene with Xenopus transcription factor IIIA. ivarure 319: 336-339. Rothe, M., U. Nauber, and H. Jackle. 1989, Three hormone receptor-like Drosophila genes encode an identical DNA- binding finger. Embo J8: 3087-94,

Rubin, G.M. and A.C. Spradling. 1982. Genetic transformation of Drosophila with transposable element vectors. Science 218: 348-53.

Rushlow. C.A.. A. Hogan, S.M. Pinchin, K.M. Howe. M. Lardelli, and D. Ish-Horowitz. 1989. The Drosophila hairy protein acts in both segmentation and bnstle patteming and shows homology to N-myc. Embo J8: 3095-103.

Ryoo. H.D. and R.S. Mann. 1999. The control of mink Hox specificity and activity by Extradenticle. Genes Dev 13: 1704-16.

Saavedra, C., K.S. Tung, D.C. Amberg, A.K. Hopper, and C.N. Cole. 1996. Regulation of mRNA export in response to stress in Saccharomyces cerevisiae. Grnes Dev 10: 1608-20.

Sadowski, 1.. J. Ma, S. Triezenberg, and M. Ptashne. 1988. GALJ-VPl6 is an unusually potent transcriptional activator. Nature 335: 563-4.

Sanchez-Herrero, E., 1. Vernas. R. Marco, and G. Morara. I985. Genetic organization of Drosophila bithomv complex. ~Varure313: 108-1 13.

Saulier-Le Drean, B.. A. Nasiadka I. Dong, and HM. buse. 1998. Dynamic changes in the functions of Odd-skipped during eariy Drosophila embryogenesis. Developmrnr 125: 485 1- 61.

Schier. A.F. and W.J. Gehring. 1992. Direct homeodomain-DNA interaction in the autoregulation of the fushi tarazu gene. Narure 356: 804-7.

Schier. A.F. and WJ. Gehring. 1993a Analysis of a fushi tarazu autoregulatory element: multiple sequence elemenn conmbute to enhancer activity. Embo J 12: 11 1 1-9.

Schier. A.F. and W.J. Gehnng. 1993b. FunctionaI specificity of the homeodomain protein fushi tarazu: the role of DN A-binding speci ficity in vivo. Proc Nat1 dcad Sci U S -4 90: 1450-4.

Schneider. 1. 1972. Cell lines derived from Iate embryonic stages of Drosophila melanogaster. J Embtyol Exp Morpho1 27: 353-65.

Schneuwly, S., R. Klement and W.J. Gehring. 1987. Redesigning the body plan of Drosophila by ectopic expression of the homaeotic gene Antennapedia. lVarure 325: 816-8.

Schneuwly, S., A. Kuroiwa. P. Baumgarmer, and WJ. Gehring. 1986. Structural organization and sequence of the homeotic gene Antennapedia of Drosophila melanogaster. Embo JS: 733-9.

Schroder, C., D. Taua. E. Seife~and H. Jackle. 1088. Differential regulation of the two transcripts hmthe Drosophila gap segmentation gene hunchback. Embo Jl: 2881-7.

Schubiger. G. and M. Schubiger. 1978. Distd transformation in Drosophila leg imagina1 disc fragments. Dev Biot 67: 286-95.

Schupbach, T. and E. Wieschaus. 1986a. Gennline autonomy of matemal-effect mutations altering the embryonic body panem of Dmsophila Dev Biol 113: 343-8.

Schupbach. T. and E. Wieschaus. 19865. Maternal-effect mutations altering the antenor-posterior pattern of the Drosophila ernbryo. Roux's Archives of Developmenrat Biologv 195: 302-3 17. Schupbach, T. and E. Wieschaus. 1989. Female sterile mutations on the second chromosome of Drosophila melanogaster. 1. Matemal effect mutations. Generics 121: 10 1-1 7.

Schupbach, T. and E. Wieschaus. 1991. Female sterile mutations on the second chromosome of Drosophila melanogaster. II. Mutations blocking oogenesis or altering egg morphology. Generics 129: 11 19-26.

