REGULATION OF THE BMP SIGNALLING PATHWAY BY BMPl-RELATED METALLOPROTEASES

FIONA CLAIRE WARDLE

A Thesis Submitted for the Degree of Doctor of Philosophy at the University of London 1998

Department of Anatomy and Developmental Biology University College, London. ProQuest Number: 10016134

All rights reserved

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest.

ProQuest 10016134

Published by ProQuest LLC(2016). Copyright of the Dissertation is held by the Author.

All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. Microform Edition © ProQuest LLC.

ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 ABSTRACT

Bone Morphogenetic Proteins (BMPs) 2-8 are members of the TGFp superfamily of secreted signalling molecules. During BMPs are involved in many processes including cell fate determination, morphogenesis, growth and programmed cell death, all of which are essential for normal development.

BMP activity may be regulated in variety of ways. Of particular interest is the finding that three Xenopus proteins. , Chordin and are able to bind BMP4 and prevent it activating its receptor (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). Inhibition by Chordin can be alleviated by Xolloid, a BMP 1-related metalloprotease that cleaves Chordin (Piccolo et al., 1997). Similarly, in the Drosophila Dpp, a BMP2/4 homologue, is inhibited by the Chordin homologue, Sog, but activated by Tolloid, a BMP 1-related metalloprotease (Marqués et al., 1997). Since BMPl- related metalloproteases are present during development of many animal species, an attractive hypothesis is that release from inactive complexes by these metalloproteases is a general mechanism for regulating BMP activity.

Mesodermal patterning during Xenopus development, a process known to require BMP activity, is used as a test system to investigate the role of BMP 1-related metalloproteases in BMP signalling. Recent experiments have shown that overexpression of XBMPl and Xolloid partially ventralizes dorsal mesoderm (Goodman et al., 1998). In contrast, expression of the sea urchin BMP 1-related metalloprotease, SpAN, completely ventralizes dorsal mesoderm. This ventralizing activity requires a functional BMP signalling pathway, indicating that these metalloproteases may be activating endogenous BMPs in the Xenopus embryo. In addition, SpAN, XBMPl and Xolloid inhibit the activity of Chordin and SpAN is also inhibits Noggin. To further test the action of these metalloproteases, putative dominant-negative constructs were made, lacking the metalloprotease domain but retaining the protein-protein interaction domains. These truncated constructs dorsalize ventral mesoderm, consistent with the truncated protein binding its target, but being unable to process it. Finally, two sea urchin TGF(3s expressed during early sea urchin development, Univin and suBMP2/4, were tested in the Xenopus embryo and found to be functionally homologous to Vgl and BMP4 respectively, suggesting that similar molecules may act in both sea urchins and Xenopus to pattern the early embryo. TABLE OF CONTENTS

ABSTRACT...... 2 TABLE OF CONTENTS...... 3 FIGURES AND TABLES...... 8 ABBREVIATIONS...... 10 ACKNOWLEDGMENTS...... 13

Chapter 1: GENERAL INTRODUCTION 1.1 The Transforming Growth Factor (3 Family ...... 14 1.1.1 Sub-groups of the TGFp superfamily ...... 14 1.1.1.1 The TGFp subgroup ...... 14 l.l.l.ii The activin subgroup ...... 15 1.1.1.111 The nodal-related subgroup ...... 15 l.l.l.iv The 60A and Dpp subgroups ...... 15 1.1.1.v The Vgl subgroup ...... 16 1.1.2 The structure of TGFPs ...... 16 1.1.3 TGpp signalling pathways ...... 17 1.1.3.1 Receptors ...... 17 1.1.3.Ü Signal transduction ...... 18 1.1.4 Control of TGFp activity ...... 19 1.2 Zinc Metalloproteases ...... 20 1.2.1 The matrix metalloprotease family of zinc metalloproteases (matrixins)...... 21 1.2.2 The astacin family of zinc metalloproteases ...... 22 1.2.2.1 Domain structure of astacin metalloproteases ...... 22 1.2.2.Ü BM PI-related metalloproteases ...... 23 1.2.2.111 Activity of astacin family metalloproteases ...... 24 1.3 TGFps in Development ...... 24 1.3.1 The early development of Xenopus ...... 24 1.3.2 The three-signal model ...... 25 1.3.3 Mesoderm induction ...... 26 1.3.4 Dorsoventral patterning of mesoderm in Xenopus ...... 28 1.3.5 Dorsoventral patterning in the early Drosophila embryo ...... 30 1.4 Sea Urchin Development ...... 31 1.4.1 Early development of the sea urchin embryo ...... 32 1.4.2 Cell fate specification in the sea urchin embryo ...... 33 1.4.2.1 Fate maps ...... 33 1.4.2.Ü Inductive interactions ...... 33 1.5 Summary...... 36 Chapter 2: MATERIALS AND METHODS 2.1 Materials ...... 51 2.1.1 Vectors ...... 51 2.1.2 DNA constructs ...... 51 2.1.3 Solutions and media ...... 52 2.2 Methods ...... 60 2.2.1 Xenopus techniques ...... 60 2.2.1.1 Obtaining XeAio/JMj eggs ...... 60 2.2.1.Ü In vitro fertilization ...... 60 2.2.1.111 Dejelling eggs ...... 60 2.2.1.iv Microinjection of mRNA into Xenopus eggs ...... 60 2.2.1.v Culturing embryos ...... 61 2.2.1.vi Staging and scoring embryos ...... 61 2.2.1.VÜ Microdissection of embryos and dissociation of cells ...... 61 2.2.1 .viii Obtammg Xenopus oocytes ...... 61 2.2.1.ix Injection of mRNAs and culture of oocytes ...... 63 2.2.2 DNA preparation and analysis ...... 63 2.2.2.1 Miniprep DNA preparation ...... 63 2.2.2.Ü Midiprep DNA preparation ...... 63 2.2.2.111 Determining concentration of nucleic acids ...... 63 2.2.2.ÎV Restriction digests ...... 63 2.2.3 Gel electrophoresis ...... 64 2.2.3.1 Agarose gel electrophoresis ...... 64 2.2.3.Ü Acrylamide electrophoresis ...... 64 2.2.4 DNA cloning ...... 64 2.2.4.1 Preparation of vector DNA for cloning ...... 64 2.2.4.Ü Preparation of insert DNA ...... 65 2.2.4.111 Ligation of vector and insert DNA ...... 65 2.2.4.iv Transformation of competent E. Coli cells...... 65 2.2.4.V Analysis of transformants ...... 66 2.2.5 Polymerase Chain Reaction (PCR) ...... 66 2.2.6 DNA sequencing ...... 67 2.2.7 In vitro transcription ...... 67 2.2.8 Protein analysis ...... :...... 68 2.2.8.1 In vitro translation ...... 68 2.2.8.Ü Extraction of proteins from embryos and oocytes ...... 69 2.2.8.111 SDS polyacrylamide electrophoresis (SDS PAGE) ...... 69 2.2.8.iv Coomassie Blue staining ...... 69 2.2.8.V Western Blotting and ECL detection ...... 69 2.2.9 RNase Protection Analysis (RPA) ...... 70 2.2.9.1 Total RNA isolation form embryos and explants ...... 70 2.2.9.Ü Radioactive labelling of RNA probes ...... 70 2.2.9.111 Purification of RNA probes ...... 71 2.2.9.iv Hybridization ...... 71 2.2.9.V RNase digestion ...... 71 2.2.10 Whole Mount In Situ Hybridization (WISH) ...... 72 2.2.10.i. Digoxygenin labelling of RNA probes ...... 72 2.2.10.Ü Fixation of embryos and explants ...... 72 2.2.10.111 Prehybridization ...... 72 2.2.10.iv Hybridization ...... 73 2.2.10.V Immunohistochemistry ...... 73 2.2.11 Histology ...... 74 2.2.11.1 Fixation of embryos and explants ...... 74 2.2.1 l.ii Staining and embedding ...... 74 2.2.11.111 Sectioning ...... 74 2.2.1 l.iv Counterstaining, dewaxing and mounting ...... 74 2.2.12 Photography ...... 75

Chapter 3: OVEREXPRESSION OF A SEA URCHIN BMPl-RELATED METALLOPROTEASE, SpAN, IN XENOPUS EMBRYOS 3.1 Introduction ...... 76 3.1.1 A conserved system for dorsoventral patterning ...... 76 3.1.2 BMP 1-related metalloproteases in Xenopus development ...... 77 3.1.3 BMP 1-related metalloproteases and BMP signalling in sea urchin embryos ...... 77 3.2 Results ...... 79 3.2.1 Injection of mRNA for the sea urchin metalloprotease, SpAN, suppresses dorsal differentiation in Xenopus embryos ...... 79 3.2.2 Dorsal injection of SpAN gives more extreme defects than ventral injection ...... 80 3.2.3 SpAN suppresses dorsal marker expression ...... 80 3.2.4 Injection of mRNA encoding the metalloprotease domain of SpAN disrupts morphogenetic movements ...... 81 3.2.5 SpAN ventralizes Xenopus embryos during gastrula stages ...... 82 3.2.5.1 Early dorsal markers ...... 82 3.2.5.Ü Early ventral markers...... 83 3.2.5.111 Other mesodermal markers ...... 83 3.2.6 SpAN blocks activin induced morphogenetic movements and dorsal mesoderm formation in isolated animal caps ...... 84 3.2.7 SpAN enhances the activity of BMP4 ...... 85 3.3 Discussion ...... 85 3.4.1 SpAN disrupts morphogenetic movements ...... 85 3.4.2 SpAN ventralizes Xenopus embryos ...... 87 3.4.3 How does SpAN ventralize dorsal mesoderm? ...... 87 3.4.4 Regulation of BMP signalling in sea urchins ...... 89 Chapter 4: EFFECT OF BMPl-RELATED METALLOPROTEASES ON EXTRACELLULAR MATRIX AND CELL-CELL INTERACTIONS 4.1 Introduction ...... 108 4.1.1 Cell and ECM interactions during embryogenesis ...... 108 4.1.2 ECM and growth factor interactions ...... 110 4.2 Results...... I l l 4.2.1 SpAN and SMP decrease fibronectin levels in Xenopus embryos ...... 111 4.2.2 XBMPl decreases fibronectin levels slightly, whilst Xolloid has no effect ...... 112 4.2.3 SpAN does not affect calcium dependent cell-cell adhesion ...... 112 4.3 Discussion ...... 113 4.3.1 Regulation of the ECM composition by growth factors ...... 113 4.3.2 Regulation of ECM composition by ...... 113 4.3.3 Is fibronectin required for movements? ...... 114 4.3.4 Cell adhesion is not affected by SpAN ...... 115 4.3.5 Does SpAN act on the ECM in sea urchins? ...... 115

Chapter 5: DO BMPl-RELATED METALLOPROTEASES ACT IN A BMP SIGNALLING PATHWAY? 5.1 Introduction ...... 118 5.1.1 Regulation of ventral signalling during Xenopus development ...... 118 5.2 Results...... 119 5.2.1 SpAN requires a functional BMP signalling pathway to ventralize Xenopus embryos ...... 119 5.2.2 SpAN suppresses dorsalization by Chordin and Noggin ...... 120 5.2.3 XBMPl and Xolloid require a functional BMP signalling pathway to ventralize Xenopus embryos ...... 120 5.2.4 XBMPl and Xolloid suppress dorsalization by Chordin, but not Noggin ...... 121 5.3 Discussion ...... 122 5.3.1 Requirement for a functional BMP signalling pathway ...... 122 5.3.2 A different specificity of SpAN and XBMPl/Xolloid for Chordin and Noggin ...... 122 5.3.3 Noggin and Chordin are very different proteins ...... 123

Chapter 6: THE ROLE OF THE C-TERMINAL DOMAIN OF BMPl- RELATED METALLOPROTEASES 6.1 Introduction ...... 127 6.1.1 Domain structure of BMP 1-related metalloproteases ...... 127 6.1.2 The role of EGF-like and CUB repeats in protein-protein interactions ...... 127 6.2 Results...... 129 6.2.1 Secretion and activity of myc-tagged XBMPl and Xolloid ...... 129 6.2.2 Secretion of XBMPl and Xolloid C-terminal domains ...... 129 6.2.3 BMPI Cub, XolloidCub and SpANCub dorsalize ventral mesoderm ...... 130 6.2.4 XBMPl Cub and SpANCub cause ectopic expression of dorsal ...... 131 6.2.5 XBMPl Cub induces expression of neural marker genes in isolated animal caps 132 6.2.6 XBMPl Cub rescues anterior stmctures in xbmpJ mRNA-injected embryos ...... 133

6 6.3 Discussion ...... 134 6.3.1 Full-length and C-terminal domain constructs of XBMPl and Xolloid are secreted...... 134 6.3.2 The activity of the C-terminal domain constructs is consistent with BMP signalling being inhibited ...... 134 6.3.3 XBMPl Cub has greater activity than XldCub or SpANCub ...... 136

Chapter 7: CHARACTERIZATION OF UNIVIN AND suBMP2/4, TWO TGFP FAMILY MEMBERS FROM SEA URCHINS 7.1 Introduction ...... 149 7.1.1 TGPp-related genes in sea urchins ...... 149 7.2 Results ...... 151 7.2.1 Sea urchin BMP2/4 ventralizes Xenopus embryos ...... 151 7.2.2 Bunivin dorsalizes Xenopus embryos ...... 151 7.2.3 Bunivin induces dorsal mesoderm in isolated animal caps ...... 152 7.3 Discussion ...... 153 7.3.1 Univin is a Vgl homologue and suBMP2/4 is a vertebrate BMP2/4 homologue 153 7.3.2 What is the role of univin and suBMP2/4 in sea urchin development? ...... 154 7.3.3 Does SpAN regulate suBMP2/4 and univin activity? ...... 155

Chapter 8: GENERAL DISCUSSION 8.1 Effects of overexpressing metalloproteases in the Xenopus em bryo ...... 163 8.2 Effects of inhibiting metalloprotease activity in the Xenopus em bryo ...... 166 8.3 The role of BMP 1-like metalloproteases and BMP signalling in sea urchin em b ry o s ...... 167 8.4 A model for the action of metalloproteases in the BMP signalling pathway 169

REFERENCES...... 175 APPENDIX: BMPl-Related Metalloproteases Promote the Development of Ventral Mesoderm in Early Xenopus Em hryos ...... i FIGURES AND TABLES

Figure 1.1 Phylogenetic tree illustrating relationships between different members of the TGFp family ...... 39 Figure 1.2 (A) Domain structure of TGFp superfamily members ...... 41 (B) TGFp signal transduction by Smads ...... 41 Figure 1.3 Families of zinc metalloproteases based on the amino acid sequences around the zinc ...... 42 Figure 1.4 Domain structure of BMP 1-related metalloproteases ...... 44 Figure 1.5 Phylogenetic tree demonstrating relatedness of BMP 1-like metalloproteases based on similarities in the metalloprotease domain ...... 45 Figure 1.6 Known cleavage sites of astacin metalloproteases ...... 46 Figure 1.7 (A) Fate map of X en o p u s gastrula ...... 48 (B) The three-signal model ...... 48 Figure 1.8 (A) Schematic diagram illustrating stages of sea urchin development ...... 50 (B) Fate map of a 64 cell embryo ...... 50 Figure 2.1 Dorsoanterior index ...... 62 Figure 3.1 SpAN exerts its effect during later gastrulation ...... 95 Figure 3.2 SpAN suppresses differentiation of dorsal structures ...... 97 Figure 3.3 Dorsal injection of SpAN mRNA causes more severe anterior truncations than ventral injection ...... 97 Figure 3.4 SpAN suppresses expression of dorsal markers ...... 99 Figure 3.5 Injection of Smp mRNA disrupts morphogenetic movements at gastrulation but does not affect differentiation of dorsal structures ...... 101 Figure 3.6 SpAN ventralizes embryos during gastrulation ...... 103 Figure 3.7 SpAN alters the expression of dorsal and ventral mesodermal marker genes in late gastrulae ...... 105 Figure 3.8 SpAN alters the expression of the mesodermal marker genes, Xbra and Xpo, in late gastrulae and neurulae ...... 105 Figure 3.9 SpAN blocks activin induction of dorsal morphogenetic movements in isolated animal caps ...... 107 Figure 3.10 SpAN blocks activin induction of dorsal mesodermal markers, and elicits expression of ventral markers, in isolated animal caps ...... 107 Figure 4.1 SpAN, Smp and XBMPl cause a decrease in fibronectin levels ...... 117 Figure 4.2 SpAN does not affect calcium-dependent cell adhesion ...... 117 Figure 5.1 SpAN, XBMPl and Xolloid act in the BMP signalling pathway to ventralize dorsal mesoderm ...... 126 Figure 6.1 XBMPl and Xolloid are secreted proteins ...... 142 Figure 6.2 XBMPl Cub and XldCub are secreted ...... 142 Figure 6.3 XBMPl Cub, XldCub and SpANCub ventralize dorsal mesoderm ...... 145 Figure 6.4 XBMPlCub and SpANCub cause ectopic expression of dorsal genes.... 145 Figure 6.5 XBMPl Cub induces neural tissue in isolated animal caps ...... 148 Figure 6.6 XBMPl Cub rescues anterior truncations caused by injection of xbm pl m R N A ...... 148 Figure 7.1 Alignment of the C-terminal (CT) regions of univin and suBMP2/4 with other closely related TGFp family members ...... 158 Figure 7.2 Injection of submp2/4 mRNA causes anterior truncations in Xenopus e m b ry o s ...... 160 Figure 7.3 Bunivin weakly dorsalizes Xenopus embryos ...... 160 Figure 7.4 Bunivin induces dorsal mesoderm and endoderm in isolated animal caps ...... 162 Figure 8.1 Inhibition and activation of BMP signalling ...... 172 Figure 8.2 (A) Chordin establishes a gradient of BMP4 activity in the dorsal marginal zone ...... 174 (B) Xolloid establishes a gradient of Chordin activity in the marginal zone ...... 174

Table 1.1 TGFp ligands and their known receptors ...... 18 Table 2.1 Sequence of primers used to generate the restriction sites for fusion c o n s tru c ts ...... 66 Table 3.1 Injection of different amounts of SpAN mRNA causes a graded series of dorsoanterior defects ...... 90 Table 3.2 Dorsal injection of SpAN m R N A causes more extreme phenotypes than ventral injection ...... 91 Table 3.3 Injection of different amounts of Smp mRNA causes a graded series of dorsoanterior defects ...... 92 Table 3.4 Coinjection of SpAN mRNA with xbmp4 mRNA causes an increase in D A I s c o re ...... 93 Table 6.1 Injection of myc tag xbm pl and xolloid mRNAs causes anterior truncations ...... 138 Table 6.2 SpANCub, XBMPlCub and XldCub induce dorsal differentiation in ventral marginal zones ...... 139 Table 6.3 Injection of xbm pl cub mRNA rescues anterior structures in xbm pl mRNA-injected embryos ...... 140 Table 7.1 Injection of submp2/4 mRNA into Xenopus embryos causes anterior truncations ...... 156 Table 7.2 Injection of bunivin mRNA weakly dorsalizes Xenopus em bryos 157

9 ABBREVIATIONS

ADMP Anti-Dorsalizing Morphogenetic Protein AmOAc Ammonium Acetate ATP Adenosine triphosphate BMP Bone Morphogenetic Protein bp base pair BSA Bovine Serum Albumin BuOH Butanol cDNA complementary DNA CHO Chinese Hamster Ovary cpm counts per minute CTP Cytidine triphosphate CUB Complement Clr/Cls; sea urchin Uegf; BMPl DMZ Dorsal marginal zone DNase Deoxyribonuclease DNA Deoxyribonucleic acid EDTA Fthylenediaminetetraacetic acid EOF Epidermal Growth Factor EtOH Ethanol FGF Fibroblast Growth Factor GDF Growth and Differentiation Factor GTP Guanosine triphosphate HCI Hydrochloric acid HRP Horseradish peroxidase ITP Inosine triphosphate IMS Industrial methylated spirits IPA Isopropyl alcohol (propan-2-ol) MAB Maleic Acid Buffer MBS Modified Barth's Saline mRNA messenger RNA NaCl Sodium Chloride NaOAc Sodium Acetate NaOH Sodium Hydroxide NTP Nucleotide triphosphate ODC Ornithine decarboxylase PBS Phosphate Buffered Saline PCR Polymerase Chain Reaction PDGF Platelet-Derived Growth Factor PFA Paraformaldehyde

10 PMSF Phenylmethylsulphonylfluoride RNA Ribonucleic acid RNase Ribonuclease RPA RNase protection assay RT Room Temperature SDS Sodium Dodecyl Sulphate tBR truncated type IBMP2/4 receptor TBS Tris Buffered Saline TGFP Transforming growth factor beta tRNA transfer RNA TTP Thymidine triphosphate UTP Uridine triphosphate UTR Untranslated region UV Ultraviolet VMZ Ventral marginal zone WISH Whole mount in situ hybridization

Unit abbreviations “C degrees Celsius Da Dalton g gramme 1 litre min(s) minute(s) M Molar concentration pH -log concentration psi pounds per square inch rpm revolutions per minute sec(s) second(s) V Volt v/v volume per unit volume w/v weight per unit volume

Prefixes used in International System of Units k kilo (10') m milli (10 ') 1^ micro (10^) n nano (10'^) P pico (10 '^)

11 Symbols used for Amino Acids A Ala Alanine C Cys Cysteine D Asp Aspartic Acid E Glu Glutamic Acid F Phe Phenylalanine G Gly Glycine H His Histidine I Ile Isoleucine K Lys Lysine L Leu Leucine M Met Methionine N Asn Asparagine P Pro Proline Q Gin Glutamine R Arg Arginine S Ser Serine T Thr Threonine V Val Valine W Tip Tryptophan Y Tyr Tyrosine

12 ACKNOWLEDGMENTS

I would like to thank my supervisor, Dr Les Dale, for his encouragement, help and advice throughout, and for always having time.