Schwaq C.J., H.M. Sampson, D. Hlousek, A. Percival-Smith, J.W. Copeland, AJ. Simmonds, and H.M. Krause. 200 1. FTZ-Factor1 and Fushi tarani interact via conservcd nuclear receptor and coactivator motifs. Embo J2O: 5 10-5 19.

Schwarz-Sommer. Z.. P. Huijser, W. Nacken, H. Saedler, and H. Sommer. 1990. Genetic control of flower development by homeotic genes in Antirrhinum majus. Science 250: 93 1-6.

Scott, M.P- 1993. A rational nomenclature for vertebrate homeobox (HOX)genes. iVucleic Acids Res 21: 1687-8.

Scott, M.P. and A.J. Weiner. 1984. Structurril relationships among genes that control development: sequence homology behveen the Antennapedia, Ultrabithomu, and fushi tatant loci of Drosophila, Proc Narl kad Sci USA 81: 41 15-9.

Scott. M.P.. A.J. Weiner, T.I. Hazelrigg, B.A. Polisky, V. Pirrotta, F. Scalenghe. and T.C. Kauhan. 1983. The molecular organization of the Antennapedia locus of Drosophila. CeII 35: 763-76.

Seed, B- and J.Y. Sheen. 1988. A simple phase-extraction assay for chlorarnphenicol acy ttransferase activity. Grne 67: 27 1-7.

SerfIing, A.C., M. lasin, and W. Schafier. 1985. Enhancers and eukaryotic gene transcription. Trendr Gena 1: 224-230.

Serrano. N., H.W. Brock, C. Demeret. J.M. Dura, N.B. Randsholt. T.B. Komberg, and F. Maschat, 1995. polyhomeotic appears to be a target of engrailed regulation in Drosophila, Drvelopmenr 12 1: 169 1-703,

Sharkey, M.. Y. Graba and M.P. Scott. 1997. Hox genes in evolution: protein surfaces and paralog groups. Trends Grner 13: 145-5 1.

Shen, W.F.. C.P. Chang. S. Rozenfeld, G. Sauvageau. R-K, Humphries, M. Lu, H.J. Lawrence, ML. Cleary, and C. Largman. 1996. Hox homeodomain proteins eAibit selective cornplex stabilities with Pbx and DNA. Nucleic Acids Res 21: 898-906.

Shepherd, J-C., W. McGinnis. A.€. Curasco, E.M. De Robertis, and WJ. Gehring. 1984, Fly and hg homoeo domains show homologies with yeast mating type regulatory proteins. Nature 310: 70- 1.

Shiau, A.K.. D. Bantad. P.M. Loria, L. Cheng, PJ. Kushner, DA. Agard, and G.L- Greene. 1998. The stnicmral basis of esuogen receptor/coactivator recognition and the antagonism of this interaction by tamoxifen. Cell95: 927-37.

Shivdasani, RA. and S.H. Orkin. 1996. The transcriptional control of hematopoiesis [see comments]. Blood 87: 4025-39.

Siegfried. E., TB. Chou, and N. Pemmon. 1992. wingless sigaling acts through zeste-white 3. the Drosophila homolog of glycogen synthase kinase-3, to regdate en-piled and establish ce11 fate. CelI7l: 1167-79. Siegfried, E., LA. Perkins, T.M. Capaci, and N, Pecrimon. 1990. Putative protein kinase product of the Drosophila segment-polarïty gene zeste-white3. Nature 345: 825-9.

Simon, J. 1995. Locking in stable scates of gene expression: transcriptional connol during Drosophila development. Curr Opin Ce11 Bi01 7: 376-85.

Simon, MA., D.D. Bowtell, G.S. Dodson, T-R Laverty, and G.M. Rubin. 1991. Ras1 and a putative guanine nucleotide exchange factor perform crucial steps in signaling by the sevenless protein tyrosine kinase. Ce11 67: 70 1-16.

Simpson-Brose, M., J. Treisman, and C. Desplan, 1994. Synergy between the hunchback and bicoid morphogens is required for anterior patterning in Drosophila, Cell78: 855-65.

Singh, H., R. Sen, D. Baltimore, and PA. Sharp. 1986. A nuclear factor that binds to a conserved sequence motif in transcriptional control elemenrs of immunogIobulin genes. Nature 319: 154-8.

SIack, J.M.W. 199 1. From rgg to embryo.

Srnall, S., D.N. Arnosti, and M. Levine. 1993. Spacing ensures autonomous expression of different stripe enhancers in the even-skipped prornoter. Development 119: 762-72.

Small, S.. A. Blair, and M. Levine. 1992. Regulation of even-skipped stripe 2 in the DrosophiIa embryo. Embo J 11: 1047-57.

Srnall, S., A. Blair, and M. Levine. 1996. Regulation of two pair-rule stripes by a single enhancer in the Drosophila embryo. Dev Bi01 175: 3 14-24.

Smalt. S., R. Kraut, T. Hoey, R. Warrior, and M. Levine. 1991. Transcriptional regulation of a pair- rule stripe in Drosophila. Genes Dev 5: 827-39.

Smith, D.L- and A.D. Johnson. 1992. A molecular mechanism for combinatorial control in yeast: MCMl protein sets the spacing and orientation of the homeodomains of an alpha 2 dirner. Ce11 68: 133-47.

Smith. S.T. and J.B. Jaynes. 1996. A conserved region of engrailed, shared among al1 en-, gsc-, NkI-. Nkl- and msh-class horneoproteins. mediates active transcriptional repression in vivo. Devrlopment 122: 3 141-50.

Smolik-Utlaut, S.M. 1990, Dosage requirements af Ultrabithora.~and bithomuoid in the determination of segment identiq in Drosophila melanogaster. Genetics 121: 357-66,

Seller, W.C., S.J. Poole. and T. Komberg. 1988. In vitro transcription of the Drosophila engrailed gene. Genes Dev 2: 68-8 1.

Sprenger, F. and C. Nusslein-Voihard. 1993. The terminal system ofaxis determination in the Drosophila embryo. In The development of Drosophila melanogastrr (ed. M. Bate and A. Martinez Arias), pp. 365-386. Cold Spring Harbor Laboratory Press.

St Johnston, D. and C. Nusslein-Volhard. 1992, The origin of pattern and polarity in the Drosophila embryo. Cell68: 201-19.

St. Johnston. D., W. Dnever, T. Berleth, S. Richstein, and C. Nusslein-Voihard. 1989. Multiple steps in the Iocalization of bicoid RNA to the anterior pole of the Drosophila oocyte. Developntent IO?: 13-9. Stanojevic, D., S. Small, and M. Levine. 199 1. Regulation of a segmentation stripe by overlapping activators and repressors in the Drosophila embryo. Science 254: 1385-7.

Stargell, LA- and K. Struhi. 1996. Mecbanisms of transcriptiona1 activation in vivo: hvo steps fonvard. Trends Genet 12: 3 11-5.

Steelman, S., J.J. Moskow, K- Muzynski, C. North, T. Druck, J.C. Montgomery, K. Huebner, 1.0. Dm,and A.M. Buchberg. 1997. Identification of a conserved family of MeisI-related homeobox genes [letter]- Genome Res 7: 142-56.

Steingrimsson, E., F. Pignoni, GJ. Liaw, and JA. Lengyel. 1991. Dual role of the Drosophila pattern gene tailless in embryonic termini, Science 251: 41 8-2 1.

Stem, S. and W. Hem. 1991. The herpes simplex virus trans-activator VP16 recognizes the Oct-I homeo domain: evidence for a homeo domain recognition subdomain. Genes Dev 5: 2555-66.

Stem, S., M. Tanaka, and W. Hem. 1989. The Oct-I homoeodomain directs formation of a multiprotein-DNA cornplex with the HSV transactivator VP 16. Na(zire 341: 634-30.