Many thanks to Dr Shelley Goodman for putting up with me, her technical advice on all things molecular and for making several of the constructs used in this study.

Finally, thanks to Dr Duncan Harding for his support and useful advice, especially in the final throes.

I gratefully acknowledge the support of the Wellcome Trust in funding this Ph.D.

13 Chapter 1

GENERAL INTRODUCTION

1.1 The Transforming Growth Factor-p Family

The transforming growth factor (3 (TGFp) superfamily is a large family of secreted signalling molecules with diverse activities during development, normal growth, tissue repair and the immune response (reviewed by Massagué, 1990; Kingsley, 1994; Wall and Hogan, 1994)

1.1.1 Sub-groups of the TGFp superfamily

A pile up analysis of TGFP superfamily members shows the family can be subdivided into several different subgroups, depending on sequence similarity in the bioactive C-terminal domain (figure 1.1). The main sub-groups are the TGFps, the Activins, the Nodal-related subgroup, the 60A subgroup, the Dpp subgroup and the Vgl subgroup. Some workers group the Vgl, Dpp, 60A and Nodal-related subgroups, along with other proteins that fall between these groups (e.g. ADMP, GDF3 and Dorsalin), into a larger group named the Dpp-Vgl-related (DVR) group or the BMP superfamily (Lyons et al., 1991; Wall and Hogan, 1994; Hogan, 1996a). However, since many new members are being identified all the time, and proteins falling within this group have very different activities, this larger grouping is becoming increasingly cumbersome and indefinite. Other members of the TGFp superfamily (e.g. inhibin a) do not fit well into any specific subgroup.

1.1.1.1 The TGFp subgroup The prototype member of the TGFp superfamily is TGFp-1 and since its discovery four other related proteins (TGFps 2-5) have been cloned from many different species (reviewed in Massagué, 1990). TGFps are expressed at numerous sites during development and in adult tissues, and have activities which fall broadly into three main categories (Lin and Lodish, 1993): (i) effects on the cell cycle, (ii) effects on other growth factors and (iii) effects on the extracellular matrix (ECM). TGFps are able to stimulate the proliferation of some cells, whilst inhibiting proliferation of others (reviewed by Roberts and Sporn, 1992). In general TGFp interrupts the cell cycle, holding cells in a non-proliferative state, but where TGFp promotes cell growth it may do so by inducing the expression of growth- promoting factors, e.g. PDGF. TGFp also induces many cell types to increase production of ECM components, such as Fibronectin, Tenascin, Biglycan and Decorin, and up- regulate ECM receptors, such as integrins, on the cell surface (review: Massagué, 1990). In addition, TGFps play a role in immune response and inflammation; for instance, mice lacking the TGFpl gene die two to three weeks after birth from an inflammatory disease

14 which is probably due to an autoimmune response (Shull et al., 1992; Kulkami et al., 1993; Yaswen et al., 1996). However, only approximately 50% of embryos survive to birth, since many die in utero from defective heamatopoiesis and yolk sac vasculogenesis (Dickson et al., 1995; Bonyadi et al., 1997). TGFp may also have further roles in early development that are masked by maternal rescue of the embryo via the placenta (Letterio et al., 1994). l.l.l.ii The Activin subgroup Activins were originally isolated as the p subunits of inhibin, a component of ovarian follicular fluid that is able to inhibit the production of FSH from pituitary cells. Activins consist of homo- or hetero- dimers of p chains, of which five (Pa - Pe ) have been identified (Mason et al., 1985; Ling et al., 1986; Vale et al., 1986; Rotten et al., 1995; Lau et al., 1996; Fang et al., 1996; Oda et al., 1995), while inhibins consist of a heterodimer of a p chain with an a chain. Activin has been implicated in early development and mesoderm formation in Xenopus embryos (see later; reviewed by Smith, 1995). However, Activins are apparently not involved in early development and mesoderm formation in mouse embryos, since mice homozygous for a null in the Pa chain, the Pb chain, or both, form mesoderm normally. Instead Pa deficient mice have eyelid deformities and reduced fertility (Matzuk et al. 1995), while pg deficient mice die within 24 hours of birth with no whiskers, lower incisors and a cleft palate (Vassalli et al., 1994). Mice deficient in both Pa and Pb have the sum of these defects, suggesting that Activins cannot substitute for each other’s function (Matzuk et al. 1995). The function of Activins C and E have yet to be determined, but Activin D, which is expressed in the oocyte and early embryos of Xenopus, is able to induce mesoderm in Xenopus embryos (Oda et al., 1995). l.l.l.iii The Nodal-related subgroup nodal is a mouse gene which is expressed at the anterior tip of the primitive streak, surrounding the node (Zhou et al., 1993; Conlon et al., 1994). Mouse embryos homozygous for a null mutation in nodal contain little or no mesoderm, do not form a primitive streak and arrest early during gastrulation, indicating that Nodal is essential for mesoderm formation and patterning in the early mouse embryo (Conlon et al., 1991 ; 1994; lannaccone et al., 1992). Four Xenopus Nodal-related genes have been isolated, X n rl-4 , which may also play a role in mesodermal patterning in the early Xenopus embryo (Smith et al., 1995; Jones et al., 1995; Joseph and Melton, 1997). Nodal may also be involved in establishing the left-right axis (reviewed by Hogan, 1996a)

l.l.l.iv The 60A and Dpp subgroups The Drosophila gene, 60A, named after its chromosomal location, was isolated in a specific search for TGFp related genes in flies (Wharton et al., 1991 ; Doctor et al., 1992), although its function in development is not yet known. Other members of the 60A subgroup are: BMP5 (Celeste et al., 1990), BMP6 (Vgrl; Lyons et al., 1989; Celeste et al. , 1990), BMP7 (Osteogenic Protein 1 (OPl); Celeste et al., 1990; Ozkaynak et al., 1990) and BMP8 15 (OP2; Ozkaynak et al., 1992). Ail these genes are expressed at different times and sites during embryonic development, and are likely to be important in normal development. The ‘short ear’ mutant mouse, for example, has been shown to have a mutation in the BMP5 gene (Kingsley et al., 1992) that leads to skeletal and soft tissue defects such as short external ears, missing ribs and lung cysts (reviewed by Kingsley, 1994).

The Dpp subgroup includes BMPs 2 and 4 as well as Drosophila Decapentaplegic (Dp^. BMPs 1-4 were originally isolated from demineralized bone extracts that induce ectopic cartilage and endochondrial bone formation when implanted into rodents. When the cDNA sequences were determined, BMPl was found to be a metalloprotease of the astacin family (discussed later in section 1.2), whereas the others were related to TGFp family members (Wozney et al., 1988). BMP2 and BMP4 are required at many different stages during development of vertebrates (reviewed by Hogan 1996a, b), including dorsal-ventral patterning in early embryos which will be discussed in more detail later (see section 1.3.4). Like, BMPs 2 and 4, Dpp is necessary for several different processes during Drosophila embryogenesis, including dorsal/ventral patterning of the embryo, establishment of the proximo-distal axis of appendages, eye development and mid-gut morphogenesis (Spencer et al., 1982; Irish and Gelbert, 1987; Panganiban et al. 1990).

1.1.l.v The Vgl subgroup

Vg] is 2L Xenopus gene expressed at the vegetal pole of the oocyte and early embryo, and the mature protein has been implicated in mesoderm formation in Xenopus embryos (Weeks and Melton, 1987; Dale et al., 1993; Thomsen and Melton, 1993; see section 1.3.3). Other Vgl-like genes have been isolated from the zebrafish and chick, and the mature gene products have all been found to have the same mesoderm-inducing abilities when ectopically expressed in Xenopus embryos (Dorhmann et al., 1996; Seleiro et al., 1996). Univin (Stenzel et al., 1994), a sea urchin protein, also falls within this group, and evidence presented in this thesis suggests that sea urchin Univin has the same activity as Xenopus V gl.

1.1.2 The structure of TGFps

TGF-p superfamily members are synthesized as larger precursor molecules with an N- terminal hydrophobic secretory signal sequence, a proregion of varying size and a mature C-terminal domain (figure 1.2). The proregion, which is poorly conserved across different family members, appears to be required for the correct synthesis, folding and secretion of the protein (Gray and Mason, 1990). TGFp y are active only after cleavage of the proregion from the C-terminal domain, usually at an Arg-X-X-Arg site (where X is any amino acid. The proregion may also remain associated with the mature domain after cleavage to regulate activity of the mature domain, as in the case of TGFpl (Gentry and Nash, 1990; see below).

16 In contrast to the poorly conserved N-terminal domain, the 110-140 amino acid C-terminal bioactive domain is highly conserved between members of the TGF(3 superfamily. Most members have seven cysteine residues in the C-terminal domain at conserved locations, although TGFp 1-5 have nine, whilst GDF9 (McPherron and Lee, 1993) and Xnr3 (Smith et al., 1995) have only six. Crystallographic studies on TGFp2 (Daopin et al. 1992; Schlunegger and Grutter, 1992) and BMP7 (Griffith et al., 1996) have shown that all but one of these cysteine residues are involved in intrachain disulphide bonds which form a ‘cysteine knot’ (McDonald and Hendrickson, 1993). Due to the highly conserved cysteine structure of TGFp superfamily members it is assumed that they all adopt this structure. The cysteine knot consists of a ring held together by two disulphide bonds with another disulphide bond threaded through the middle, forming a strong core to the protein. The remaining cysteine residue forms an interchain disulphide bond with a second chain, to form the functional dimer. At present it is not clear whether TGFp ^ h ich have an even number of cysteine residues are able to form interchain disulphide bonds, although there are enough potential hydrogen bonds to allow dimerization in the absence of interchain cysteine residues.

1.1.3 TGFp signalling pathways l.l.S.i Receptors TGFp ^family members interact with heteromeric complexes of transmembrane serine/threonine kinase receptors (reviewed by Massagué, 1996; Lin and Lodish, 1993; Derynck, 1994; ten Dijke et al., 1996). Based on sequence similarity these receptors fall into two groups, type 1 and type 11 (table 1.1), and both receptors are required for normal TGFp activity (Wranaet al., 1994; Nellen et al., 1994; Letsou et al., 1995; Ruberte et al., 1995). Both type 1 and 11 receptors contain an extracellular domain that is cysteine rich, a transmembrane domain, and a cytoplasmic domain that contains the kinase. Type 1 receptors contain a conserved motif in the intracellular juxtamembrane region, the GS domain, so-called because it is rich in glycine and serine residues. As well as being structurally different, type 1 and type 11 receptors differ in their ligand binding properties. The type 11 receptors for TGFp and Activin can recognize and bind their ligand by themselves, whereas type 1 receptors must be associated with type 11 receptors for ligand binding (Wranaet al., 1992). This is in contrast to the BMP receptor system, where both type 1 and type 11 receptors have low affinity for the ligand individually, but high affinity when both receptors are present (Letsou et al., 1995; Rosenzweig et al., 1995). This suggests that the specificity of TGFg(signalling may be governed by the affinity of different receptors for varying ligand and ligand/receptor complexes. Once the ligand is bound, a signal is propagated by phosphorylation of the GS domain in the type 1 receptor by the constit itively active kinase domain of the type 11 receptor (Wrana et al., 1994; Wieser et al., 1995).

17 Another TGFp binding protein, receptor type III or Betaglycan (reviewed in Lin and Lodish, 1993), is expressed on most cells which respond to TGFp. Betaglycan is a large transmembrane proteoglycan with a short intracellular domain that has no known signalling motif. Cells expressing Betaglycan without type I or II receptors are unable to respond to TGFp, and so it does not seem to be directly involved in signal transduction; however, expression of Betaglycan in myoblasts increases the ability of the type II receptor to bind TGFpl, perhaps by presenting the ligand to the type II receptor.

Table 1.1 TGFp ligands and their known receptors

LIGAND TYPE 1 RECEPTOR TYPE 11 RECEPTOR D pp Thick veins Punt S a x o p h o n e B M P2/4 ActR-1 BMPR-ll BMPR-IA D af4 BMPR-IB BM P7 ActR-1 BMPR-ll BMPR-IA ActR-11 Activin ActR-1 ActR-11 ActR-lB ActR-llB TGFp TpR-1 TpR-ll

Adapted from Massagué, 1996 and ten Dijke et al., 1996.

1.1.3.Ü Signal transduction Until recently very little was known about pathways within the cell that are activated upon TGFp^ligand binding. A ras/raf signalling pathway and a MAPK pathway have both been implicated in the transduction to the nucleus of TGFp signals. BMP4 signalling in the Xenopus embryo can be blocked by expressing a dominant negative form of ras or raf, whilst coexpressing ras or raf reverses the effect of a dominant negative BMP4 receptor, indicating both can work downstream of the BMP4 receptor (Xu et al., 1996)., Yamaguchi et al. (1995) isolated TAKl, which is related but distinct from members of the MAPK family. Its kinase activity is stimulated by TGFp signalling and an increase in transcription of a reporter gene is seen in response to BMP and TGFp signalling.

More recently, a novel set of proteins have been identified that transduce TGF(^signals (Figure 1.2B; reviewed by Massagué, 1996; Derynck and Zhang, 1996; Wrana and Attisano, 1996; Whitman, 1997). In Drosophila a genetic screen for modifiers of Dpp activity identified Mothers against Dpp (Mad) as being required for Dpp activity (Raftery et al., 1995; Sekelsky et al., 1995). Sma 2, 3 and 4, which have similar mutant phenotypes to

18 those of the BMP receptor, daf 4, and act downstream of this receptor (Savage et al., 1996), were isolated from Caenorhabditis elegans and found to be structurally related to Mad. Since then many Smads have been isolated by homology from humans, mice and Xenopus.

Smads 1-5 are activated by binding of TGF(3^1igands to their receptors (Hoodless et al., 1996; Wiersdorff et al., 1996; Graff et al., 1996; reviewed by Whitman, 1997). Smad 1, and probably Smad 5, act mainly downstream of BMP receptors, whereas Smad 2 and Smad 3 act downstream of Activin and TGFp. Smad 9 has also been reported, which is most closely related to Smads 1 and 5, and is also likely to act downstream of BMP receptors (Watanabeet al., 1997). When the upstream receptors are stimulated by binding of the appropriate ligand, these pathway-specific Smads are directly phosphorylated at a C- terminal consensus site by the type I receptor (Hoodless et al., 1996; Macias-Silva et al., 1996; Zhang et al., 1996; Kretzschmar et al., 1997). Smad 4, in contrast, is not phosphorylated but forms a stable complex with phosphorylated Smads 1, 2, 3 and 5. This complex is then translocated to the nucleus where it activates transcription (Baker and Harland, 1996; Hoodless et al., 1996; Liu et al., 1996). Once in the nucleus, Chen et al. (1996; 1997) have shown that Smad 2 and Smad 4 forms a complex with FASTI (forkhead Activin-sensitive transcription factor), a novel forkhead/winged helix DNA binding protein, and binds the promoter of mix2 (figure 1.2B), an immediate early response gene to Activin induction of mesoderm in animal caps (Vize, 1996) The N-terminal domain of Drosophila Mad, however, directly binds DNA in the promoter of the vestigial gene, indicating that the C-terminus of Mad masks an N-terminal DNA binding domain (Kim et al. 1997). This binding site in vestigial is required for its correct activation by Dpp in the wing imaginai disc, so Mad, at least, may not require other, intermediary factors to bind DNA, although the C-terminal repression must still be overcome in some way.

Other Smads, Smad 6, Smad 7 and Smad 8 and Drosophila Daughters against Dpp (Dad), have recently been reported, which rather than transducing a TGFp^signal to the nucleus, inhibit this signalling (Imamura et al., 1997; Tsuneizumi et al., 1997; Nakao et al., 1997; Hayashi et al., 1997; Heldin et al., 1997). In addition, they are induced by TGFp signalling, suggesting there is a negative feedback loop that regulates TGFp activity. Both Smad 6 and Smad 7 interact with type I receptors and overexpression of these Smads inhibits ligand-stimulated phosphorylation of pathway specific Smads, suggesting that inhibitory Smads block signalling by preventing activation of pathway specific Smads (Whitman, 1997).

1.1.4 Control of TGFp activity.

One way to regulate TCFj^activity is to limit the amount of mature peptide cleaved from its proregion. An extreme example of this is provided by the Xenopus protein Vgl, where in oocytes and embryos detectable Vgl protein is in the larger, precursor form and mature

19 protein is not seen. Forms of Vgl in which the proregion has been replaced by the proregion of BMP2, BMP4 (BVgl) or Activin (AVgl) are, however, processed by the embryo and oocytes, and the resulting peptide is active in induction assays (Thomsen and Melton, 1993; Dale et al., 1993; Kessler and Melton, 1995).

Another regulatory mechanism involves the association of the mature TGFp peptide with other, inhibitory proteins. For instance, members of the TGFp subgroup are secreted from most cells in latent, inactive complexes that must be proteolytically cleaved in the extracellular space to be activated (reviewed in Taipale and Keski-Oja, 1997). The latent complex consists of the TGFp dimer non-covalently associated with its proregion (also a disulphide linked dimer), called TGFp latency associated protein (LAP). The latent TGFp complex can also associate with a latent TGFp binding protein (LTBP) of which four have been isolated to date. Latent TGFpl binds to the ECM via LTBPl (Taipale et al., 1994), thus providing a mechanism for sequestering inactive TGFps in the ECM.

The formation of latent complexes outside the cell may be a common control mechanism for all TGFp superfamily members. A secreted protein called Follistatin was identified originally as a factor that binds and inactivates Activin (Nakamura, et al., 1990) but has recently been shown to bind and inactivate BMP4 (Fainsod et al., 1997). In addition, two other secreted proteins. Noggin and Chordin, have been isolated from Xenopus embryos and shown to bind BMP4 and BMP2 (Piccolo et al., 1996; Zimmerman et al., 1996). The importance of this in embryonic development is discussed later (see section 1.3.4). Sifperfan^ y The activity of proteases may be important in regulating TGF^activity by releasing bound and inactive TGFps from the ECM or other inhibitory proteins. The results presented in this thesis and other recent studies (Blader et al., 1997; Piccolo et al., 1997; Marqués et al., 1997; Goodman et al., 1998) indicate that a family of zinc metalloproteases, related to BMPl, may regulate the activity of BMPs during embryonic development.