Stroeher, V.L., E.M. Jorgensen, and R.L. Garber. 1986. Multiple cranscripts from the Antennapedia gene of Drosophila melanogaster. Mol CeIl Biol6: 4667-75.

StmhI, G. 198 la. A gene product required for correct initiation of segmental determination in Drosophila. mure 293: 36-4 I .

Struhl, G. 1981b. A homoeotic mutation transforming leg to antenna in Dmsophila. iVature 292: 635-8.

Struhl, G. 1984. Splitting the birhorar complex of DrosophiIa. Narrrre 308: 45447.

Stmhl, G. 1985. Near-reciprocal phenotypes caused by inactivation or indiscriminate expression of the Drosophila segmentation gene ftz. iVuture 318: 677-80.

Stnihl. G. 1989- Differing snategies for organizing anterior and posterior body pattern in Drosophila embryos. Nature 338: 74 1-4.

StnihI, G., P. Johnston, and P.A. Lawrence. 1992. Control of Drosophila body pattern by the hunchback morphogen gradient. Ce11 69: 237-349.

Smihl, G., K- SmhL and P.M. Macdonaid, 1989. The gradient morphogen bicoid is a concentration- dependenr transcriptionai activator. Ce11 57: 1259-73.

Smihl, G. and R.A. White. 1985. Regulation of the Ultrabithonu gene of Drosophila by other bithorau cornplex genes. Cd1 43: 507-19.

Shrm, RA. and W. Hem 1988. The POU domain is a bipartite DNA-binding structure. Nature 336: 60 1-4.

Sun, B., DA. Hursh. D, Jackson, and PA. Beachy. 1995. UInabithorav protein is necessary but not suficient for fuI1 activation of decapentaplegic expression in the visceral mesodem. Embo J 14: 520-35.

Svendsen, P-C.and J.D. McGhee. 1995. The C. elegans neuronally expressed homeobox gene ceh-10 is cIosdy related to genes expressed in the vertebrate eye. Developmenr 121: 1253-62. Takemaru, K., F.Q. Li, H. Ueda, and S. Kuose. 1997. Multiprotein bridging factor 1 (MBFI)is an evolutionarily conserved transcriptional coactivator that connects a regdatory factor and TATA element-binding protein. Proc Nat1 Acad Sei U S A 94: 725 1-6.

Tamkun, J.W., R. Deuring, M.P. Scott, M. Kissinger, A.M. Pattatucci, T.C. Kauhan. and JA. Kennison. t992. brahma: a regulator of Drosophila homeotic genes stnicturdly related to the yeast transcriptional activator SNFYSWi2. Ce11 68: 561-72.

Tanaka, M., U. Grossniklaus, W. Herr, and N. Hernandez. 1988, Activation of the U2 snRNA promoter by the octamer motif defines a new class of RNA polymerase II enhancer elernents [pubIished erratum appears in Genes Dev 1989 Apr;2(4):584]. Genes Dev 2: 1764-78.

Tanaka, M, and W. Herr. 1990. Differential transcriptional activation by Oct-1 and Oct-2: interdependent activation domains induce Oct-2 phosphorylation. CeIl 60: 375-86.

Tanaka, M., J.S. Lai, and W. Herr. 1992. Promoter-selective activation domains in Oct-l and Oct-2 direct differential activation of an snRNA and mRNA prornoter. Cell68: 755-67.

Tautz. D. 1988. Regulation of the Drosophila segmentation gene hunchback by two maternai morphogenetic centres, Nature 332: 28 14.

Tautz. D. 1992. Redundancies, development and the flow of information. Bioessays 14: 763-6.

Tautz. D., R. Lehmann, R. Schnurch, E. Schuh, A. Seifert, K. Kieniin, K. Jones. and H. Jackle. 1987. Finger protein of novel structure encoded by hunchback, a second member of the gap class of Drosophila segmentation genes. Nature 327: 383-389.

TenHamsel, A.. R.J. Austin, N. Savenelli, and M.D. Biggin. 1993. Cooperative binding at a distance by even-skipped protein correlates with repression and suggests a mechanism of silencing. Mol Ce11 Bi01 13: 2742-52.

Thali, M.. M.M. Muller, M. DeLorenzi, P. Matthias, and M. Bienz. 1988. Drosophila homoeotic genes encode transcriptional activators similar to rnammalian OTF-2 [published erratum appears in Nature 1989 Jan 19;337(6204):290], Nature 336: 598-60 1.

Thummel, CS. 1992. Mechanisms of transcriptionat timing in Drosophila. Science 255: 39-40.

Thummel, C.S. 1995. From embryogenesis to metamorphosis: the regdation and function of Drosophila nuclear receptor superfarnily members. Cd83: 87 1-7.

Thurnmel, C.S. 1996. Files on steroids-Drosophila metamorphosis and the mechanisms of steroid hormone action. Trends Grnrr 12: 306-10.

Thuringer, F. and M. Bienz. 1993. Indirect autaregulation of a homeotic Drosophila gene rnediated by e.utracellular signaling. Proc lVad Acad Sei USA 90: 3899-903.

Thuringer. F., S.M. Cohen, and M. Bienz. 1993. Dissection of an indirect autoregulatary response of a homeotic Drosophila gene. Embo J 12: 2419-30.

Tiong, S., L.M. Bone, and S.R. Whittle. 1985. Recessive lethal mutations within the bithomu- complex in Drosophila, ,Mol Gen Genet 200: 33542.

Topol, J.. C.R. Dearolf, K. Pnkash, and C.S. Parker. 1991. Synthetic oligonucieotides recreate Drosophila hshi tarani zebra- smpe expression. Genes Dev 5: 855-67. Topol, J., G. Wiederrecht, and C.S. Parker. 1987. Biochemical Analysis of the hshi tarazu and U1trabithora.x Promoter Regions. In Genetic Regdarion of Developmenr, pp. 3-12. AIan R. Liss. Inc.

Treacy, MX., X. He, and M.G. Rosenfeld. 1991. 1-POU: a POU-domain protein that inhibits neuron-specific gene activation [see comments]. Nature 350: 577-84.

Treacy, M.N., L.I. Neilson, E.E. Turner, X. He, and M.G. Rosenfeld. 1992. Twin of 1-POU: a two amino acid difference in the [-POU homeodomain distinguishes an activator From an inhibitor of transcription. Cell68: 491-505.

Treisman, J., P. Gonczy, M. Vashishtha, E. Harris, and C. Desplan. 1989. A single amino acid can determine the DNA binding specificity of homeodomain proteins. CeII 59: 553-62.

Treisman, J., E. Harris. and C. Desplan. 199 1. The paired box encodes a second DNA-binding domain in the paired homeo domain protein. Genes Dev 5: 593-604.

Treisman, J., E. Harris, D. Wilson, and C. Desplan. L992. The homeodomain: a new face for the helix-turn-helix? Bioessqs 14: 145-50,

Tremml. G. and M. Bienz. 1992. Induction of labial expression in the Drosophila endoderm: response elements for dpp signalling and for autoregulation. Developmenr 116: 447-56.

Triezenberg, S.J. 1995. Structure and function of transcriptional activation domains, Curr Opin Genet Dev 5: 190-6.

Triezenberg, S.J., R.C. Kingsbury, and S.L. McKnight. 1988. Functional dissection of VP16, the trans-activator of herpes simpIex virus immediate early gene expression. Genes Dev 2: 7 18- 19.

Tsai, C. and J.P. Gergen. 1994. Gap gene properties of the pair-rule gene runt during Drosophila segmentation. Developmenr 120: 167 1-83,

Tsai, C. and P. Gergen. 1995. Pair-rule expression of the Drosophila fushi tarazu gene: a nuclear receptor response element mediates the opposing regdatory effects of runt and hairy. Developmenr 121 : 453-62.