1.2 Zinc Metalloproteases

Zinc metalloproteases are a large family of proteases that coordinate a zinc atom essential for their activity. These have been classified on both sequence similarity and, where the crystal structure has been solved, on stmctural similarity (e.g. Jongeneel et al., 1989; Jiang and Bond 1992; Bode et al., 1992; Bode et al., 1993; Rawlings and Barrett, 1993). Hooper (1994) integrated much of the data and proposed four superfamilies, the zincins, inverzincins, carboxypeptidases and DD-carboxypeptidases, which are further subdivided according to the scheme in figure 1.3.

This review will concentrate on the matrix metalloprotease (matrixin) family and the astacin family of zinc metalloproteases, which have been implicated in regulation of growth factor

20 activity. Both the matrixins and the astacins are part of the metzincin subgroup of the zincin superfamily. The zincins are defined by the HEXXH zinc-binding motif, where the two histidine residues (bold) act as zinc ligands. The 'metzincins' possess a so called 'Met- turn', and a longer concensus zinc binding motif, HEXXHXXGXXH where the third histidine also acts as a zinc ligand. The 'Met-turn' is a conformational structure involving a methionine residue downstream of the , which bends the amino acid chain back on itself, thereby introducing the fourth zinc ligand (a tyrosine residue) to the active site (Bode et al., 1993; Stocker et al., 1995).

1.2.1 The matrix metalloprotease family of zinc metalloproteases (matrixins)

The matrix metalloproteases (MMPs; also known as matrixins) are key regulators of matrix degradation (reviewed by Birkedal-Hansen, 1995). They are found at low levels in adults but are up-regulated during normal morphogenesis, e.g. embryonic development, and in pathological tissue remodelling e.g. inflammation, tumour invasion and metastasis. Naturally occurring inhibitors (tissue inhibitors of matrix metalloproteases, TIMPs) are important in controlling MMP activity and tissue degradation in disease is often correlated with an imbalance of MMPs over TIMPs (reviewed by Docherty et al., 1992). As well as post-translational control by TIMPs, MMP expression is controlled at the transcriptional level. Many growth factors and cytokines induce expression of MMPs, e.g. Activin, Interleukin 1, Tumour Necrosis Factor-a (TNF- a), whilst TGF-p, in cooperation with other growth factors, represses transcription of MMPl (Interstitial ) and induces TIMPl expression (Docherty et al., 1992). Many of these factors are secreted by macrophages at the sites of tissue damage/inflammation allowing ECM degradation and tissue repair, and several MMPs (stromelysin 1, 2, 3 and 9, A and B and MTl- MMP) have been found at healing wound sites (Azar et al., 1996; Okada et al., 1997).

There are three main groups of MMPs, stromelysins, and , which between them are able to degrade all components of the ECM (reviewed by Overall, 1991; Docherty et al., 1992; Birkedal-Hansen, 1995). Stromelysins have a broad substrate specificity including proteoglycan, elastin, Fibronectin and basement membrane components. The specific collagenases cleave Collagen types I, II, III, VII, VIII, and X, while gelatinases degrade Elastin, Fibronectin, Collagen IV, V, VII and XI and Gelatins (denatured Collagen). Recently, MMP2 (Gelatinase A) was also shown to cleave Laminin 5, a component of basement membranes, thus allowing cell migration of epithelial cells (Giannelli et al., 1997). Other MMPs which do not belong to these groupings are matrilysin and membrane-bound metalloproteases (MT-MMPs; Sato et al., 1994).

MMPs, however, are not involved in cleavage of ECM components exclusively. McGeehan et al. (1994) and Gearing et al. (1994) showed that synthetic MMP inhibitors block processing of the TNF- a precursor, and subsequently that several purified MMPs are able

21 to cleave proTNF-a (Gearing et al., 1995). MMPs can also regulate the activity of other MMPs, for example MT-MMPs activate progelatinase A (proMMPZ) by cleavage of its proregion (Sato et al., 1997). In addition degradation of ECM components may release bound growth factors (reviewed in Taipale and Keski-Oja, 1997).

1.2.2 The astacin family of zinc metalloproteases

Astacin, a crayfish digestive , was the first characterized member of this family (Titani et al., 1987; Dumermuth et al., 1991). Since then many more members have been identified from species ranging from hydra to humans, and in both adult and developmental systems. Recently, another member of the family has been isolated from the bacterium Flavobacterium meningosepticum (flavastacin; Tarentino et al., 1995), showing that the astacin family is very ancient and was present before the divergence of prokaryotes and eukaryotes.

1.2.2.i Domain structure of astacin metalloproteases Astacin family members are characterized by the 18 amino acid sequence HEXXHXXGFXHBXXRXDR, within the 200 amino acid metalloprotease domain, which coordinates the zinc atom required for catalytic activity (reviewed by Bond and Beynon, 1995; Stocker and Zwilling, 1995). In astacins the zinc atom is penta-coordinated by three histidine residues (H above), a water molecule (which is itself coordinated to a conserved glutamic acid residue, E above) and a tyrosine residue (Y), within the consensus sequence SXMHY (Bode et al., 1993). This sequence also contains the methionine involved in the 'Met-turn'. Sequence comparison of family members shows that 28 other residues are totally conserved within the metalloprotease domain, and these are probably crucial to the overall conformation of the protein (Bond and Beynon, 1995). Astacin family members are synthesized as larger precursor molecules with an N-terminal hydrophobic sec retory signal, a proregion, the metalloprotease domain and usually further domain C- terminal to this (figure 1.4). The proregion is probably cleaved to release the mature peptide by subtilsin-like serine proteases, such as furin (Barr, 1991), since a consensus cleavage site (Arg-X-X-Arg) for these proteases is present at the C-terminus of the proregion. Removal of the proregion is probably required for activation of the proenzyme since the crystal structure of astacin shows that the N-terminus of the mature domain folds back and forms a salt bridge with a Glutamine residue near the active site of the metalloprotease domain (Bode et al., 1992; Stocker et al., 1993). The presence of the proregion would prevent formation of this salt bridge, thereby blocking activity (Stocker et al., 1993).

The astacin family can be sub-divided into smaller groups depending on sequence similarities in the metalloprotease domain and the structure of additional C-terminal domains. Broadly speaking there are three sub-groups; the BMPl-related metalloproteases which have arrangements of 'CUB' and 'EGF-like' domains in the C-terminus (see

22 below), the meprins which have 'MAM' dimerization domains in the C-terminus, and those proteins, including astacin, which have no extra domains C-terminal to the metalloprotease domain.

EGF (Epidermal Growth Factor)-like repeats and CUB repeats are non-catalytic domains found in many different groups of proteins apart from BMPl-like metalloproteases (Appella et al., 1988; Bork and Beckmann, 1993). EGF-like repeats consist of approximately 40 amino acids, including 6 highly conserved cysteine residues that are likely to be involved in intradomain disulphide bonding (Appella et al., 1988). In addition, these EGF-like repeats contain a consensus sequence (D/N, D/N, D/N, Y/F, E) that can form a high affinity calcium binding site (Rees et al., 1988). CUB repeats were first identified in the complement subcomponents C lr and Cls (Leytus et al., 1986; Tosi et al., 1987), and later in an embryonic sea urchin protein, Uegf (Delgadillo-Reynoso et al., 1989) and human BMPl (Wozney et al., 1988). These repeats consist of approximately 110 amino acids including 4 conserved cysteine residues which form two intrachain disulphide bonds (Einspanier et al., 1994) and it has been suggested that CUB repeats form anti-parallel p barrels similar to those found in immunoglobulins (Bork and Beckmann, 1993). Both EGF-like and CUB repeats have been implicated in protein-protein and protein-substrate interactions (reviewed by Appella et al., 1988; Bond and Beynon, 1995). For instance, in the complement proteases (Clr and Cls) there is evidence that EGF-like and CUB repeats mediate the tetramization of two C lr and Cls dimers and the association of this complex with Clq to form the mature Cl molecule (Arlaud et al., 1987; Thielens et al., 1990)

1.2.2.1! BMPl-related metalloproteases BMPl is a short splice variant of a single gene found in humans, mice, Xenopus and sea urchins (Wozney et al., 1988; Maeno et al., 1993; Takahara et al., 1994; Hwang et al., 1994; Goodman et al., 1998), and has a CUB-CUB-EGF-CUB arrangement in its C- terminal domain. The longer splice variant (mammalian or Xenopus Tolloid), has an additional EGF-like and two additional CUB repeats at the C-terminal end (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997). This longer domain structure is the most common motif and is also found in the products of the Drosophila genes tolloid {tld\ Shimell et al., 1991) and tolloid-related (tlr-1; Nguyen et al., 1994; Finelli et al., 1995), the mouse gene mammalian tolloid-like {mtll\ Takahara et al., 1996), the Xenopus gene xolloid {xld, Goodman et al., 1998) and the Aplysia gene apTBL-1 (Liu et al., 1997). The sea urchin proteins SpAN and BP 10 are more distantly related to BMPl, having an EGF-like repeat, rather than a CUB repeat, immediately following the metalloprotease domain. Following the EGF-like repeat are two CUB repeats which sandwich a threonine-rich domain, consisting of four repeats of the sequence STTTLQTT in SpAN (Reynolds et al., 1992) and four repeats of TTT in BP 10 (LePage et al., 1992). Pile up analysis of BMPl- related metalloproteases, based on similarities in the metalloproteases domain (figure 1.5),

23 indicates several different branches in the family, with SpAN, BP 10, and a C. elegans hatching enzyme (hchl; Hishida et al., 1996) being the most highly diverged members.

1.2.2.iii Activity of astacin family metalloproteases Very little is known about the activity of individual members of the astacin family, although recent studies have begun to identify substrates for some members. Astacin itself is secreted into the stomach-like cardia of the crayfish where it digests Collagens. Stocker and Zwilling (1995) have analyzed the preferred substrate cleavage sites of astacin in synthetic peptides and found that many different sites are recognized, with certain preferences (see figure 1.6). Recently, BMPl was found to be the same as pro-Collagen-C-Proteinase (PCP; Kessler et al., 1996; Li et al., 1996), an activity isolated in the 1980’s which cleaves the C- terminal peptide from pro-Collagens I, II and III (Hojima et al., 1985; Kessler et al., 1986). Since its discovery, PCP/BMPl has also been found to cleave pro-Lysyl Oxidase (Panchenko et al., 1996), an enzyme involved in crosslinking Collagen fibres, and pro- Laminin y2 (Amano et al., 1997). The known cleavage sites recognized by BMPl (both long and short splice variants) are shown in figure 1.6. More recently, the Drosophila BMPl-related protein Tolloid was shown to cleave Short gastrulation (Sog) in vitro (Marqués et al., 1997), whilst a related Xenopus protein, Xolloid, cleaves Chordin, a Sog homologue (Piccolo et al., 1997). However, the exact cleavage sites of these metalloproteases are not known. Xolloid is thought to cut Chordin within two cysteine-rich domains, one domain at the N-terminus (CHI) and one near the C-terminus (CH3); whilst there are no obvious cleavage sites in CHI, based on BMPl/astacin preferences, within the CH3 domain is a Gly-Asp motif which may be a potential Xolloid cleavage site, if it is able to recognize sites similar to BMPl.

1.3 TGFps in Development

S o p S r - f t l AjU Uy /Vj&Ait.436*3 TGppj/are found during embryonic development in many different species, from flies to humans. For example, TGFps are expressed in the oocyte, during early embryonic stages and throughout development at many sites, such as in the neurectoderm of ascidians (Miya et al., 1996), in the imaginai discs of Drosophila, and in the condensing bone in vertebrates (reviewed by Kingsley, 1994; Hogan, 1996b). This description of the role of TGFp family members in development concentrates on early inductive interactions in the embryo of the frog, Xenopus laevis, although other embryos and systems are discussed where important information has been obtained.

1.3.1 The early development of Xenopus embryos.

During its development in the ovary, the Xenopus egg becomes polarized along its animal- vegetal axis, which is most obviously seen by its darkly pigmented animal hemisphere and lightly coloured vegetal hemisphere. The second axis (dorsal-ventral) is set up at

24 fertilization when sperm entry breaks the radial symmetry of the egg and causes an event known as ‘cortical rotation’, when the plasma membrane of the egg, plus a thin shell of underlying cytoplasm (the cortex), shifts relative to the inner cytoplasm. The point opposite the site of sperm entry, where the vegetal cortex meets the animal cytoplasm, will become the future dorsal side of the embryo (Vincent et al., 1986). Cortical rotation is essential for the formation of the dorsal axis; when cortical rotation is prevented, for example by UV irradiating the fertilized egg during the first cell cycle, the embryo develops with no dorsal structures (e.g. Gerhart et al., 1989).

As the embryo goes through cleavage divisions, a cavity, the blastocoel, forms in the animal hemisphere, and the embryo becomes known as a bias tula. During bias tula stages the cell cycle lengthens and zygotic transcription starts, an event known as the ‘mid-blastula transition’ (MET). By late blastula stages the 3 definitive germ layers (endoderm, mesoderm and ectoderm) have been specified and the next major developmental phase, gastrulation, brings these layers into their correct positions with respect to each other, setting up the basic body plan of the embryo, including the anterior-posterior axis. Gastrulation in the Xenopus embryo is initiated on the dorsal side, just below the equator (reviewed by Gerhart and Keller, 1986), when epithelial cells, ‘bottle cells’, contract at their apices and elongate basally (becoming bottle shaped). The apical constriction causes pigment granules to become concentrated in a small area, so that a line of black pigment is seen externally which marks the dorsal blastopore lip. Cells of the involuting marginal zone (IMZ), move through the blastopore lip, whilst cells of the deep zone, which lay internal to the IMZ, spread along the blastocoel wall and roof, leading the involution of IMZ cells. As gastrulation proceeds the blastopore extends laterally and ventrally, and eventually becomes a complete circle. As cells involute through the blastopore they converge on the dorsal midline and extend towards the animal pole, a process known as convergent extension. At the same time yolky vegetal cells are rolled inwards towards the prospective ventral side of the embryo, as the blastopore closes over its surface. Although is initiated, and is more extensive, on the dorsal side of the embryo, invagination occurs all round the blastopore, so that by the end of gastmlation the yolky endoderm has become internalized, the ectoderm surrounds the outside of the embryo, and the mesoderm is sandwiched between these two layers (figure 1.7A). The next step in the development of the Xenopus embryo is neurulation, when the neural plate (dorsal ectoderm) rolls up to form the neural tube and the mesoderm adjacent to the notochord, which underlies the neural tube, begins to segment to form somites. At this stage too, terminal differentiation of tissues has already begun, although histological differences are not really detectable until late tailbud stages (shown schematically in figure 1.7A).

1.3.2 The three-signal model

At early cleavage stages the Xenopus embryos consists of two cell types, prospective ectoderm in the animal hemisphere and prospective endoderm in the vegetal hemisphere. By 25 blastula stages, however, mesoderm has been specified in the marginal zone by a process known as mesoderm induction (reviewed by Slack, 1991). Mesoderm induction can be mimicked by juxtaposing vegetal pole explants with animal pole explants (animal caps). When cultured alone animal caps will form , however, when cultured with vegetal explants, mesoderm is induced in the animal cap, with the vegetal pole releasing mesoderm inducing signals as early as the 32-64 cell stage (Jones and Woodland, 1987). Moreover, ventral-vegetal tissue induces ventral-type mesoderm while dorsal-vegetal tissue induces dorsal mesoderm in the animal cap (Dale et al., 1985). Isolation experiments during blastula stages show that ventral-type mesoderm (blood, mesenchyme) is induced throughout the ventral and lateral marginal zone, however, fate map studies show that cells of the lateral marginal zone give rise to muscle in the whole embryo (Dale and Slack, 1987a). This indicates that after initial induction, the mesoderm is further patterned into a wider range of mesodermal subtypes. This patterning process, known as dorsalization, can be demonstrated by culturing lineage-labelled ventral marginal zone explants adjacent to dorsal marginal zone explants, when muscle is seen to form in the labelled ventral tissue (Dale and Slack, 1987b). A simple model to explain these observations is the ‘three signal model’ (figure 1.7B) proposed by Slack and colleagues in the early 1980’s (reviewed by Slack, 1991). This model, with some modifications, can still be used to describe mesoderm formation and patterning in the Xenopus embryo.

As a modification to the three-signal model, Kimelman et al. (1992) proposed that a ‘competence prepatterning signal’ (probably a Wnt-like signal) is activated on the dorsal side of the embryo as a result of sperm entry and cortical rotation. Instead of there being a separate dorsal-inducing and ventral-inducing signal, this prepattern would synergize with a ventral mesoderm inducing signal to induce dorsal-type mesoderm on the dorsal side of the embryo.

In addition, there is much evidence to indicate that a fourth signal is also responsible for patterning mesoderm during gastmlation (reviewed by Sive, 1993). This ventralizing signal, released from cells of the ventral and lateral marginal zone, antagonizes the dorsal signal emanating from the organizer. Over the last two years it has become clear that at least part of the dorsalizing signal from the organizer is involved in blocking this ventral signal, and this is discussed in greater detail later (see section 1.3.4).

1.3.3 Mesoderm induction

It is still unclear what signalling molecules are responsible for induction of mesoderm in the marginal zone of Xenopus embryos, however, members of the TGF-P family are strongly implicated in this process.

One assay for mesoderm inducing activity is to add the putative inducer to isolated animal caps, to test its ability to mimic the effect of vegetal pole tissue and induce mesoderm.

26 Activin, a TGFp superfamily member, is able to induce mesoderm in animal caps in this way and different concentrations induce different types of mesoderm (Green and Smith, 1990), indicating that Activin may be an endogenous mesoderm inducing factor. Although mRNA for Activin B is not present in the embryo until MET, and Activin A is not transcribed until late gastrula stages (Thomsen et ah, 1990), Activin protein is present in the egg, possibly supplied by maternal follicle cells (Dorhmann et ah, 1993; Rebagliati and Dawid, 1993). In addition, Activin D has been isolated recently, with transcripts for this gene present maternally and in the early embryo, and may therefore be a candidate for an endogenous mesoderm inducing factor (Oda et ah, 1995). More evidence that Activin or an Activin-like molecule is responsible for mesoderm induction comes from experiments which block Activin-like signalling. Firstly, when dominant-negative Activin receptors (ActRIIB and ALK4) which have a deleted kinase domain, and consequently can not propagate a signal, are overexpressed in the Xenopus embryo, little or no mesodermal tissue differentiates (Hemmati-Brivanlou and Melton, 1992; Chang et ah, 1997). Secondly, mesoderm formation can also be blocked by injection of Smad 7 or a dominant-negative form of Smad 4, both of which block intracellular signalling by Activin (Macias-Silva et ah, 1996; Nakao et ah, 1997), However, the receptors are not specific for Activin and inhibit other TGFp superfamily members, such as Vgl and BMP4 (Schulte-Merker et ah, 1994; Kessler and Melton; 1995; Chang et ah, 1997), and Smad 4 and Smad 7 are general for all TGFp signalling pathways (reviewed by Heldin et ah, 1997), so a TGF(3 other than Activin may be responsible for mesoderm induction in the Xenopus embryo. The mesoderm inducing signal, however, is unlikely to be a BMP, since injection of a dominant-negative BMP receptor, or blocking BMP signalling using antisense RNA or dominant-negative ligands does not block mesoderm formation (Graff et ah, 1994; Suzuki et ah, 1994; Steinbeisser et ah, 1995; Hawley et ah, 1995).

Perhaps the best candidate for the endogenous mesoderm inducing signal is Vgl, which is expressed at the right time and place, i.e. the vegetal hemisphere of the oocyte and early blastula (Weeks and Melton, 1987). In addition, the mature protein is able to induce mesoderm in animal caps, rescue dorsal axial structures in UV irradiated embryos, and induce a secondary axis in whole embryos (Thomsen and Melton, 1993; Dale et ah, 1993; Kessler and Melton, 1995). Furthermore, Vgl signalling is blocked by the dominant- negative Activin receptor that also inhibits mesoderm formation (Schulte-Merker et ah, 1994; Kessler and Melton, 1995). However, Vgl protein is detected in the embryo only as an inactive, unprocessed precursor. Even when massively overexpressed, by injection of mRNA into the embryo, barely any processed, mature Vgl protein is detectable (Tannahill and Melton, 1989; Dale et ah, 1993). Since some members of the TGFp family are typically active at very low concentrations, there may be enough mature Vgl protein in the embryo to induce mesoderm while being below the detection levels of current methods. These levels would have to be regulated by an as yet unknown post-translational mechanism.

27 As described above, a competence factor present on the dorsal side of the early embryo may modify a general mesoderm inducing signal so that dorsal mesoderm is induced (Kimelman et al., 1992). Recently, very strong evidence has emerged that p-catenin is this competence modifier, p-catenin accumulates opposite the site of sperm entry during the first cell cycle and continues to accumulate throughout cleavage divisions (Larabell et al., 1997). Moreover, depletion of p-catenin, using antisense oligonucleotides, causes embryos to develop without dorsal structures (Heasman et al., 1994a). p-catenin is both a component of the cytoskeleton at the cell membrane and a signal transducing molecule that associates with transcription factors such as XTcfS (Molenaar et al., 1996), activating transcription of target genes in response to the Wnt family of secreted signalling molecules. One gene that can be activated in early Xenopus embryos by Wnt family members is goosecoid {gsc), a homeobox gene that is transcribed in the dorsal marginal zone after MET as an immediate early response to mesoderm induction (Cho et al., 1991). In the promoter of gsc are at least two elements, a distal element (DE) that responds specifically to a Vgl/Activin-like signal and a proximal element (PE) which responds to Wnt-like signalling (Watabe et al., 1995), and may therefore be a binding site for p-catenin/XTcf3. The DE is necessary and sufficient to respond to Activin induction, and when a reporter construct under the control of the DE is expressed in Xenopus embryos it is activated all over the marginal zone (Watabe et al., 1995). When the same reporter gene is under the control of both the DE and PE, activity is greater on the dorsal side of the embryo. Taken together these results indicate that a Vgl/Activin-like signal may induce mesoderm all over the marginal zone, and that a Wnt- like signal is responsible for modifying the response so that dorsal mesoderm is induced. However recent evidence suggests that Wnts may not be responsible for inducing dorsal mesoderm in the early Xenopus embryo. Injection of dominant-negative constructs for either XwntS or dishevelled, an intracellular component of the Wnt signalling pathway, has no effect on dorsal mesoderm formation in the early Xenopus embryo (Hoppler et al., 1996; Sokol, 1996). In contrast, a dominant-negative construct for GSK3p, a component of the Wnt signalling pathway responsible for inactivating p-catenin, induces a secondary dorsal axis in Xenopus embryo, while overexpression of GSK3p blocks dorsal mesoderm formation (He et al., 1995; Dominguez et al., 1995). While these results argue for a role for p-catenin in dorsal axial development, they argue against a role for the Wnt family in activating p-catenin on the dorsal side of the embryo.

1.3.4 Dorsoventral patterning of mesoderm in Xenopus

The three signal model implies that the ventral-lateral mesoderm plays no active role in patterning the mesoderm during gastmlation. However, experiments over the last five years have indicated that this is not the case, and that a ventralizing signal is released by cells of the ventral marginal zone. Active ventralization was suggested by the finding that injection of bmp4 mRNA into dorsal blastomeres of early cleavage stage Xenopus embryos results in severe anterior truncations and reduced or absent dorsal axial structures, such as somites,

28 notochord and neural tissue (Dale et al., 1992; Jones et al., 1992a). In addition, BMP4, which is expressed in the ventral and lateral marginal zone (Fainsod et al., 1994; Hemmati- Brivanlou and Thomsen, 1995), may act as a morphogen since different concentrations of bmp4 mRNA induce different types of mesoderm (Dosch et al., 1997). Whereas the highest concentration of bmp4 mRNA induces blood differentiation, intermediate doses give pronephros and low doses give muscle (Dosch et al., 1997). Ventralization of dorsal mesoderm by BMP4 occurs during gastmlation, rather than earlier, during mesoderm induction. This is indicated by the fact that transcription of dorsal genes, such as goosecoid, is normal in bmp4 mRNA-injected embryos at the start of gastmlation, but transcription is subsequently down-regulated (Jones et al., 1996b). In addition, bmp2 and bmp7 are expressed throughout development and have the same properties as BMP4 when overexpressed in embryos (Clement et al., 1995; Hawley et al., 1995; Hemmati-Brivanlou and Thomsen, 1995). It is possible that BMP4/BMP7 heterodimers are the ventralizing signal, since these heterodimers form readily and have more potent mesoderm inducing capacity than either homodimer (Suzuki et al., 1997). Further indication that BMPs are involved in ventralizing mesoderm comes from experiments that inhibit the activity of BMPs in Xenopus embryos, using either (i) injection of a dominant-negative receptor which binds BMP4 and BMP2 (Graff et al., 1994; Suzuki et al., 1994), or (ii) injection of antisense RNA directed against BMP4 (Steinbeisser et al., 1995), or (iii) injection of BMP4 and BMP7 cleavage mutants which dimerize with endogenous BMPs and prevent proper processing (Hawley et al., 1995). In all cases this leads to ectopic dorsal mesoderm formation, and indicates that disruption of BMP signalling is required for dorsal fates. This is also indicated by experiments that show three dorsalizing proteins secreted by the organizer. Noggin, Chordin and Follistatin, are able to bind BMP4, BMP2 and BMP4/BMP7 heterodimers, and thereby prevent them from interacting with their receptors (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997; Piccolo et al., 1997). Thus it seems that at least part of the dorsalizing signal emanating from the organizer is involved in actively preventing BMP4 signalling. In fact, inhibition of ventral signalling may be a general role of the organizer, since XwntS signalling from the ventral side of the Xenopus embryo is inhibited by Frzb, whose transcripts are expressed in the organizer. Injection oifrzb mRNA into Xenopus embryos causes mild hyperdorsalization and blocks lateral mesoderm formation (Leyns et al., 1997; Wang et al., 1997). Frzb shares homology with the cysteine-rich extracellular domain of a the Wnt receptor Frizzled, and is probably able to inhibit Wnt activity by competing with the cell-surface receptor for Wnt binding.

Another layer of regulation in mesodermal patterning has been revealed with the discovery that Xolloid, a Xenopus BMP 1-related metalloprotease, is able to cleave Chordin and release active BMP4/7 from a complex of Chordin and BMP4/7 (Piccolo et al., 1997). Thus while dorsal signals such as Chordin are binding BMPs in the extracellular space, and preventing interaction with their receptors, metalloproteases, such as Xolloid are able to release that inhibition and allow BMP signalling.

29 The organizer also expresses several TGF-p related genes. One is admp, which is related to BMPs and despite being expressed exclusively in the organizer is able to ventralize dorsal mesoderm (Moos et al., 1995); BMP2 and BMP? are also expressed in the organizer, as well as throughout the ventral-lateral mesoderm and animal cap. In addition, four genes related to mouse nodal, named Xnrl-4 (Smith et al., 1995; Jones et al., 1995; Joseph and Melton, 1997), have been isolated in Xenopus and are expressed in the organizer. Xnrl, Xnr2 and Xnr4 can induce mesoderm in animal caps and pattern ventral mesoderm to form dorsal tissues (Jones et al., 1995; Joseph and Melton, 1997). Xnr3 can also dorsalize ventral mesoderm when expressed in the ventral marginal zone, but does not induce mesoderm in animal caps (Smith et al., 1995). Unlike Xnrl, Xnr2 and Xnr4, however, Xnr3 also induces neural tissue in animal caps (see below), and its action is direct because it occurs in the absence of mesoderm induction (Hansen et al., 1997). In addition, Xnr3 inhibits mesoderm formation in animal caps by coinjected BMP4 but not Xnr2 (Hansen et al., 1997). These activities of Xnr3 suggest it interferes with BMP signalling, although the way it does this is not clear. Although a major role of the organizer is to prevent BMP signalling, dorsalization by Xnrl, Xnr2 and Xnr4, which do not block BMP signalling, suggests that another, active dorsalizing mechanism may be involved in organizer activity.

Antagonism between BMPs and inhibitory factors also occurs in the ectoderm and endoderm. Neurons differentiate from ectoderm in the absence of BMP signalling, for instance when BMP mature ligand synthesis or receptor activation is blocked (Graff et al., 1994; Suzuki et al., 1994; Hawley et al., 1995), or ectodermal cells are dispersed in single cell culture (Godsave and Slack, 1989; Wilson and Hemmati-Brivanlou, 1995), or when Noggin, Follistatin or Chordin are expressed in the ectoderm (Lamb et al., 1993; Hemmati- Brivanlou et al., 1994; S as ai et al., 1995). Re-supplying BMP4 ligand to dispersed cells or restoring BMP signalling prevents neuralization and maintains epidermal fates. Thus BMPs are involved in specifying epidermal fate, while inhibition of this activity leads to expression of neural marker genes. A recent set of experiments also point to a role for Noggin and Chordin in the induction of endodermal tissues. Sasai et al. (1996) showed that injection of noggin or chordin mRNA can induce endoderm in anima^. This induction can be inhibited by BMP4, suggesting that the same kind of ligand antagonism that occurs in mesoderm and ectoderm is operating in the endoderm.

1.3.5 Dorsoventral patterning in the early Drosophila embryo.

Cells of the Drosophila blastoderm embryo differentiate according to their dorsal-ventral position. Those nearest the dorsal midline become the extra-embryonic amnioserosa, dorsal-lateral cells differentiate as dorsal epidermis, ventral-lateral cells form the ventral neurogenic region, which becomes the ventral neural cord and ventral epidermis, while the ventral-most cells give rise to mesoderm. Genetic analysis in Drosophila has led to the independent identification of homologues of BMP4 and Chordin: Dpp and Sog respectively (Gelbert, 1989; Francois et al., 1994; Francois and Bier, 1995). The phenotypes caused by 30 loss-of-function in dpp and sog demonstrate that the activities of both are necessary for proper dorsoventral patterning in Drosophila, dpp, which is expressed in the dorsal 40% of the Drosophila embryo at the blastoderm stage (St. Johnston and Gelbert, 1987), is required for the production of all dorsal cell types and a gradient of Dpp activity specifies the type of tissue formed (Ferguson and Anderson, 1992a). For example, null mutations in dpp cause a loss of dorsal most tissues, the amnioserosa and dorsal epidermis, while overexpression of dpp in wild-type embryos causes expansion of the dorsal epidermis, or at higher concentrations, the amnioserosa (Ferguson and Anderson, 1992a). Conversely, sog is expressed in the ventral-lateral cells that give rise to the ventral neurogenic ectoderm and loss of Sog activity reduces, but does not eliminate, this tissue. dpp/sog double mutants have a phenotype identical to dpp single mutants, indicating that Sog acts upstream of Dpp. In addition, Holley et al. (1996) showed that Noggin is able to inhibit Dpp activity when expressed in Drosophila embryos, in a similar way to sog. By analogy with Chordin in the Xenopus embryo, Sog is likely to inhibit Dpp activity by binding Dpp and preventing interaction with its receptors; it is also probable that Noggin is able to recognize and bind Dpp, as it recognizes BMP4, thereby inhibiting Dpp activity. A Drosophila Noggin homologue has not yet been identified. dpp and sog are not the only zygotic genes involved in dorsoventral patterning in the early Drosophila embryo. Five other genes, screw, shrew, zerkniillt, twisted gastrulation and tolloid are also required for generation of the normal pattern in the dorsal 40% of the embryo (Ferguson and Anderson, 1992b). Embryos that lack any of these genes display varying degrees of ventralization, similar to dpp mutants, tolloid (tld) is a BMP-1 related metalloprotease expressed in the dorsal 40% of the embryo at the blastoderm stage, in the same pattern as dpp. By studying genetic interactions between tld and dpp Ferguson and Anderson (1992b) showed that Tolloid acts upstream of Dpp to enhance its activity. This is in contrast to Sog, which negatively regulates Dpp activity. Tolloid appears to enhance Dpp activity by relieving the inhibition of Sog, since Marqués et al. (1997) have demonstrated that Tolloid is able to cleave Sog in the presence of Dpp, similar to the way Xolloid cleaves Chordin. Thus metalloproteases have a role in maintaining the balance of ventralizing and dorsalizing activity in the both the Drosophila blastoderm and in the Xenopus gastrula.

1.4 Sea Urchin Development

The embryonic development of sea urchins was first described by Derbès ( 1847) and von Baer (1847; cited in Davidson 1989); many classical sea urchin experiments such as cell lineage tracing and grafting experiments are reviewed by Horstadius (1973).

Many different sea urchin species, which are highly diverged from each other, are used for experimental work. Despite this, their general development and many general mechanisms are similar between the species studied. There are of course differences and what applies to 31 one species will not always apply to another, which must be born in mind when reviewing experimental data. An example of a difference which has caused some confusion is the specification of the oral-aboral (ventral-dorsal) axis (reviewed by Jeffery, 1992). In some cases the first cleavage plane correlates with the oral-aboral axis (e.g. Lytechinus pictus), while in other species (e.g. Paracentrosus lividus) the oral-aboral axis is randomly oriented with respect to the first cleavage plane. However, since the molecular mechanisms underlying dorsal-ventral axis formation have yet to be elucidated, it may be that the underlying molecular mechanism is conserved in sea urchins despite the differences observed above.

1.4.1 Early development of the sea urchin emhryo

As with the Xenopus egg, the animal-vegetal axis is established before fertilization in the sea urchin egg. During early cleavage stages, divisions are not equal so that at the 16 cell stage the embryo consists of 4 small micromeres at the vegetal pole, overlying these are 4 large macromeres, and above these, in the animal tier, 8 mesomeres (figure 1.8A; reviewed by Slack, 1991). The mesomeres then divide to produce two tiers of cells (the anl and an2 tiers; Horstadius, 1973). In the vegetal half the macromeres divide twice to produce the vegl and veg2 tiers, and the micromeres divide unequally to form the micromeres (skeletogenic or primary mesenchyme) and the small micromeres. These early cleavage divisions are predictable and invariant, which has lead to suggestions that localized maternal determinants are a key feature of early sea urchin development and cell fate specification (reviewed by Davidson, 1989). However, recombination and grafting experiments clearly indicate that this is also an inductive system in which most of the cell fates remain plastic and easily alterable even up to gastmlation stages; models of sea urchin development are discussed below (section 1.4.2.ii).

As cleavage continues a central cavity, the blastocoel, enlarges until a spherical blastula stage embryo forms, the walls of which are one cell thick. Cilia then develop on the apical cell surface of every cell, and a few hours later the blastula hatches out of its fertilization membrane (hatching blastula stage). After hatching, a small tuft of long stereocilia develop at the animal pole, the apical tuft, and at the vegetal pole the blastula wall thickens to form the vegetal plate, which will give rise to the gut, pigment cells, muscle, coelomic pouches and skeleton. The micromeres (primary mesenchyme cells, PMCs), the skeletogenic progenitor cells, located in the centre of the vegetal plate begin to migrate as individual cells into the blastocoel (known as the mesenchyme blastula stage). Gastmlation begins as the vegetal plate invaginates to form the . The PMCs form a ring around the invagination, whilst others cluster at two ventral-lateral sites, fuse to form a syncytium, and begin to deposit calcium carbonate (reviewed by Wilt, 1997). Secondary mesenchyme cells at the tip of the archenteron extend filopodia into the blastocoel, where they contact with the inner surface of the ectoderm. Once in contact with the ectoderm the filopodia contract and 'puli' the archenteron to the blastocoel wall, where it will fuse with the stomadeum formed 32 by the oral ectoderm. The apical tuft disappears and a band of cilia (the ‘ciliated band’) surrounds the oral area at the junction with aboral ectoderm. The archenteron becomes subdivided into the fore-, mid- and hind-gut by two constrictions, and by the end of gastmlation (the prism stage) the tip of the archenteron has bent to one side of the embryo, such that the original animal-vegetal axis is also bent. The prism embryo then elongates to form a pluteus with two oral arms and two anal arms, and secondary mesenchyme cells (SMCs) form the coelomic pouches. Inside the coelomic pouches are found the small micromeres that will contribute to the echinus mdiment, from which the adult metamorphoses.

1.4.2 Cell fate specification in the sea urchin emhryo

1.4.2.i Fate maps By the blastula stage, five territories, oral ectoderm, aboral ectoderm, vegetal plate, skeletogenic mesenchyme and small micromeres, can be distinguished either by morphology, e.g. oral ectoderm is a thickened epithelium, whereas aboral ectoderm is a squamous epithelium, or molecular markers, e.g. Endo 1 recognizes an antigen expressed by endodermal cells (Coffman et al., 1985) and the Cyllla cytoskeletal actin gene is expressed in aboral ectoderm (Cox et al., 1986). Until recently it was thought that these lineages segregate during cleavage stages by the 60 cell stage. This assumption was based on old fate maps using vital dyes (Horstadius, 1973) and a fate map of the 8-cell embryo using modern lineage tracers (Cameron et al., 1987; reviewed by Davidson, 1989; Cameron and Davidson, 1991); a cartoon drawing of this fate map is shown in figure 1.8B (Old). However, a recent fate map by Logan and McClay (1997), using the lipophilic dye Dil as a lineage tracer, shows that the ectodermal and endodermal lineages do not separate until after the 64 cell stage and that the position the cell finds itself in later in development is more important than earlier interactions. A cartoon drawing of this new fate map is shown in figure 1.8B (New). However, since Logan and McClay only traced the progeny of vegl and veg2 tier cells, other tiers may in the future be found to contribute to other lineages and so this fate map may need further modification.

1.4.2.Ü Inductive interactions Very little is known about the mechanisms of sea urchin development, for instance what are the molecular pathways that determine different cell fates and how is the oral-aboral axis is determined? Based on early isolation experiments and his own findings, Runnstrom (1929) proposed a two gradient model of sea urchin development in which an animalizing signal came from the animal pole and a vegetalizing signal from the vegetal pole. In combination these two gradients specified fates along the animal-vegetal axis (Horstadius, 1973). Later, Horstadius extended this earlier work by performing many isolation and recombination experiments, explaining his results in terms of a double gradient (reviewed by Horstadius, 1973). Isolated animal halves form ciliated balls, whereas vegetal pole explants form ovoids containing spicules, gut and pigment cells. When micromeres are recombined with 33 animal pole cells (anl and an2), normal, or near normal embryos, with a complete range of structures form. These isolation/recombination experiments have been repeated recently using modem molecular markers, giving the same basic results (reviewed by Livingston and Wilt, 1990). Micromeres are also able to induce a second gut if transplanted to the animal pole (Ransick and Davidson, 1993), so the micromeres are clearly a source of an inductive signal. However, if the micromeres are removed from a cleavage stage embryo, a normal embryo forms (with SMCs laying down the skeletal spicules in the absence of primary mesenchyme cells), demonstrating that the micromeres are not absolutely essential for development.

Inductive interactions are clearly operating in the early embryo, as shown by these isolation and recombination experiments, however, as described above, the cleavage planes of the early embryo were previously thought to correspond to lineage boundaries. In order to reconcile these ideas, Davidson (1989) proposed a model in which the cleavage boundaries are sites of inductive interactions which segregate different lineages. In Davidson’s model, the micromeres act as a vegetal signalling centre, which signal to the overlying veg2 tier of cells. The veg2 tier then signals to the vegl tier, and so on as the signal is propagated towards the animal pole; these signals would then induce lineage specific gene expression by modulating the activity of maternal transcription factors. In this model there is no animalizing signalling, since the evidence from classical experiments for such a signal is considered weak (Davidson, 1989; Livingston and Wilt, 1990).

Until recently it was unclear whether secreted signalling molecules could act over a distance of several cell diameters without the need for the signal to be relayed from cell to cell, as proposed by Davidson. However, recent work in Xenopus and Drosophila has shown that at least in the case of Activin and Dpp, a relay system is not used, since expressing constit itively active Activin or Dpp receptors does not result in long range activation of downstream target genes, as would be expected if a relay system were working (Jones et al., 1996a; McDowell et al., 1997; Nellen et al., 1996; Lecuit et al., 1996). In fact, the work of Logan and McClay (1997) demonstrates that, at least for the ectodermal/endodermal (vegetal plate) lineage, cleavage boundaries (at the 64 cell stage) do not correspond to cell fate boundaries, and so induction of supposed invariant cell fates does not need to be explained as signalling across these boundaries. As pointed out by Slack (1991), Davidson’s model is probably ‘unnecessarily complicated’ given the available data.