Tsukiyama. T.. P.B. Becker, and C. Wu- l994. An-dependent nucleosome disruption at a heat- shock promoter mediated by binding of GAGA transcription factor [see comments]. ~Varure 367: 525-32.

Tsukiyarna, T.? C. Daniel, J. Tamkun, and C. Wu, 1995, ISWI, a member of the SWIUSNF2 ATPase family, encodes the 140 kDa subunit of the nucIeosome remodeling factor. Cd1 83: i021-6.

Tramarias, D. and K. Struhl. 1994. Functional dissection of the yeast Cyc8-Tupl transcriptional CO- repressor complex. ~Vature369: 758-6 1.

Ueda, H. and S. Hirose. 1990. identification and purification of a Bomb-yx mori homologüe of FTZ- FI. lVucleic rlcids Res 18: 7229-34.

Ueda, H., S. Sonoda, J.L. Brown, M.P- Scott, and C. Wu- 1990. A sequence-specific DNA-binding protein that activates hshi tarani segmentation gene expression. Genes Dev 1: 624-35.

Un, M., C. Li, and J.L. Manley. 1995. The transcriptional repressor even-skipped interacrs directly with TATA- binding protein. Mol CeII Bi01 15: 5007-16. Urness, L.D. and C.S. Thummel. 1990. Molecular interactions within the ecdysone regdatory hierarchy: DNA binding properties of the Drosophila ecdysone-inducible E71A protek. Ce11 63: 47-6 1.

Utley, KT., K. Ikeda, PA. Grant, J. Cote, DJ. Steger, A. Eberhaner, S. John, and J.L. Workman. 1998- Transcriptional activators direct histone acetyltransferase complexes to nucIeosomes. Nature 394: 198-502.

Vachon, G., B. Cohen, C. Pfeifle, M.E. McGufin, J. Botas, and S.M. Cohen. 1992. Horneotic genes of the Bithora complev repress limb development in the abdomen of the Drosophila embryo through the target gene Distal-less. CeIl 71: 437-50. van den Heuvel, M., R. Nusse, P. Johnson, and P.A. Lawrence. 1989. Distribution of the wingless gene product in Drosophila embryos: a protein involved in cell-cell communication. Celf59: 739-49. van Dijk, M.A. and C. Murre. 1994. extradenticle raises the DNA binding specificity of homeotic selector gene products. Ce11 78: 6 11-24. van Dijk, M.A.. L.T. Peltenburg, and C. Murre. 1995. Hox gene products modulate the DNA binding activity of Pbxl and Pbd. Mech Dev 52: 99-108.

Verrijzer, C.P.. A.J. Kal, and P.C. van der Vliet. 1990. The oct-1 homeo domain contacts only part of the octamer sequence and hl1 oct-1 DNA-binding activity requires the POU-specific domain. Genes Dev 1: 1964-74.

Vincent, I.P., J.A. Kassis, and P.H. O'Farrell. 1990. A synthetic homeodomain binding site acts as a cell type specific, promoter specific enhancer in Drosophila embyos. Embo J9: 3573-8.

Vincent, J.P. and P.H. O'Farrell. 1992. The state of engrailed expression is not clonally rransrnirted during early Drosophila development. Cefl68: 923-3 1.

Waddington, C.H. 1962. !Vmv Patterns in Generics and Devefopmenr. Columbia University Press, New York London.

Wagner-Bemholz, J.T., C. Wilson, G. Gibson, R, Schuh, and W.1- Gehring, 199 1. identification of target genes of the horneotic gene Antennapedia by enhancer detection [published errata appear in Genes Dev 1992 Feb;6(2):328 and 1992 Apr,6(4):690]. Genes Dev 5: 2467-80.

Wakimoto, B.T. and T.C. Kauhan. 1981. Analysis of larval segmentation in lethal genotypes associated with the antennapedia gene complex in DmsophiIa rneianogaster- Dev Biol81: 5 1- 64.