Davidson’s model also proposes a mechanism by which the oral-aboral axis would be specified. The aboral ectoderm contributes to the epidermis covering the majority of the embryo and part of the cilary band, while the oral ectoderm forms oral epithelium around the mouth, part of the ciliated band, neurons and the stomadeum. According to Davidson’s model, an initial polarization, which occurs sometime after fertilization, localizes a determinant to the future oral side of the embryo; as early as the 8 cell stage a polarity in the 34 aboral-aboral axis is seen with the oral side having high cytochromoxidase activity compared to the aboral side (Czihak, 1961; cited in Horstadius, 1973). Once an initial asymmetry is established the oral ectoderm would induce aboral ectoderm. Later signalling between the oral and aboral ectoderm would maintain the boundary and induce stmctures such as the cilary band and stomadeum in the oral ectoderm. However, experiments by Wikramanayake and Klein (1997) on L. pictus embryos suggest that, at least in this species, a vegetal signal is required to induce aboral fate. In these experiments, animal halves, which normally develop into embryoids expressing oral markers but not aboral markers, were isolated and the vegetalizing agent, lithium, was added. This treatment caused the animal halves to become patterned and express aboral markers. In light of this, Wikramanayake and Klein (1997) propose a variation of Davidson’s model which incorporates a vegetal signal that activates aboral-specific genes in animal hemisphere, whilst a negative signal suppresses oral-specific gene activation. Localized maternal determinants would make the prospective oral ectoderm refractory to these inductive and suppressive activities.

Although it is not clear how lineage is restricted in the sea urchin embryo, or how pattern is elaborated, there may be some parallels with Xenopus development. An inductive signal is released from the vegetal pole and induces aboral ectoderm, and this may be analogous to the vegetal signal which induces mesoderm in the Xenopus embryo. In the Xenopus embryo both ventral and dorsal mesoderm is induced, however, in sea urchins only aboral ectoderm is induced and oral ectoderm is apparently autonomously specified (Wikramanayake et al., 1995; Wikramanayake and Klein, 1997). Once an oral-aboral difference has been established in the sea urchin this axis can be further elaborated by signalling between the oral and aboral ectoderm, which may be analogous to dorsal-ventral patterning in the mesoderm of the Xenopus gastrula.

Of course, without knowledge of the molecules involved in vegetal signalling and the elaboration of the oral-aboral axis, we cannot say whether the processes are conserved between sea urchins and Xenopus. However, a BMP signalling pathway may be functioning during early sea urchin development, since several components of the putative signalling pathway have been identified in sea urchins. For instance, three BMP 1-related metalloproteases, suBMPl, SpAN and BPIO (Hwang et al., 1994; Reynolds et al., 1992; LePage et al., 1992), and two TGF-p-related BMPs, Univin (Stenzel et al., 1994) and suBMP2/4 (C. Logan and D. McClay, unpublished), have been isolated. Overexpression of Xenopus BMP4 (XBMP4), or suBMP2/4, by microinjection of mRNA into S.purpuratus or Lytechinus pictus embryos, causes defects in both the animal-vegetal and dorsal-ventral axes (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted). These embryos have an expanded aboral ectoderm domain and lack a ciliated band, a structure which normally forms at the boundary of oral and aboral ectoderm, indicating that patterning along the oral-aboral axis is disturbed by overexpression of BMP4. In addition,

35 overexpression of BMP4 causes a reduction in the length of the gut, although it is differentiated into fore-, mid- and hind-gut as normal. Thus BMP4 is able to suppress formation of vegetal derived structures and direct ectoderm to a more aboral (dorsal) fate. A similar phenotype is produced when SpAN, a sea urchin BMP 1-related metalloprotease, is overexpressed in sea urchin embryos, a result consistent with BMP 1-related metalloproteases activating the BMP signalling pathway. Further evidence that these metalloproteases are important in sea urchin development comes from experiments done by LePage et al. (1992), where the activity of BPIO was inhibited by injecting an antibody to BPIO into P. lividus embryos. This results in an expansion of vegetally derived structures, such as gut, at the expense of animal structures, a phenotype opposite to that seen when SpAN or BMP4 are overexpressed.

In addition, injection of Xenopus noggin mRNA into sea urchin embryos causes a vegetalized phenotype, the opposite of BMP overexpression (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted). Noggin binds to, and inhibits, BMP4 and BMP2 in Xenopus embryos (Zimmerman et al., 1996) and also inhibits Dpp activity when expressed Drosophila embryos (Holley et al., 1996). Sea urchin embryos expressing Noggin have enlarged guts, and the number of spicules, which are derived from the vegetal plate, increases, noggin mRNA-injected embryos also fail to establish oral-aboral polarity, and in contrast to bmp4 mRNA-injected embryos, the ectoderm is diverted towards an oral fate. In addition, a ciliated band marker, which also stains a subset of neural cells (Wikramanayake and Klein, 1997), is expressed throughout the ectoderm. This is interesting since in both Xenopus and Drosophila inhibition of BMP signalling leads to neural fate (Holley et al., 1995; Sasai et al., 1995; Wilson and Hemmati-Brivanlou, 1995; Biehs et al., 1996).

1.5 SUMMARY

Bone Morphogenetic Proteins (BMPs) 2-8 are members of a large family of secreted signalling molecules, the TGPp superfamily, and play an important role during normal embryonic development.

The activity of TGPPs is controlled at many levels, for instance activity may be attenuated before receptor binding by holding TGPps as inactive complexes in the extracellular space until required. Of particular interest is the finding that three proteins. Noggin, Chordin and Follistatin, are able to bind and inactivate BMP4, a TGP(3 important for dorsal-ventral patterning during early development. Homologues of Chordin and BMP4, Sog and Dpp respectively, are expressed in the embryo of the fruit fly Drosophila melanogaster, where they are involved in early patterning events. Also expressed at this time and acting in the same signalling pathway is a metalloprotease, Tolloid, which enhances the activity of Dpp by cleaving Sog and thus releasing Sog’s inhibition on Dpp. In addition, a Xenopus homologue of Tolloid, named Xolloid, is able to cleave Chordin and release active BMPs.

36 Since other metalloproteases, related to Tolloid, are present during early development of Xenopus and other animals, an attractive hypothesis is that release from inactive complexes by these metalloproteases is a general mechanism for regulating TGFP-like BMP activity. In addition sea urchin homologues of BMPl and BMPs 2 and 4 have been isolated from early embryos, raising the possibility that the BMP signalling pathway is conserved in sea urchins, and may be used to pattern the early embryo as it is in Xenopus, zebrafish and Drosophila.

The aim of this thesis is to further characterize the role of metalloproteases in BMP signalling. Since the effects of perturbing the BMP signalling pathway in Xenopus embryos are well characterized, this is an ideal system in which to investigate any role for BMPl-related metalloproteases in regulating BMP signalling. Reagents that perturb BMP signalling, such as dominant-negative receptors and inhibitory binding proteins, are available and active in Xenopus embryos. However, with the exception of Noggin, these reagents have no phenotype when expressed in the sea urchin embryo, probably because the amount of mRNA that would be required to elicit a phenotype in the sea urchin is toxic. The Xenopus embryo therefore has an advantage over the sea urchin as a model system in which to study regulation of the BMP signalling pathway.

This thesis also investigates the possibility that the BMP signalling pathway is conserved between Xenopus and sea urchins. By comparing the activity of sea urchin genes, for instance SpAN, with endogenous Xenopus homologues, such as XBMPl and Xolloid, evidence may be provided that these homologues have similar functions. This would suggest that the BMP signalling pathway is highly conserved, being found not only during development of Xenopus, zebrafish and Drosophila embryos, but also in sea urchins.

To these ends, two Xenopus metalloproteases, XBMPl and Xolloid, as well as the sea urchin homologue, SpAN, were overexpressed in Xenopus embryos and their effects on dorsal-ventral patterning and morphogenesis were analyzed. In addition, putative dominant- negative constructs of SpAN, XBMPl and Xolloid were made and used in an attempt to block endogenous metalloprotease activity. Finally, two sea urchin TGFp BMP-like proteins, which may be endogenous targets of SpAN, were tested in the Xenopus embryo for functional homology with Xenopus TGFp BMPs.

37 Figure 1.1

Phylogenetic tree illustrating relationships between different members of the TGFp superfamily based on homology in the mature, C-terminal domain.

To construct this tree Pileup analysis on GCG was used. For references see text, except: GDFl (Lee, 1990), GDF3 (McPherron and Lee, 1993), GDF5, GDF6 and GDF7 (Storm et al., 1994), GDFIO (Cunningham et al., 1995), Dorsalin (Easier et al., 1993), Radar (Rissi et al., 1995), Screw (Aroraet al., 1994).

38 Figure 1.1

TGF-P1 TGF-P5 TGF-p TGF-P2 subgroup TGF-P3 Activin C Activin D Activin Activin E subgroup Activin A Activin B Xnr 1 Xnr2 Nodai-reiated Xnr 3 subgroup nodal Xnr 4 BMP 3 GDF10 ADMP GDF6 Radar GDF5 GDF7 BMP 5 BMP 6 60A F BMP 7 subgroup BMP 8

BMP 2 BMP 4 Dpp rE SUBMP2/4 subgroup

ZÜVR1 Vg1 □ subgroup Univin GDF3 Dorsa in GDF1 Screw GDF9 Inhibin A

39 Figure 1.2

(A) Domain structure of TGF(3 superfamily members. Each member has an N-terminal hydrophobic secretory signal (the preregion), a proregion involved in protein folding and secretion, and a mature, C-terminal bioactive domain. The prepro region is proteolytically cleaved from the bioactive domain at an RXXR site. The C-terminal domain generally contains 7 conserved cysteine residues which form the TGFp knot, a structural motif.

(B) TGFp signal transduction by Smads. Activin is given as an example since many of the components involved have been identified. On binding of Activin to the type II receptor, this receptor/ligand complex interacts with the type I receptor which is phosphorylated by the constitutively active type II receptor kinase. Smad 2 is in turn phosphorylated by the type I receptor kinase and forms a complex with Smad 4. This Smad 2/4 complex then associates with FASTI, a winged-helix transcription factor, and is translocated to the nucleus. The Smad 2/4/FAST 1 complex binds to the promoter of mix2 and activates transcription. Smad 6 and Smad 7 are able to inhibit signal transduction, perhaps by binding to the type I receptor kinase and blocking phosphorylation of Smad 2.

40 A. Domain structure of TGFp superfamily members

cc cc cc Signal Proregion Mature bio-active sequence domain

B. Activin signal transduction

Activin

CYTOPLASM

Smad6/7

Smad4

NUCLEUS

FASTI mix.2

41 ZINC METALLOPROTEASES

ZINCINS INVERZINCINS CARBOXYPEPTIDASES DD-CARBOXYPEPTIDASES A. HEXXH HXXEH HXXE HXH

1 GLUZINCINS METZINCINS B. HEXXH. HEXXHXXGXXH. 1 Insulinase Thermoiysin Endopeptidase-24.11 Aminopeptidase Astacin Reprolysin Family Family Family Family Family Family (inicl. snake venom C. ISXMHY proteases)

Angiotensin Converting Endopeptidase-24.15 Tetanus and Serratia Matrixin Enzyme Family Family botulism Family Family neurotoxins meprlns astacin I------BMP1-related collagenases stromelysins gelatinases

Figure 1.3 Families of zinc metailoproteases based on the amino acid sequences around the zinc binding site.

Bold letters represent identified zinc ligands. X stands for any amino acid. Boxed residues on line A indicate the two ligands in the short zinc binding motif; boxed residues on line B indicate the third ligand and, in the c a se of metzincins, the 'met-turn' methionine; boxed residues on line C indicate the fourth ligand and conserved motif in the case of astacins. Adapted from Hooper (1994).

42 Figure 1.4 Domain structure of BMPl-reiated metalloproteases

BMPl-related metalloproteases have an N-terminal hydrophobic secretory signal, a proregion which may be involved in activation of the protease, a metalloprotease domain and a C-terminal domain containing calcium-binding EGF-like repeats, CUB repeats and other variable domains. BMPl exists as at least two splice variants, a long and a short form.

For references, see text.

43 Figure 1.4

SpAN/BP10 n

hchi

BMP1 (short splice variant)

Xld / mTII / apTBL / BMP1 (long splice variant) □

suBMPI

Tld I l I I I

Tlr-1 n

XHE

KEY; Prepro region Thr domain [ I Metalloprotease Ser/Thr Domain r ) CUB domain I I Variable I 1 EGF-Ilke domain

4 4 BP10

SpAN

hch1

r - hBMP1

mBMP1

XBMP1

Xolloid

mTII

suBMPI

Tlr-1

Tolloid

apTBL

XHE

Figure 1.5 Phylogenetic tree demonstrating relatedness of BMPl-like metalloproteases based on similarities in the metalloprotease domain.

To construct this tree Pileup analysis on GCG was used. See text for references

45 Figure 1.6 Known cleavage sites of astacin metalloproteases.

Astacin Some examples of astacin cleavages sites in synthetic peptides:

Cleavage site P3 P2 PI i P I ’ P 2 ’ Leu Lys Tyr Ala Pro Pro Lys Arg Ala Pro Gly Arg Arg Ala Pro

Astacin degrades fibrillar Collagens, which have a proline every three residues. Not surprisingly astacin has a preference for a proline residue (Pro) in position P2’ of the substrate. In addition, short chain amino acids (such as Ala) are preferred in the P I’ position, and hydrophobic residues in P3’ and P4’. Arg, Lys, Asn and Tyr are preferred in positions PI and P2, and in position P3, Pro> Val> Leu> Ala> Gly are preferred in that order (Stocker and Zwilling, 1995).

BMPl Known cleavage sites:

______Cleavage site P3 P2 PI i pr P 2’ Substrate Tyr Arg Ala Asp Asp pro-Collagen al (I) Tyr Arg Ala Asp Gin pro-Collagen a2 (I) Tyr Tyr Gly Asp Glu pro-Collagen al (III) M et Val Gly Asp Asp pro-Lysyl Oxidase

XBMPl cleaves pro-Collagens type I, II and III, pro-Lysyl Oxidase and the pro-Laminin chain y2. In some cases (e.g. human pro-Collagen a l (II), pro-Lysyl Oxidase) these cleavage site also have a proline (Pro) in the P2’ position, although this is not required for cleavage since human pro-Collagen al (I) does not contain a proline residue at this position. (References: Hojima et al., 1985 ; Kessler et al., 1986 ; Panchenko et al., 1996; Amano et al., 1997).

PI, P2, etc. and PI’, P2’, etc. designate substrate residues N-terminal and C-terminal to the scissile bond respectively.

46 Figure 1.7

(A) Fate map of X enopus gastrula. During gastmlation the three embryonic germ layers become rearranged so that the endoderm runs through the centre of the embryo, the ectoderm encases the outside and the mesoderm is sandwiched between these two layers. As the embryo continues to develop, the neural tubes forms from the dorsal ectoderm and comes to overlie the notochord. Mesodermal derivatives include blood islands, lateral plate, somites and notochord. At tailbud stages these tissues are identifiable histologically and their arrangement in the tailbud embryo is shown here schematically.

(B) The three-signal model. During blastula stages two signals from the vegetal pole induce mesoderm in the overlying marginal zone; a ventral-vegetal (VV) signal induces ventral type mesoderm (mesenchyme, blood) whilst a dorsal-vegetal signal (DV) induces dorsal-type mesoderm (notochord). Later, during late blastula and gastrula stages, a dorsalizing signal from the newly-induced dorsal mesoderm (the organizer; O) induces lateral mesoderm such that the marginal zone now contains a full range of mesodermal tissues types (blood, lateral plate mesoderm, pronephros, somites and notochord).

47 FateJVIaji

Gastrula Tailbud

Epidermis Notochord

H Neural Tube Somite HZl Endoderm Lateral Plate Blood Island

B Three signal model

Cleavage ^ Blastula Early gastrula Late gastrula

VP

Mesoderm Dorsalization induction

4 8 Figure 1.8

(A) Schematic diagram illustrating stages of sea urchin development. Early cleavage divisions are unequal and give rise to micromeres, macromeres and mesomeres. The mesomeres divide to give the anl and an2 tiers, which will become specified as ectoderm. The macromeres divide to give the vegl and veg2 tiers which will contribute to several tissues of the ectodermal and endodermal lineage. The micromeres divide to give the small micromeres which will form part of the echinus rudiment, and the skeletogenic (primary) mesenchyme, which as the name suggests, will form the larval skeleton. Cleavage continues until a small blastula is formed which then hatches out of its fertilization membrane. The primary mesenchyme cells (PMCs) then begin to ingress, and the embryo is known as a ‘mesenchyme blastula’. Next the embryo begins to gastrulate, the vegetal plate invaginates, and the PMCs secrete the skeletal rods. At the end of gastmlation, the embryo is referred to as the prism stage, which elongates to become a free-feeding pluteus larva.

(B) Fate map of a 64 cell embryo. The ‘old’ fate map, based on early experiments with vital dye, indicates that the vegl tier does not contribute to the endodermal lineage (vegetal plate). Recent experiments have produced a ‘new’ fate map which shows that the vegl tier may contribute to either the ectodermal or endodermal lineage. This fate map, however, is based on lineage tracing experiments for the vegl and veg2 tier only, and is likely to change as the fate of other blastomeres is examined using modem lineage tracers. Adapted from Logan and McClay, 1997.

49 Figure 1.8

A

mesomeres an 1 \ an 2 macromeres

micromeres 16 cell embryo 32 cell embryo

apical tuft

primary mesenchyme cells

micromeres small micromeres (1 “ mesenchyme) vegetal plate

64 cell embryo mesenchyme blastula

secondary oral arm mesenchyme cells spicule (oral rod)

gut

mouth j L spicules

archenteron anal arm gastrula pluteus

oral ectoderm aboral ectoderm

vegetal plate

small micromeres skeietogenic (primary) mesenchyme Old New

5 0 Chapter 2

MATERIALS AND METHODS

2.1 Materials

2.2.1 Vectors

Vectors used to make constructs for injection pBluescript.KS+: Stratagene pSP64A: Promega pSP64T : Kreig and Melton, 1984 pRN 3: Lemaire et al., 1995 pCS2+: Rupp et al., 1994; Turner and Weintraub, 1994 pRN3.1in is based on pRN3 and is used for making capped transcripts. The Pstl-Kpnl fragment of pRN3 was replaced with the Pstl-Kpnl fragment of pNEB 193 (New England Biolabs). This deletes the Sfil site of pRN3 and provides additional sites for linearization. pTVlink is based on pGEM2 (Promega) into which the 5’ untranslated region (UTR) of (3- globin has been inserted (gift of Caroline Hill).

CS2+.MT is based on CS2+. 6 copies of the myc epitope recognized by the 9E10 monoclonal antibody were inserted between the Clal and EcoRl sites of the CS2+ polylinker (D. Turner and R. Rupp).

2.1.2 DNA constructs

The following constructs were made as part of this study for injection into Xenopus embryos. Other constructs used for injection or for probes are referenced in the text.

Metalloprotease constructs pSP64A.SpAN and pSP64T.SpAN contain the coding region of SpAN (Reynolds et al., 1992) with an in frame myc tag attached to the 3' end. pSP64T.ASpAN is the same as pSP64T.SpAN except that glutamic acid 191 is converted to alanine. These constructs were a gift of Dr Lynne Angerer (Rochester University, USA). pBluescript.SpAN was made by ligating a Smal/Sall fragment from pSP64A.SpAN, which contains the entire coding region of SpAN and the 3' myc tag, into the Smal/Sall site of pBluescript KS+.

pRN3.Xld was made by Dr Rodolpho Alb an o and contains the full coding region of Xolloid. pCS2+.Xld was a gift of Dr Stefano Piccolo (UCLA, USA). JC2.BMP1 was made by Dr Rodolpho Albano and contains the full coding region of Xenopus B M Pl.

51 pCS2+.XldMT and pCS2+.BMPlMT were made by Dr Shelley Goodman and contain six in frame myc tag sequences at the C-terminus with the full length coding region. pCS2+.XldCub, pCS24-.BMP 1 Cub and pCS2+.SpANCub contain the BMPl signal sequence fused to the C-terminal domain of Xolloid, BMPl or SpAN respectively. An EcoRl restriction site was introduced at the 5’ end of the C-terminal domain of each gene by PCR, and the PCR product was fused to the native EcoRl site at the 3’ end of the BMPl signal sequence. pCS2+.BMP 1 Cub was made by Dr Shelley Goodman. pCS2+.XldCub.MT was made by replacing the Sphl/Xbal fragment from pCS2+.XldCub with the Sphl/Xbal fragment from pCS2+.XldMT which contains the six myc tags. pCS2+.BMPlCub.MT was made by replacing the Sstl/Xbal fragment from pCS24-.BMP 1 Cub with the Sstl/Xbal fragment form pCS2-i-.BMP 1 MX. pSP64T.Smp (SpAN metalloprotease domain) was made by ligating a PCR product of the SpAN metalloprotease domain, which had an EcoRl site introduced at the 3’ end, into pTTlink. This was then blunt end cloned into pSP64T.

Univin constructs pSP64T.univin (1 AUG) was gift of Dr Lynne Angerer. It contains the full coding region of Univin (Stenzel et al., 1994) but the 5’ UTR is truncated so that only one AUG with a favourable Kozak sequence is present. This ensures that only one translation product is made. pBluescript.Bunivin was made by Dr Shelley Goodman. An Sphl site was introduced at the 5' end of the bioactive domain of Univin by PCR. The PCR product was ligated into pBluescriptKS+.BVgl, replacing the bioactive domain of Vgl, so that the Univin bioactive domain is fused to the preproregion of BMP4. This was then subcloned into the EcoRl site of pRN31in.

Other constructs used for injection pBluescript.suBMP2/4 was a gift of Cati Logan and Dr Dave McClay (Duke University, USA), and contains the full coding region. This was subcloned into the EcoRl site of pRN31in.

2.1.3 Solutions and media

Unless otherwise stated chemicals were obtained from BDH (AnalaR grade) or Sigma, general enzymes were obtained from Promega and restriction enzymes from Boehringer Mannheim. Where stated solutions were autoclaved at 121°C for 15 mins.

52 General buffers and reagents

TE pH 8.0 lOmM Tris-HCi pH 8.0 ImM disodium EDTA Autoclaved and stored at RT.

Neutralization and preparation of TE-equilibrated phenol 8-hydroxyquinoline (0.1% ) was added to liquefied phenol to prevent oxidation. The phenol was then extracted with equal volumes of IM Tris-HCl pH 8.0, until the phenol was pH 8.0. Phenol was then mixed with an equal volume of O.IM Tris-HCl pH 8.0/0.2% 2-mercaptoethanol and allowed to separate. This was replaced with a layer of TE and stored at 4°C in the dark.

Phenol pH 4.3 Molecular biology grade phenol (pH4.3; Sigma) was aliquoted and stored at -20°C, and used for protein extraction from RNA.

IX PBS tpH 7.5k P-Tw 80mM disodium hydrogen orthophosphate (anhydrous) IX PBS 20mM Sodium dihydrogen orthophosphate 0.1% Tween 20 lOOmM Sodium chloride (NaCl) Autoclaved and stored at RT.

IX TBS T-Tw 25mM Tris-HCl pH8.2 IX TBS 144mM NaCl 0.1% Tween 20 Autoclaved and stored at RT.

Proteinase K stock Proteinase K was dissolved in sterile water at a concentration of 50mg/ml and stored at - 20°C.

Xenopus solutions

Folligon (FSH substitute: Intervetl 200iU/ml Lyophilized Folligon was dissolved in 5ml phosphate buffered water (supplied) and used immediately, or frozen on dry ice and then stored at -70°C.

Chorulon (Chorionic Gonadatrophin: Intervetl 1000-1500iU/ml Lyophilized Chorulon was dissolved in phosphate buffered water (supplied) and used immediately.

53 IPX Modified Barth's Saline TMBS^ Solution A (pH 7.6) Solution B 0.88M Sodium chloride 3mM Calcium nitrate. 4 H 2O lOmM Potassium chloride 4mM Calcium chloride.2H20 24mM Sodium bicarbonate 8mM Magnesium sulphate.7H20 lOOmM HEPES Filter sterilized and stored in 50ml aliquots at -20°C.

IX MBS 50ml solution A 50ml solution B 400mls sterile water 1.25ml lOmg/ml Gentamycin (Gibco BRL). Stored at 4°C.

IX calcium, magnesium-free MBS Made up as above but without Solution B.

2% Cvsteine-HCl pH 7.8Z8.2 4g L-Cysteine hydrochloride 1-hydrate 1.33-1.36g NaOH 200ml water Made up immediately before use.

Media and solutions for DNA preparation

Ampicillin stock Ampicillin (sodium salt) was dissolved in sterile water to a concentration of lOOmg/ml and stored at -20°C.

LB (Luria BertanO medium 4g Tryptone (Gibco BRL) 2g Yeast extract (Gibco BRL) 4g Sodium chloride 400ml water Autoclaved and stored at RT. LB, with appropriate antibiotic selection, was used for the culture of E.coli.

LB agar plates LB agar was made as above but with the addition of 15g agar per litre. After autoclaving the agar was allowed to cool to 55°C and ampicillin was added to a concentration of lOOpg/ml. Poured plates were stored at 4°C for less than a month.

54 Solution I (GTE") Solution II 25mM Tris-HCl pH8 0.2M Sodium chloride lOmM EDTApHS 1% Sodium Dodecyl Sulphate (SDS) 50mM Glucose Made up fresh before use. Stored at RT.

Solution III 60ml 5M potassium acetate 11.5ml glacial acetic acid 28.5ml water Stored at RT.

IPX TBE 40XTAE 0.89M Trizma 1.6M Trizma 0.89M Boric acid 0.8M Sodium acetate.3H 2O 20mM EDTA 40mM EDTA Stored at RT. pH adjusted to 7.2 with acetic acid. Stored at RT.

6X DNA Dves 2X Formamide Dves 0.25% Bromophenol blue 95% Formamide 30% Glycerol in sterile water 20mM EDTA Stored at RT. 0.05% Bromophenol blue 0.05% Xylene Cyanol stored at -20°C

Ikb ladder (Gibco BRLl 4|il 1 kb ladder 20)11 6X DNA dyes 76p,l sterile water Typically 6|il of the Ikb ladder was run on an agarose mini-gel. Stored at -20°C.

1 % Agarose gel 25ml IX TBE or IX TAE 0.25g agarose Heated in microwave to melt. Ipl lOmg/ml ethidium bromide (Gibco BRL) was added and the gel cast in mini-gel apparatus.

55 Solutions for sequencing

40% acrvlamide mix 38g acrylamide 2g bis-acrylamide Made up to lOOmls with sterile water, stored at 4°C.

7M urea. 6% acrvlamide gel mix for sequencing gels lOmls 1 OX TBE 15ml 40% acrylamide mix 42g urea Made up to lOOmIs with sterile water, then filtered and de-gassed. 600|il freshly made 10% Ammonium persulphate (APS) and 100|il TEMED were added to lOOmIs of gel mix, to polymerize the gel, and the gel poured immediately.

DNA Denaturing Solution 2M Sodium hydroxide 2mM EDTA Stored at RT.

USB 5X Labelling mix rdlTP) 15pM dITP 7.5pM dCTP 7.5|iM dTTP

USB Termination nucleotide mixes Stored at -20°C ddG: 160pM dITP ddA: 80p.M dITP 80|iM dATP 80|iM dATP 80|lM dCTP 80^M dCTP 80|iM dTTP 80 )liM dTTP

1.6|liM ddGTP 8|iM ddATP 50mM Sodium chloride 50mM Sodium chloride

ddT: as ddA but 8 fiM ddATP replaced with 8|liM ddTTP. ddC: as ddA but 8|iM ddTTP replaced with 8}iM ddTTP.

Solutions for protein work

5X Electrode Buffer (pH8.3) 9g Trizma 43.2g Glycine 3g SDS Make up to 600ml with water. Stored at RT. 56 2X Sample Buffer Transfer buffer tpH8.3) 1ml l.OM Tris-HCl (pH 6.8) 25mM Tris 1.6ml Glycerol 192mM Glycine 3.2ml 10% SDS 20% (v/v) M ethanol 0.4ml 0.05% (w/v) Bromophenol blue 9.0ml sterile water 0.8ml p-mercaptoethanol Stored at 4°C.

30% acrylamide mix G7.5:l Acrvlamide:Bis-acrvlamide) 29.2g Acrylamide 0.8g N'N'-Bis-methylene-acrylamide Make up to lOOmIs with sterile water. Stored at 4°C.

Homogenization buffer A 50mM Tris-HCl pH 7.5 5mM EDTA pH 8.0 0.5% SDS Stored at -20°C

Coomassie Blue staining solution Coomassie Blue destaining solution 50% (v/v) MetOH 10% (v/v) MetOH 10% (v/v) Acetic acid 5% (v/v) Acetic acid 0.25% (w/v) Coomassie Blue

RNase Protection Assav Solutions

Proteinase K buffer (for 40 reactions) 10ml Homogenization Buffer A 1.5ml 10% SDS 0.48ml 5M Sodium Chloride 20mg Proteinase K Incubated at 37°C for 15 mins before use to remove any RNases which may be present in the solution.

Elution buffer 0.1% SDS ImM EDTA 0.5M Ammonium acetate 1 OmM Magnesium acetate

57 Ambion RPAII Hybridization Buffer 80% Deionized formamide lOOmM Sodium citrate pH 6.4 300mM Sodium acetate pH 6.4 ImM EDTA

Whole Mount in situ Hybridization Solutions

MEMFA: O.IM MOPS pH7.4 2mM EGTA 1 mM Magnesium sulphate 3.7% Formaldehyde Made up fresh before use.

M ABtpH 7.81 lOOmM Maleic acid 150mM Sodium chloride Adjusted to pH 7.8 with sodium hydroxide and autoclaved. When the solution had cooled to RT 0.1 % Tween 20 was added. Stored at RT.

Hybridization buffer 5OX Denhardt's solution 50% Deionized formamide 1% (w/v) BSA (Pentax Fraction V) 5X SSC pH7 1 % (w/v) Ficoll 1 mg/ml Torula RNA 1 %(w/v) Polyvinylpyrolidone 100p.g/ml Heparin Stored at -20°C. IX Denhardt's solution 0.1% Tween 20 0.1% CHAPS lOmM EDTA pH7.5 Stored at -20°C.

20XSSC pH 7.0 3M Sodium chloride 0.3M tri-sodium citrate Adjusted to pH 7.0 with sodium hydroxide. Autoclaved and stored at RT.

Heat-treated Lambs' Serum: (Gibco BRL) Heat treated at 55°C for 30 mins. Stored at -20°C.

58 AP buffer : lOOmM Tris-HCl pH 9.5 lOOmM Sodium chloride 5OmM Magnesium chloride 0.1% Tween 20 5mM Levamisole Made up fresh before use.

Solutions for Histology

20% Paraformaldehyde stock 20g paraformaldehyde 100ml water The mixture was heated to 65°C and drops of IM NaOH added until the paraformaldehyde dissolved. Stock solution stored at -20°C. 20% paraformaldehyde was diluted in IX PBS to make 4% paraformaldehyde for fixing tissues.

Polv-l-lvsine coated slides Slides were racked and then cleaned by soaking in detergent (Decon Laboratories) for several hours, rinsing in tap water and soaking in distilled water for several hours. Poly-1- lysine solution (Sigma, 0.1% w/v) was diluted 1:10 with distilled water and the racks of slides dipped in this solution for 5 mins. The slides were then drained and dried overnight atRT.

Borax carmine 8g Carmine lOg Sodium borate 250ml sterile water The mixture was boiled for 30 mins, cooled overnight and filtered. This was evapourated to dryness on a hot plate and a saturated solution made up with 35% ethanol. Stored at RT.

Picro-blue-black 5g Picric acid 390ml sterile water 9.75g Naphthol blue black

59 2.2 Methods

2.2.1 Xenopus techniques

2.2.1.1 Obtaining Xenopus eggs Adult Xenopus males were killed by placing in anaesthetic (0.2% 3-aminobenzoic acid ethyl ester; MS222) for 1 hour and the testes dissected out. Testes were stored at 4°C in L I5 Medium with L-glutamine (Gibco BRL) containing 25p,g/ml Gentamycin (Gibco BRL for up to one week. Adult X enopus females were injected with 25-lOOiu of 'Folligon' into the dorsal lymph sac three to seven days before eggs were required. The evening before eggs were required these frogs were induced to lay eggs by injecting 500- 750ÎU of 'Chorulon' into the dorsal lymph sac.

2.2.1.Ü In vitro fertilization Eggs were squeezed from the female into a petri dish to form a monolayer. Part of a testis was macerated in the dish and the opened testis wiped over the eggs. The eggs were then left for 5-10 mins before flooding with O.IX MBS.

2.2.1.iii Dejelling eggs Eggs were placed in a beaker containing 2% cysteine (pH 7.8/8.2) and swirled for approximately 3 mins until the jelly coat had dissolved, then rinsed twice in tap water and placed in O.IX MBS.

2.2.1.iv Microinjection of mRNA into Xenopus embryos Needles were made from 1mm external diameter glass capillary tubes. Needles were pulled using a micropipette puller and the tip of the resulting needle was broken off using watchmakers forceps so that the external diameter at the tip was approximately 25|im. This generally led to a needle which would dispense 5nl of liquid/second with a pressure of 2 psi.

Embryos were placed in a 1% agar coated dish containing O.IX MBS/4% Ficoll and injected with 10-16nl of synthetic mRNA at the concentration indicated in the text. Agar prevents the embryos sticking to the plastic dishes. The high molecular weight Ficoll in the MBS solution draws water from between the vitelline space and facilitates injection.

The site of RNA injection can effect the phenotype of the resulting embryo. mRNA injected before the first cleavage division diffuses throughout the egg and so most cells inherit the mRNA injected (see for example Ruiz i Altaba and Melton, 1989). Localizing mRNA to say the dorsal side of the embryo will result in only dorsal cells inheriting the mRNA. In this study mRNAs were injected into the equatorial or animal pole region of the embryo at the 1-cell stage, 2-cell stage or 4-cell stage. For membrane proteins, or those less efficiently secreted, a stronger phenotype results if injections are done into both blastomeres at the two cell stage, which presumably distributes RNA more evenly

60 throughout the embryo. Injections were also localized to the dorsal or ventral side at the four cell stage. At this stage dorsal and ventral blastomeres can be distinguished since dorsal blastomeres are generally smaller and have less pigment than ventral cells.

2.2.1.v Culturing embryos Injected embryos were incubated in O.IX MBS/4% Ficoll for several of hours, after which the medium was replaced with O.IX MBS. Once embryos had completed gastrulation they were transferred to IX MBS. Before gastrulation incubating embryos in a high salt solution (i.e. IX MBS) leads to exogastrulation. During incubation any dead embryos were removed to prevent spoiling surviving embryos.

2.2.1.VÜ Staging and scoring embryos Injected embryos were scored using the Dorsoanterior Index (DAI) of Kao and Elinson (1988). In this index normal embryos are given a score of 5. Ventralized embryos score 0- 4 and dorsalized embryos score 6-10. See figure 2.1. Embryos were staged according to Nieuwkoop and Faber (1967).

2.2.1.viii Microdissection of embryos and dissociation of cells Embryos were placed in IX MBS in agarose-coated dishes and the vitelline envelop was removed from each embryo using watchmakers forceps. Animal pole explants (animal caps) were dissected out at stage 8-9 (mid-blastula) using watchmakers forceps. Ventral marginal zone explants (VMZs) and dorsal marginal zone explants (DMZs) were dissected out with needles at stage 10-10.5. Explants were incubated in IX MBS, since they heal more quickly and survive better in a high salt solution.

Medium conditioned with human Activin A from Chinese Hamster Ovary (CHO) cells was a gift of Jim Smith. This was diluted 1 in 1000 with IX MBS and used for Activin induction assays.

VMZs and DMZs were cut from stage 10 embryos and dissociated in calcium, magnesium-free medium by pipetting up and down in a drawn out, serum-coated Pasteur pipette. Dissociated cells were then transferred to medium containing Ca^+ and Mg^+ (IX MBS) and blown into a loose heap with a stream of medium from the Pasteur pipette.

2.2.1.viii Obtaining Xenopus oocytes Adult Xenopus females were sacrificed in 0.2% MS222, and the ovaries removed. Oocytes were dissociated from the ovary tissue by incubating in 2mg/ml collagenase for 2 hours with gentle shaking, then washed several times in IXMBS.

2.2.1.ix Injection of mRNAs and culture of oocytes Mature oocytes were injected with 20-30|il of the indicated mRNAs and incubated in IX MBS overnight at 18°C. Any dead or unhealthy oocytes were removed to prevent them spoiling the surviving oocytes. Oocytes were then placed in 96 multiwell plates (5

61 > Q. -n 03 O \ Bauchstuck (Q‘ ■E c 2. o o to 3 Acephalic

0) ZJ Û. 0 o 3 3\ Microcephalic o 2 3 8 &> ^ Cyclopic S i 0

3 Reduced eyes Q. and forehead g to0) CJl Normal

OA Bent axis

Short axis

No axis

Multiple eyes, f Janus twin

Radial proboscis oocytes per well) with 50 |Li 1 IX MBS and incubated for 48 hours. After 24 hours medium was collected and frozen, and replaced with fresh IX MBS.

2.2.2 DNA preparation and analysis

2.2.2.i Miniprep DNA preparation This protocol is a modification of the alkaline lysis method described in Sambrook et al. (1989). A 1.5 ml culture of E.coli carrying plasmid was grown overnight at 37°C in LB with ampicillin (50|ig/ml). Bacteria were harvested by spinning at 13,000 rpm for 5 mins in a benchtop microfuge. The supernatant was completely removed and the cells were resuspended in lOOjil Solution I by gentle pipetting. 200)il Solution II was added to lyse the cells and incubated for 5 mins at RT. Cell debris and chromosomal DNA were precipitated by adding 150|il Solution III and incubating on ice for 10 mins. The lysate was cleared by spinning at 13,000 rpm for 5 mins and the supernatant transferred to a clean tube. Plasmid DNA in the supernatant was precipitated by the addition of 2 volumes 100% ethanol (EtOH) and incubated for 10 mins on ice. DNA was pelleted by spinning at 13,000 rpm. The supernatant was then removed and the DNA pellet washed with 70% EtOH and spun for a further 5 mins. Finally the supernatant was completely removed and the pellet allowed to dry before resuspending in 30p,l sterile water.

2.2.2.Ü Midiprep DNA purification When larger amounts were required, plasmid DNA was prepared using Qiagen Plasmid Midi kit according to the manufacturer's instructions. This uses an anion exchange resin which allows the DNA to be separated from contaminants. DNA is prepared by the alkaline lysis method, passed down the resin column and eluted using high salt concentrations, whereas contaminants such as proteins and RNA are eluted at lower salt concentrations.

2.2.2.iii Determining concentration of nucleic acids DNA or RNA concentration were determined by spectrophotometry. Ipl DNA or RNA was diluted in 1ml RNase-free water and the absorbance at 260nm and 280nm measured. The concentration of the DNA or RNA was calculated from:

1 A26O = 50pg/ml DNA or 40|ig/ml RNA

The purity of DNA was calculated from the A 260/A 280 ratio where DNA with no contamination from protein has a ratio of 0.8.

2.2.2.iv Restriction digests Typically 0.5-10pg of DNA was restricted for 1 hour to overnight at the appropriate temperature, with 2-1 OU enzyme/pg DNA. The appropriate buffer for the restriction enzyme was used and conditions adjusted as indicated by the supplier (Boehringer Mannheim).