Wakimoto, B.T., F.R. Turner, and T.C. Kauhan. 1984. Defects in embryogenesis in mutants associated with the antennapedia gene complex of Drosophila melanogaster. Dev Biol 102: 147-72.

Walldorf. LI. and WJ. Gehring. 1992. Ernpty spiracles, a gap gene containing a homeobox invoived in Drosophila head developrnent. Embo J 11: 2247-59.

Walter, 1- and M.D. Biggin. 1996. DNA binding specificity of two homeodomain proteins in vitro and in Drosophila embryos. Proc Nat1 Acad Sci USA 93: 2680-5.

Walter, J., CA. Dever, and M.D. Biggin. 1994. Two homeo domain proteins bind with simiIar specificity to a wide range of DNA sites in Dmsophila embryos- Genes Dm 8: 1678-92. 204 Wang, B.B., M.M. Muller-Immergluck, I. Austin, N.T. Robinson, A. Chisholm, and C. Kenyon. 1993. A homeotic gene cluster patterns the anteroposterior body mis of C. elegans. Cell74: 29-42.

Wang, C. and R. Lehmann. 1991. Nanos is the Iocalized posterior determinant in Drosophila [published erratum appears in Ceil 1992 Mar 20;68(6):1177]. CelI66: 637-47.

Wedeen, C., K. Harding, and M. Levine. 1986. Spatial reguIation of Antennapedia and bithorrtv gene expression by the Poiycomb locus in Drosophila. Ce11 14: 739-48.

Weigel, D. and H. Jackle, 1990. The fork head domain: a novel DNA binding motif of eukaryotic transcription factors? [letter]. Cell63: 455-6.

Weigei, D., G. Jurgens, M. Klingler, and H. Jackle. 1990. Two gap genes rnediate materna1 terminal pattern information in Drosophila. Science 248: 495-8.

Weiner, A.J., M.P. Scott, and T.C. Kaufman. 1984. A molecular analysis of hshi tarazu, a gene in Drosophila melanogaster that encodes a product affecting embryonic segment number and ceil fate. Cell37: 843-5 1.

Weir, M.P. and T. Komberg. 1985. Patterns of engrailed and fushi tarazu transcripcs reveal novel intermediate stages in Drosophila segmentation. Nature 318: 433-9.

Wentworth, B.M., M. Donoghue, J-C. Engert, E.B, BergIund, and N. Rosenthal. 1991. Paired MyoD- binding sites regulate myosin light chah gene expression. Proc Nat1 Acad Sci USA88: 1242-6.

White, R.A. and R. Lehmann. 1986. A gap gene, hunchback, regulates the spatial expression of Ultrabithorax Cell-47: 3 1 1-2 1.

White, R.A. and M. Wilcox. 1985a. Distribution of UItrabithorax pmteins in Drosophila. Embo J4: 203 5-3043.

White, R.A. and M. Wilcox. 1985b. Regulation of the distribution of the Ultrabithorm proteins in Drosophila. Nature 318: 563-566.

Wieschaus, E., C. Nusslein-Volhard, and G. Jurgens. 1984. Mutations affecting the pattern pf the larval cuticle of Drosophila melanogaster. [II Zygotic loci on the X- chromosome and the fourth chromosome. Rourk Archives of Developmenral Biolog), 193: 296-307.

Wilcox, M. and R.J. Smith, 1980. Compartrnents and distal outgrowth in the Drosophila imagina1 wing disc. WiIhelm Roux Arch. Dev. Biol. 188: 157-161.

Winslow, G.M., S. Hayashi, M. Krasnow, D.S. Homess. and M.P. Scott. 1989. Transcri~tional activation by the Antennapedia and fushi proteins in cultured Drosophila Cells. CelI 57: 1017-30.

Wirz, J., L.1. Fessler, and W. Gehring. 1986. The Iocalization of the Antennapedia protein in Drosophila embryos and imaginai discs. Embo J5: 3327-3334.