63 2.2.3 Gel electrophoresis

2.2.3.1 Agarose gel electrophoresis

DNA fragments were separated in gels of varying concentrations from 0.75% to 1.5% agarose, depending on the size of fragments to be resolved. Loading dyes were added to the DNA to be analyzed and the sample loaded into the gel. Ikb ladder was also loaded onto the gel along side the sample DNA as molecular weight standards. Gels were electrophoresed at 50-100V for as long as required. Ethidium bromide intercalated DNA was visualized under UV light using a transilluminator and photographed using a video copy processor.

2.2.3.Ü Acrylamide gel electrophoresis Gels for separating DNA sequencing fragments (2.2.6), and protected RNA fragments (2.2.9) were 6% acrylamide, 7M urea made up as indicated in materials (2.1.3). Glass plates were prepared by cleaning with detergent, ethanol and acetone, and coating the short plate with silane. Gels were poured with a 50ml syringe and allowed to set for 1 hour. Gels were run in IX TBE at 60 Watts for the required amount of time, then fixed in 10% methanol/10% acetic acid, transferred to 3MM Whatmann paper and dried at 80°C for 1-2 hours under vacuum. Dried gels were then exposed to X-ray film (Biomax, Kodak) with 2 intensifying screens (DuPont) at -70°C.

2.2.4 DNA cloning

2.2.4.! Preparation of vector DNA for cloning

The required vector DNA was digested with the appropriate restriction enzyme to generate compatible ends for ligation. When the ligation was not directional it was necessary to dephosphorylate the 5' ends of the vector using Calf Intestinal Alkaline Phosphatase (CIAP; Boehringer), to prevent self-ligation of the vector and thus reduces background non-recombinants. 10 pi lOX CIAP buffer and Ipl diluted CIAP (see below) were added to the completed restriction reaction and made up to lOOpl with sterile water. This reaction was incubated at 37°C for 30 mins, another Ipl diluted CIAP was added and incubated for a further 30 mins at 37°C. The reaction was then cleaned by extracting with phenol/chloroform and chloroform, precipitated with ethanol and resuspended in an appropriate volume of water.

Dilution of CIAP: CIAP was diluted so that 0.0lu CIAP per pmol of ends was added to the reaetion. The amount of pmol ends for linear double stranded DNA was calculated using the following formula; UgDNA X 3.04 = pmol of ends kb size of DNA

64 2.2.4.Ü Preparation of insert DNA The DNA required was digested with the appropriate restriction enzyme to generate a fragment with compatible ends for cloning. The digested insert was cleaned by phenol/chloroform extraction and concentrated by EtOH precipitation. The DNA was run on a 1% agarose, IX TAE gel to separate the insert of interest from the rest of the plasmid. The correct sized DNA band was excised from the gel using a scalpel blade and purified from the agarose using the Qiaex II gel extraction kit, according to manufacturers instructions. The Qiaex II kit uses silica gel particles to adsorb DNA whilst contaminants such as protein, agarose and dyes are washed away. The DNA is then eluted and ethanol precipitated. This precipitation step was found to be necessary for efficient ligation of the insert.

In order to construct fusion proteins it was necessary to introduce a RE site at the fusion point using PCR (see 2.2.5). The resulting PCR products were digested with the appropriate restriction enzyme and purified as above. A description of the constructs made appears in section 2.1.2

2.2.4.iii. Ligation of vector and insert DNA A ligation reaction (below) was set up so that the ratio of vector: insert was 1:3, with lOOng of vector per reaction and incubated at 16°C for 4 hours - overnight. lOOng vector DNA

X ng insert DNA l|ii 1 OX ligation buffer \\i\ 1 OmM ATP l|il lU/p.1 made up lOpl with sterile water the amount of insert required was calculated according to the following formula:

lOOng X kb size of insert x ratio insert = ng of insert kb size of vector vector

A control ligation containing without DNA gave an indication of the amount background recombinants present in each reaction.

2.2.4.iv Transformation of competent E.coli cells Competent cells (XL 1-Blue cells) were made by either Mary Rahmen or Dr Shelley Goodman. Typically they had a transformation efficiency of 1x10^. Competent cells were thawed on ice, each lOjil ligation reaction was transferred to 15ml falcon tubes on ice, then lOOp.1 of competent cells were added and gentky swirled. The cells and DNA were incubated together on ice for 20 mins then heat-shocked at 42°C for 90 sec. The tubes were then cooled on ice for 3 mins before the addition of 200|il of LB. The cells were

65 incubated at 37°C with shaking for an hour and then plated onto LB agar plates with ampicillin selection. Colonies were grown up overnight at 37°C.

2.2.4.V Analysis of transformants Single colonies were picked into 1.5ml of LB broth with SOjig/ml ampicillin and grown overnight at 37“C. DNA was made by the miniprep method and resuspended in 30|il sterile water. 3p.l of this was used for each diagnostic restriction digest to confirm the plasmid contained the correct insert.

2.2.5 Polymerase Chain Reaction (PCR)

PCR was used to create fusion proteins, by introducing a restriction site at one end of the insert to be ligated. The sequence of primers (Genosys) used to introduce these restriction sites are shown in table 2.1. PCR reactions were carried out as follows: IjLil Template DNA 2|il lOmM dNTPs 10|il 1 OX Pfu buffer (Stratagene) 0.5)il 1 mg/ml forward primer 0.5|il 1 mg/ml reverse primer 0.5p,l 2.5U/ml Pfu (Stratagene) made up to lOOp.1 with sterile water.

The PCR reaction was overlaid with mineral oil and incubated according to the following scheme: 1 cycle 95°C 3 mins 35 cycles 95°C 45 sec. 55"C 45 sec. ITC 2 mins 1 cycle ITC 7 mins

Table 2.1 Sequence of primers used to generate the restriction sites for fusion constructs.

PCR Sequence PCR product RE

primer generated site

SpANM 5'GTCTTC GAATTC GTCGCACTCG3' SpAN C-terminal EcoRl domain

SpAN ssC 5'CGAC GAATTC GAAGACTGTTCC3’ SpAN Metalloprotease EcoRl

domain

BMPl C 5'CTGTAC GAATTC TTCCC AGCTTGCGG3 ’ BMPl C-terminal EcoRl domain

66 2.2.6 DNA sequencing

To check the fusion constructs had correctly ligated, they were sequenced. DNA was made by the miniprep method, except that the cleared lysate was extracted with phenol/chloroform and chloroform and precipitated with 1ml 100% EtOH on ice for 5 mins. DNA was then pelleted, washed in 1ml 70% EtOH and resuspended in 50jJ,l H 2O. 3|il sterile water and 2p.l 2M NaOH/2mM EDTA were added to 15pi DNA to denature it, and incubated at RT for 10 mins. The denatured DNA was precipitated with 100% EtOH and 5M AmOAc by incubating at -20°C for at least 20 mins, then pelleted, washed in 0.5ml 70% EtOH and resuspended in 7 pi of sterile water. The sequencing reaction was started immediately to prevent renaturation of the DNA.

Sequencing was carried out according to manufacturer's instructions using Sequenase II kit (USB) which utilizes the chain-termination method (Sanger et al., 1977). dITP was used instead of dGTP to reduce compression in the sequence. All reagents for sequencing reaction were supplied with kit, and primers were made by Genosys. 2pl Sequenase reaction buffer and Ipl of the appropriate primer (Ipmol/pl) were added to 7pi of denatured DNA. Annealing was carried out by heating the samples to 65°C for 2 mins in a hot block, then allowing the samples to cool to less than 35°C over 30 mins. Meanwhile for each sample 2.5 pi of each termination mix (ddG, ddA, ddT and ddC) was aliquoted into a multiwell plate and pre-warmed to 37°C. Next, Ipl O.IM DTT, 2pl labelling mix

(diluted 1:4 with H 2 O), 0.5pi [^^SJdATP (5pCi; Dupont NEN) and 3.25U Sequenase Polymerase were added to the annealed DNA sample(s) and incubated at RT for 2-5 mins. To terminate the reaction, 3.5pl of the extension reaction was mixed with each pre­ heated termination mix (ddG, ddA, ddT and ddC) in the multiwell plate. These were incubated at 37°C for 2-5 mins before the reaction was stopped by adding 4pl of Stop Solution. Sequencing gels (see 2.2.3.1) were pre-run at 60 Watts for 30 mins - 1 hour. Immediately prior to loading on sequencing gel samples were heated to 75°C for 2 mins 3pl of each sample was loaded per track. The gels was run for between 1.5 and 6 hours depending on the length of sequence required. Sequences were transferred to computer and analyzed using Mac Vector software.

2.2.7 In vitro transcription

Vector DNA was prepared by the miniprep method and linearized to make a template for transcription with the appropriate restriction enzyme. The template DNA was cleaned by Proteinase K digestion (0.5% SDS, 0.5pg/ml Proteinase K, 37°C, 1 hour) followed by phenol/chloroform extraction and then chloroform extraction, which removes proteins from the solution, before precipitating with 100% EtOH and 5M AmOAc. Finally the

DNA was pelleted, washed with 70% EtOH and resuspended in lOpl RNase-free water

ready for transcription.

67 Most mRNAs in a cell are ‘capped’ with a 7-methyl guanosine residue at their 5' end, this protects the transcript from degradation and aids initiation of translation. To this end synthetic transcripts were capped with m"^G(5')ppp(5')G (cap analogue) during the transcription reaction (Kreig and Melton, 1987). All transcription reagents were supplied by Promega unless otherwise stated.

Each transcription reaction was assembled at RT as follows:

SP6 T7/T3

5|il DNA(l|ig/pl) 5pl DNA(lp.g/|xl) 10)Lil 5X transcription buffer 10)il 5X transcription buffer 2.5\i\ 0.25M DTT 2.5p.l 0.25M DTT 5|il lOmM ATP 5|Ltl 15mM ATP 5jil lOmM CTP 5|il 15mM CTP 5pl lOmMUTP 5p,l 15mM UTP 5pl l.OmMGTP 5pl 1.5mMGTP 5pl 5mM cap analogue (Ambion) 7.5pl 5mM cap analogue (Ambion) 2.5p,l RNasin 2.5p,l RNasin made up to 50pl with RNase-free H 2O made up to 50|il with RNase-free H 2O

The reaction was preheated to 37“C for 2 mins, then 2.5p.l of the appropriate RNA polymerase added and incubated for 30 mins at 37°C. Next, 2.5 p,l lOmM (SP6) or 15mM (T7/T3) GTP was added and the reaction incubated for 1 hour at 37°C. 5jil RQ DNase 1 was then added and incubated at 37°C for a further 30 mins. The reaction was cleaned by phenol/chloroform extraction and precipitated. The RNA was pelleted at 4°C, washed in 70% EtOH and resuspended in 15p,l sterile water. The concentration of the RNA measured by spectrophotometry (see section 2.2.2.iii). In addition, to check integrity of RNA and estimate amount of RNA made, lp.1 was mixed with 2X formamide dyes and run on a 1% agarose gel. It was not found necessary to check the RNA by running it on a formamide gel. Instead a normal agarose gel was freshly poured and the RNA was run for only a few mins, these precautions were taken so that RNA would not degrade on the gel. If the RNA was of good quality it would appear as a sharp band(s) when viewed under UV light. By running other RNAs of known concentration against the sample it was also possible to estimate the amount of RNA.

2.2.8 Protein Analysis

2.2.8.i In vitro translation To check that capped transcripts efficiently translate, a rabbit reticulocyte lysate (Ambion) was used to translate the transcripts in vitro (Pelham and Jackson, 1976) as described below. Promix (Amersham) contains L-[ ] -methionine and L-[^‘’S]-cysteine.

68 4X mix 40)11 reticulocyte lysate 3)xl amino acid master mix (minus methionine) 3|il 35s-Promix(41.9^Ci/^l) 2|il RNase-free H 2O lOjLil of this mix was added to 2.5|il (0.1-0.05|ig) RNA and the reaction incubated at 30°C for 1 hour. 3|il of the lysate reaction was mixed with 3|il 2X sample buffer and boiled for 5 mins before loading on a 12-8% SDS-PAGE gel (2.2.8.iii).

2.2.8.Ü Extraction of proteins from embryos and oocytes Embryos or oocytes were homogenized in lOOmM Tris pH 7.5, ImM PMSF, 0.1% Triton X -1 GO and spun at 20,000xg to remove cell debris and yolk platelets. The supernatant was transferred to a clean tube and stored at -70°C.

2.2.8.iii SDS polyacrylamide gel electrophoresis A gel of the required concentration was poured, with a stack, using BioRad mini-Protean gel casting equipment. Samples (embryo/oocyte homogenates or reticulocyte lysates) were mixed with an equal volume of 2X sample buffer and boiled for 5 mins before loading on the gel; prestained molecular weight markers (Sigma) were also loaded. Gels were run in IX electrode buffer at 200 volts for approximately an hour. After this the gel was either Western blotted (2.2.8.iv), stained for total protein with Coomassie Blue or fixed in 40% MetOH/10% acetic acid for 30 mins. After fixing the gel was incubated in Enhance (DuPont NEN) for 20-30 mins to intensify the signal from S labelled-proteins, and the Enhance precipitated by washing in water for about 5 mins. The gel was then dried at 60°C for 1-2 hours and exposed to Xray film (X-OMAT AR, Kodak).

2.2.8.iv Coomassie Blue staining The gel was placed in Coomassie Blue staining solution immediately after electrophoresis, and stained at RT, with gentle shaking, for 30 mins. After staining, the solution was poured off and the gel rinsed briefly with destaining solution several times. The gel was then placed in fresh destaining solution and left for several hours - overnight, replacing destaining solution as required. After destaining the gel was dried at 60°C for 1- 2 hours.

2.2.8.V Western blotting Proteins were blotted from the gel onto nitrocellulose membrane (Amersham) at lOOV for 1-1.5 hours in transfer buffer. The membrane was blocked overnight in 5% skimmed milk powder/T-Tw at 4°C, then rinsed briefly and washed for 1x15 mins and 2x5 mins in T- Tw. The membrane was then incubated for 1 hour at RT in 4|ig/ml 4H2 antibody or 9E10 antibody diluted 1:1000 in T-Tw. 4H2 is a monoclonal antibody that recognizes Fibronectin (gift of Doug DeSimone; Ramos and DeSimone, 1996; Ramos et al., 1996), 9E10 is monoclonal antibody that recognizes the myc epitope (Evan et al., 1985; gift of

69 Kate Nobes). Next the membrane was washed for 1x15 mins and 2x5 mins in T-Tw and then incubated for 1 hour at RT in secondary antibody (anti-mouse coupled to horseradish peroxidase (HRP); Amersham) diluted 1:1500 in T-Tw. The membrane was washed again for 1x15 mins and then 4x5 mins in T-Tw before ECL detection (Enhanced Chemiluminescence; Amersham). This uses luminol which is oxidized by HRP and emits light as it decays that can be detected by X-ray film.

2.2.9 RNase Protection Assay (RFA)

RNase protection assays (Lee and Costlow, 1987) were carried out using the Ambion RPA II kit according to manufacturer's instructions.

2.2.9.1 Total RNA isolation from embryos and explants Whole embryos or explants were collected at the stage indicated in the text, frozen on dry ice and stored at -70°C until required. The frozen samples were homogenized in 300pl of Proteinase K buffer and incubated at 37°C for 15 mins. The samples were extracted up to three times with phenol/chloroform and precipitated with 100% EtOH and 5M AmAOc on dry ice, then spun at 13,000 rpm in a benchtop microfuge at 4°C for 20 mins. The pellet was washed in 0.5ml 70% EtOH and resuspended in RNase-free water. Samples were generally resuspended to give 1 embryo equivalent per 5 pi, or 1 explant equivalent per Ipl.

2.2.9.Ü Radioactive labelling of RNA probes DNA was prepared by the miniprep method, linearized with the appropriate restriction enzyme and cleaned by proteinase K and phenol chloroform extraction to make a template for transcription. For each probe the following reaction mix was assembled at RT and incubated for 1 hour at 37°C (all reagents supplied by Promega unless otherwise stated).

2pl DNA 0.5pl 2.5mg/ml BSA 0.5pl 0.5MDTT 1 .Opl 5mM NTPs (5mM each A,U,G) 2.0pl 5X Transcription buffer 0.5pl RNasin 2.5pl 32p CTP (DuPont NEN, SOOCi/mmol) 0.5pl appropriate RNA polymerase 1.5pl RNase-free water lOpl

Ipl RQ DNase 1 was then added to the transcription reaction and incubated at 37°C for 15 mins, in order to remove the template DNA which can cause spurious background bands.

70 Table of RPA probes used in this study.

DNA Linearized Transcribed Probe Protected Reference template with with size fragment size

a-actin EcoRl SP6 ~390bp 250bp Mohun et al, 1984 aT 4-globin EcoRl SP6 319bp 250bp Banville & Williams, 1985 goosecoid Xbal TV 440bp 36Vbp Blumberg et al, 1991 Xnot Hind3 TV ~250bp 192bp von Dassow et al, 1993 Xhox3 Ddel TV 300bp 230bp Saha & Grainger, 1992 XwntS EcoRl TV ~350bp 325bp Smith & Harland, 1991 Xbrachyury Sspl TV 293bp 214bp Smith et al, 1991 ODC Bgl2 TV -120bp 90bp Isaacs et al, 1992 XAG Ndel TV ~200bp 148bp Sive et al, 1989 nrpl Bgl2 T3 ~350bp ~300bp Richter et al, 1988 Endodermin Styl TV ~350bp 311bp Sasai et al, 1996

70a 2.2.9.iii Purification of RNA probes 1 l|il of formamide dyes were added to the transcription mix and heated to ITC for 3 mins prior to loading on a small acrylamide gel (6% acrylamide, 7M urea). The gel was run in IX TBE for 1-1.5 hours at 20 Watts. Next the gel was exposed to X-ray film for 1 min. The film was developed and realigned with the gel in order to locate the band corresponding to the major transcription product for each probe. This band was cut from the acrylamide gel using a scalpel blade. Each acrylamide band was sliced into smaller pieces and the probe eluted at 50°C in 400pl of elution buffer and 40pl lOmg/ml tRNA for 2 hours (or overnight at RT). The gel was pelleted and the supernatant containing the eluted probe transferred to a fresh tube and precipitated with 100% EtOH on dry ice. The probe was pelleted, washed with 70% EtOH and resuspended in 25|il. To determine the amount of radioactivity in the probe, Ipl of the resuspended probe was dotted onto DE81 paper and allowed to dry. This was put into a scintillation vial and counted using Cerenkov counting. The probe was diluted to 1x10^ cpm/pl (no correction was made for Cerenkov counting). Probe was used immediately or stored overnight at -20°C before use.

2.2.9.ÎV Hybridization Hybridization was set up with 10 animal cap equivalents of RNA, 5 VMZ/DMZ equivalents of RNA, or 1 embryo equivalent of RNA. 20pl of hybridization buffer and Ipl (1x10^ cpm) of each appropriate probe were added to each RNA sample. For all probes except a-actin a loading control of ODC was also added to the sample; the a-actin probe recognizes cytoskeletal actin (as well as muscle actin) and so acts as an internal loading control. Often it was possible to add more than one probe to a sample because the size of the protected fragments differed enough that there would be no interference between probes. Thus xwntS and xhox3, gsc and xnot, and nrpl and XAG were hybridized to the same sample. For each probe (or combination of probes) a tRNA control was set up in exactly the same way as described with lOpg of tRNA being used instead of sample RNA. This hybridization mix was heated to 90°C for 5 mins to denature the RNA, spun briefly, and then incubated overnight at 50°C.

2.2.9.V RNase digestion Next morning 200pl of Ambion RNase digestion solution Bx (containing 700U/ml RNase T 1 ; Gibco BRL) was added to the hybridization mix and incubated at 37°C for 30 mins 300pl of solution Dx (Ambion) and lOOpl 100% EtOH were added and incubated on dry ice until frozen. The protected RNA was pelleted at 4°C and washed in 70% EtOH. The pellet was then wetted with Ipl RNase-free water and resuspended in 5pi of formamide dyes.