Wodick- L.? H. Dong, M. Mittrnann, M.H. Ho, and D.J. Lockhart. 1997. Genome-wide expression monitoring in Saccharomyces cerevisiae. Nat Biotechnol 15: 1359-67.

Wolberger, C.? A.K. Vershon, B. Liu, AD. Johnson, and C.O. Pabo. 1991. Crystal structure of a MAT alpha 2 homeodomaùi-operator cornplex suggests a general mode1 for homeodomain- DNA interactions. CeII61: 5 17-28. Xiao, P. and J.P. Capone. 1990. A cellular factor binds to the herpes simplex vims type 1 transactivator Vmw65 and is required for Vmw65-dependent protein-DNA complex assembly with Oct-1. Mol Cell Biol 10: 4974-7.

Xue, D., M. Finney, G. Ruvkun, and M. Chalfie. 1992. Regdation of the mecJ gene by the C-elegans homeoproteins WC-86 and MEC-3, Embo J11: 4969-79.

Xue, D., Y. Tu, and M. Chalfie. 1993. Cooperative interactions between the Caenorhabditis elegans homeoproteins WC-86 and MEC-3. Science 261: 1324-8.

Yan, R., S. Small, C. Desplan, C.R. Dearolf, and J.E. Damell, Jr, 1996. Identification of a Stat gene that functions in Drosophila development. Cell84: 421-30.

Yost, H.J. and S. Lindquist. 1986. RNA splicing is intempted by heat shock and is rescued by heat shock protein synthesis. Cell15: 185-93.

Yu, Y., W. Li, K. Su, M. Yussa, W, Han, N. Pemmon, and L. Pick, 1997. The nuclear hormone receptor Ftz-FI is a cofactor for the Drosophila homeodomain protein Ftz. Nature 385: 532- 5.

Yu, Y. and L. Pick. 1995, Non-periodic cues generate seven Ftz stripes in the Drosophila embryo. Mech Dev 50: 163-75.

Yu, Y., M. Yussa, J. Song, J. Hirsch, and L. Pick. 1999. A double interaction screen identifies positive and negative ftz gene regulators and €tz-interacting proteins. Mech Dev 83: 95- 105.

Yuan, D., X. Ma, and 1. Ma. 1996. Sequences outside the homeodomain of bicoid are required for protein- protein interaction. J Biol Chem 271: 11660-5.

Zeng, C., J. Pinsonneault, G. Gelion, N. McGinnis, and W. McGinnis. 1994. Deformed protein binding sites and cofactor binding sites are required for the function of a srnall segment- specific regulatory element in Drosophila ernbryos. Embo J 13: 3362-77.

Zeng, W., D.J. Andrew, L.D. Mathies, M.A. Horner, and M.P. Scott, 1993. Ectopic expression and function of the Antp and Scr homeotic genes: the N terminus of the homeodomain is critical to functional specificity. Development 118: 39-52,

Zhang, C.C. and M. Bienz. 1992. Segmental determination in Drosophila conferred by hunchback - (hb), a repressor of the homeotic gene Ultrabithom. (Ubx). roc Nad Acad ci USA 89: 751 1-5.

Zhang, C.C.. J. Muller, M. Hoch, H. Jackle, and M. Bienz. 1991. Target sequences for hunchback in a control region conferring Ultrabithorax expression boundaries. Development 113: 1 171-9.

Zhao, J.J., R.A. Lanarini, and L. Pick. 1993. The rnouse Hox-If gene is fiinctionatly equivalent to the Drosophila Sex combs reduced gene Genes Dev 7: 343-54.

Zheng, L.B. and R. Losick. 1990. Cascade regulation of spore coat gene expression in Bacillus subtilis. J Mol Bi01 212: 645-60.

Zink, B. md R. Paro. 1989. In vivo binding pattern of a trans-regulator of homoeotic genes in Drosophila melanogaster. Nature 337: 468-7 1.

Zwilling, S., H. Konig, and T. Wirth. 1995. High mobility goup protein 2 iünctionally interacts wiih the POU domains of octamer transcription factors. Embo J 14: 1198-208.