Immediately prior to loading on a large acrylamide gel (6% acrylamide, 7M urea; see 2.2.3.Ü) the samples were heated to 90“C. All the sample was loaded on to the gel, which was run in IX TBE at 60 Watts for approximately 2 hours (2.2.3.i).

71 2.2.10 Whole mount in situ hybridization (WISH)

This method is based on that described by Harland (1991) with several modifications.

2.2.10.i Digoxygenin labelling of RNA probes Vector DNA for each transcription template was made, linearized and cleaned to make a template for transcription (2.2.7). The following transcription reaction was set up at RT (all reagents used from Promega unless otherwise stated):

2pl DNA 2pl 1 OmM ATP 2pl lOmM CTP 2pl lOmM GTP 1.3pl lOmM UTP 0.7pl lOmM digoxygenin-11-UTP (Boehringer) 4pl 5X transcription buffer Ipl RNasin 2pl RNA polymerase made up to 20p,l with RNase-free water

The reaction was incubated at 37°C for 3 hours, then Ipl RQ DNase 1 was added to remove DNA template and incubated for a further 30 mins at 37“C. The labelled probe was precipitated with 100% EtOH and 5M AmOAc, then pelleted, washed with 70% EtOH and resuspended in 12pl. The concentration of the probe was determined by spectrophotometry and its integrity checked on a freshly poured agarose gel. The probe was then diluted in hybridization buffer to 0.3 - 0.5pg/ml

2.2.10.Ü Fixation of embryos and explants Embryos or embryo explants were collected at the stage indicated in the text and fixed in MEMFA for 1-2 hours with gentle shaking. They were then washed for 5 mins in 100% EtOH and stored in fresh 100% EtOH at -20°C until required.

2.2.10.iii Prebybridization The embryos or explants were brought to room temperature before rehydrating according to the following scheme:

75% EtOH/25% P-Tw 5 mins 50% EtOH/50% P-Tw 5 mins 25% EtOH/75% P-Tw 5 mins P-Tw 3x5 mins

The embryos were then treated with lOpg/ml Proteinase K in IX PBS for 5 mins (explants and gastrulae) or 10 mins (tailbud) with gentle shaking in order to permeablize them and increase sensitivity. Next the embryos were rinsed in O.IM Triethanolamine pH 72 7.8 for 5 mins with gentle shaking, then replaced with fresh O.IM Triethanolamine (5ml). After 5 mins of gentle shaking 12.5|il of acetic anhydride was added and after a further 5 mins another 12.5|l l 1 acetic anhydride was added and gently shaken for 5 mins This treatment blocks positively charged groups which would otherwise bind the RNA probe by ion exchange (Hayashi et al., 1978). The embryos were then washed 2x5 mins, in P- Tw with gentle shaking, refixed in 10% formalin for 20 mins and finally rinsed 5x5 in P- Tw. After the last wash most of the P-Tw was removed (leaving 1ml P-Tw behind) and 250|il of hybridization buffer was added. Once the embryos had settled through the viscous hybridization buffer the mixture was replaced by 1ml fresh hybridization buffer. Embryos were then transferred to 60°C. After 10 mins the hybridization buffer was replaced with fresh pre-warmed hybridization buffer and the samples were pre-hybridized at 60°C for at least 6 hours with gentle shaking.

2.2.10.iv Hybridization After pre-hybridization the buffer was replaced by 0.5ml-1ml of the appropriate probe made up to 0.3|Lig/ml {a-actin and aT4-globin) or 0.5p,g/ml (other probes) in hybridization buffer and hybridized overnight at 60°C with gentle shaking. The next day the probe was removed at stored at -70“C for future use.

2.2.10.V Immunohistochemistry The samples were then washed according to the following scheme (where necessary solutions were prewarmed to 60°C):

1ml Hybridization buffer 60“C 2x10 mins 1ml 2X SSC/0.1% Tween 20 60"C 3 X 30 mins

1ml 0.2X SSC/0.1% Tween 20 60°C 2 X 30 mins 1ml MAB RT 2x15 mins 1ml 2% block/MAB RT 2 hour 1ml 2% block/2% serum/MAB RT 2 hour

Finally the samples were incubated in 1ml antibody solution (1/2000 dilution antidigoxygenin FAB fragments in 20% serum/2% block (Boehringer Mannheim) made up in MAB) over night at 4°C. The next day, antibody solution was removed and the samples rinsed for 5 mins in MAB at RT. The embryos were then transferred to 'Netwells' (Costar; 15mm Netwell, 74|im mesh) and washed 2x5 mins in MAB, with gentle shaking. Washing was continued at RT throughout the day, replacing MAB every hour and then incubated overnight at 4°C in fresh MAB, with gentle shaking. Next day the MAB was removed and the samples washed twice in 5ml AP buffer for Ix 3 mins then 1x10 mins. This activates the alkaline phosphatase coupled to the anti-digoxygenin antibody, whilst levamisole in the AP buffer inhibits endogenous alkaline phosphatases. The AP buffer was replaced by l-2ml BM purple precipitating solution (Boehringer) and incubated at RT until colour appeared. When colour had sufficiently developed, the colour reaction

73 was stopped by rinsing the samples twice in P-Tw for 5 mins each rinse. Samples were then bleached in clear plastic bijoux with 1% H202/5% formamide/lX PBS on top of a light box for approximately 3 hours, in order to see staining more clearly. The embryos were then rinsed twice in P-Tw for 5 min and refixed in 10% formalin overnight.

2.2.11 Histology

2.2.11.1 Fixation of embryos and explants Embryos or embryo explants were collected at the stage indicated in the text and fixed overnight in 4% paraformaldehyde at 4°C. Embryos that had been processed for whole mount in situ hybridization were embedded immediately without further fixation or borax carmine staining.

2.2.11.11 Staining and embedding Fixed embryos or explants were placed in a saturated solution of borax carmine overnight. Next morning embryos were rinsed briefly in 70% EtOH/1% HCl, then left to differentiate in fresh 70% EtOH/1% HCl for 3-6 hours. This destains the cytoplasm and leaves nuclei bright red. The embryos or explants were then embedded in paraffin wax according to the protocol below.

temperature whole embrvos explants 80% IMS RT 30 mins 10 mins 80% EtOH/BuOH 3:1 RT 30 mins 10 mins 90% EtOH/BuOH 1:1 RT 30 mins 10 mins Absolute EtOH/BuOH 1:3 RT 30 mins 10 mins BuOH RT 1 hour 20 mins BuOH/wax 1:1 60°C 20 mins 10 mins

Wax 60°C 2x15 mins 2 X 10 mins Wax 60°C embed embed

22.ll.iii Sectioning 10pm sections were cut using a manual microtome. Ribbons of sections were floated on water on poly-l-lysine coated slides and dried at 37°C overnight on a hot plate.

2.2.11.iv Counterstaining, de waxing and mounting

Sections were de-waxed and counterstained with picro-blue-black according to the following protocol (overleaf).

74 Histoclear (National Diagnostics) 2 x 2 mins Absolute EtOH 1 min IMS 1 min 95% IMS 1 min 90% IMS 1 min 70% IMS 1 min 50% IMS 1 min Water 1 min Picro-blue-black 1 min Water 2 x 1 min 70% IMS 1 min 90% IMS 1 min 95% IMS 1 min IMS 1 min Absolute EtOH 1 min Histoclear 2 x 1 min

Sections were immediately mounted with DPX mounting medium (Fisons) under coverslips.

2.2.12 Photography

Embryos, explants and dissociated cells were photographed under a dissecting microscope; sections were photographed under a compound microscope. Slides were scanned using a Polaroid Sprint Scan 35 slide scanner. Composite figures were made using Adobe Photoshop 4.0 software. Filters were used to remove any dust or scratches from figures, and in some pictures a uniform background was painted around the embryos or sections.

Autoradiographs, Western blots and Coomassie Blue stained gels were scanned using a Hewlett Packard ScanJet Ilex. These images were imported into Adobe Photoshop 4.0 and composite pictures made. Filters were used to remove any dust or scratches from figures, and in some autoradiographs and Western blots the background was brightened to give better contrast.

75 Chapter 3

OVEREXPRESSION OF A SEA URCHIN BMPl-RELATED METALLOPROTEASE, SpAN, IN XENOPUS EMBRYOS

3.1 Introduction

3.1.1 A conserved system for dorsoventral patterning

Recently it has become clear that dorsoventral patterning in early embryos of Drosophila, Xenopus and the zebrafish {Danio rerio) is mediated by a conserved system of extracellular signals (reviewed by De Robertis and Sasai, 1996; Holley and Ferguson, 1997). This system is clearly ancient, being conserved in both protostomes and deuterostomes, and is likely to be important in the early embryos of many animal species.

In the early Drosophila embryo, dorsoventral fate is specified by a graded activity of the secreted protein Decapentaplegic (Dpp), a TGF(3 superfamily member (Ferguson and Anderson, 1992a). Mutations in the dpp gene result in embryos with reduced or absent dorsal structures, whilst overexpressing dpp causes embryos to become dorsalized (Irish and Gelbert, 1987; Ferguson and Anderson, 1992a). Dpp is expressed in the dorsal 40% of the embryo, complementing the expression pattern of short gastrulation (sog) in the ventral 60% of the embryo, which inhibits Dpp activity (Ferguson & Anderson, 1992b). An old idea proposed in 1822 by Geoffrey St-Hilaire, and which has recently gained support, is that vertebrates and arthropods have a common body plan but an inverted dorsal-ventral axis (reviewed by Hogan, 1995; De Robertis and Sasai, 1996). Thus in Xenopus embryos BMP4, a structural and functional homologue of Dpp (Padgett et al., 1993; Sampath et al., 1993; Holley et al., 1996), is expressed in the ventral-lateral mesoderm and specifies ventral fate (Fainsod et al., 1994; Dale et al., 1992; Jones et al., 1992a), and the Sog homologue Chordin, which inhibits BMP4 activity, is expressed in the dorsal organizer (Sasai et al., 1994; Francois and Bier, 1995; Piccolo et al., 1996). In vitro Chordin is able to bind BMP4 (and BMP2), preventing interaction with its receptors and inhibiting signalling (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). A similar situation exists in zebrafish where overexpression of zbmp2 causes a ventralized phenotype (Nikaido et al., 1997) and a mutation in the zb m p l gene, swirl, causes dorsalization (Kishimoto et al., 1997). In addition, genetic studies also demonstrate that Chordino, a Chordin homologue expressed in the zebrafish organizer (the shield), inhibits swirl activity (Hammerschmidt et al., 1996a; Hammerschmidt et al., 1996b; Schulte-Merker et al., 1997).

76 As well as these extracellular signals, many of the components of the dpp/BMP4 signalling pathway, including receptors and intracellular signalling molecules, are structurally and functionally conserved between Drosophila, Xenopus, and many other species (for reviews see Massagué, 1996; Massagué et al., 1997). In Drosophila this signalling pathway has an additional level of regulation, upstream of Dpp. There is genetic evidence that Tolloid, a BMP 1-like metalloprotease, enhances the activity of Dpp, possibly by releasing Dpp from an inactive complex with Sog (discussed in Ferguson & Anderson, 1992b; Holley et al., 1996). This raises the possibility that BMP 1-related metalloproteases may regulate the activity of BMP4 homologues, and other TGpp superfamily members, in other species.

3.1.2 BMFl-related metalloproteases in Xenopus development.

Xenopus bmpl (xbmpl) and a closely related metalloprotease gene named xolloid (xld), were recently isolated by this lab from a Xenopus oocyte cDNA library (Goodman et al., 1998; Appendix 1). xbm pl had previously been described as a partial cDNA isolated from a tailbud cDNA library by Maeno et al. (1993). In addition a long splice variant of the xbm pl gene, xtolloid (xtld), has recently been described (Lin et al., 1997).

Both xb m p l and xld transcripts are expressed maternally and throughout development of the embryo, although transcript levels drop during gastrula and neurula stages before rising again at tailbud stages (Goodman et al., 1998). At blastula and early gastrula stages both xb m p l and xld are expressed uniformly throughout the embryo. Thus both XBMPl and Xld are present at gastrulation when BMP4 is active in mesodermal patterning. Overexpression of XBMPl and Xld by mRNA injection into Xenopus embryos results in tadpoles with anterior truncations and dorsal axial defects such as loss of notochord (Goodman et al., 1998). In addition, XBMPl and Xld are able to ventralize dorsal mesoderm induced in animal caps by Activin. This phenotype is similar, albeit weaker, to that seen when bmp4 mRNA is injected into Xenopus embryos, and may indicate that XBMPl and Xld act to ventralize mesoderm by enhancing endogenous BMP activity.

3.1.3 BMFl-related metalloproteases and BMF signalling in sea urchin embryos

The sea urchin metalloprotease, SpAN, was isolated from Strongylocentrotus purpuratus (Reynolds et al., 1992) in a screen for genes expressed at the very early blastula stage. The structure of the C-terminal domain of SpAN differs from most other BMP 1-related metalloproteases and most closely resembles the Paracentrosus lividus protein BP 10 (LePage et al., 1992). Based on structural and expression data it is likely that SpAN and BPIO are the same gene from different sea urchin species. Transcripts of SpA N and BP 10 are detected soon after fertilization and by blastula stages expression is seen in the animal three quarters of the embryo, which will give rise mainly to ectoderm and to a lesser extent endoderm and secondary mesenchyme, but are completely absent from the most vegetal

77 regions of the embryo which will give rise to endoderm and primary mesenchyme cells (Reynolds et al., 1992).

As described in Chapter 1, a BMP signalling pathway may be functioning during early sea urchin development and several putative components of the pathway have been isolated. A BMP2/4-related molecule has been identified in the early embryos of both Lytechinus variegatus and S. purpuratus (C. Logan and D. McClay, unpublished; L. Angerer, pers. comm.) and in addition, three BMP 1-like metalloproteases have been isolated: SpAN, BPIO and suBMPl (Reynolds et al., 1992; LePage et al., 1992; Hwang et al., 1994). Expression of Xenopus BMP4 (XBMP4) or suBMP2/4 in S.purpuratus or Lytechinus pictus embryos, by microinjection of mRNA, causes defects in both the animal-vegetal and dorsal-ventral axes, such that formation of vegetal derived structures is suppressed and ectoderm becomes more aboral in character (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted). A similar phenotype is produced when SpAN is overexpressed in sea urchin embryos, a result consistent with SpAN activating BMP signalling. In contrast, injection of mRNA for Noggin, the BMP4 inhibitory binding protein, causes a vegetalized phenotype, the opposite phenotype to BMP4 overexpression. This suggests that Noggin is able to recognize, and inhibit, an endogenous BMP in the sea urchin. Although a Noggin homologue has not been identified in sea urchins as yet, this evidence suggests that inhibitory binding proteins are able to function in the sea urchin embryo and may regulate BMP activity during development.

Since the effects of perturbing the BMP signalling pathway in Xenopus embryos are well characterized, this is an ideal system in which to investigate any role for BMP 1-related metalloproteases in regulating BMP signalling. Experiments that perturb BMP signalling are described in this thesis, which make use of different BMP signalling pathway components, such as dominant-negative receptors and inhibitory binding proteins. Although these constructs are active in Xenopus embryos, they have no phenotype when expressed in the sea urchin embryo (except Noggin), probably because the amount of mRNA that would be required to elicit a phenotype is toxic to the sea urchin. The X enopus embryo therefore has an additional advantage over the sea urchin as a model system in which to study regulation of the BMP signalling pathway. This chapter aims to test the effect of overexpressing a sea urchin BMP 1-related metalloprotease, SpAN, in the Xenopus embryo by microinjection of the mRNA.

78 3.2 Results

3.2.1 Injection of mRNA for the sea urchin metalloprotease, SpAN, suppresses dorsal differentiation in Xenopus embryos

To investigate the action of SpAN, Xenopus embryos were injected at the 1-cell stage with varying concentrations (13 pg - 1.6 ng) of SpA N mRNA and the development of these embryos observed. As a control, embryos were injected with mRNA encoding a mutated form of SpAN, ASpAN, in which the glutamic acid (glu^^l) of the metalloprotease active site is converted to alanine. This mutation has no biological activity in sea urchin embryos (Lynne Angerer, pers.comm.).

Embryos injected with 1.6 ng SpAN vriKNA develop normally until mid-gastrulae, forming a dorsal blastopore lip at the beginning of gastrulation (compare figure 3.1 A and 3.IB) which extends laterally and ventrally as normal. However, at mid-gastrulae the development of SpAN mRNA-injected embryos begins to slow compared to sibling control embryos and by late gastrulae (stage 12.5), when the yolk plug of sibling embryos has almost invaginated (figure 3.1C), the yolk plug of S pA N mRNA-injected embryos is still large (figure 3. ID). This effect is concentration dependent in that gastrulation movements of embryos receiving lower doses of SpAN mRNA are less retarded; in those embryos that received 13 pg SpA N mRNA there is no discernible difference in the rate of yolk plug invagination compared to sibling control embryos (not shown). Embryos injected with 1.6 ng ASpA N mRNA develop normally up to and throughout gastrulation. Histological sections of embryos injected with 1.6 ng SpA N mRNA show that at the start of gastrulation there is little difference between sibling controls and S p A N mRNA-injected embryos (figure 3.IE and F), in that bottle cells have formed on the dorsal side of both embryos (figure 3.1 G and H). However, when compared to control gastrulae at stage 12.5 (figure 3.11), the mesoderm of SpA N mRNA-injected embryos does not appear to have migrated normally through the blastopore and along the inner surface of the embryo (figure 3.1J).

Injected embryos allowed to develop to tailbud stages were scored for dorsoanterior defects using the DAI scoring system of Kao & Elinson (1988; see materials and methods), where normal embryos are given a score of 5 and completely ventralized embryos are given a score of 0. Embryos injected with 1.6 ng SpA N mRNA appear to be almost completely ventralized at tailbud stages (figure 3.2B), having an average DAI score of 0.11 (table 3.1). Again the effect is concentration dependent, since five-fold dilutions in the concentration of injected SpA N mRNA results in the development of progressively more dorsoanterior structures. Embryos receiving the lowest dose of SpA N mRNA (13 pg) have an average DAI score of 3.55 (table 3.1). Control embryos injected with 1.6 ng of A SpA N mRNA are normal (figure 3.2A) and have a DAI score of 5.00, and uninjected sibling embryos developed to tailbud stages with an average DAI score of 4.95 (table 3.1).

79 Histological analysis on SpAN mRNA-injected embryos given a DAI score of 0 confirmed that dorsal structures such as neural tube, notochord and somites fail to differentiate in these embryos (figure 3.2E). Sections of SpAN mRNA-injected embryos given intermediate DAI scores show that dorsal structures are reduced and the ventral domain expanded as the embryos become more ventralized (figure 3.2, compare C with D).

3.2.2 Dorsal injection of SpAN gives more extreme defects than ventral injection

If SpAN is ventralizing the mesoderm of Xenopus embryos then a stronger phenotype would be expected when SpAN mRNA is localized to the dorsal rather than the ventral side of the embryo, an effect that is seen with BMP4 (Dale et al., 1992) and with XBMPl and Xolloid (Goodman et al. 1998). When 64 pg of SpAN mRNA is injected into a single ventral blastomere at the 4-cell stage, the average DAI score of the resulting embryos is 4.43, while injecting the same amount of SpAN mRNA into a dorsal blastomere at the 4- cell stage results in tailbud embryos with an average DAI score of 2.53 (table 3.2; figure 3.3, compare A with B). As expected localized injection of ASpAN mRNA has no effect on development (table 3.2). Thus, the effect of SpAN is more potent when localized to the dorsal side rather than the ventral side.

3.2.3 SpAN suppresses dorsal marker gene expression.

SpAN mRNA-injected embryos were next analyzed for the expression of dorsal and ventral mesodermal markers in tailbud embryos, to confirm that dorsal differentiation is suppressed. Embryos injected with either 13 pg or 64 pg of SpAN mRNA were analyzed by whole mount in situ hybridization (WISH) with probes for a-actin (Mohun et al., 1984; figure 3.4A) or type II collagen (Amaya et al., 1993; figure 3.4B), which under the conditions used here stain somites (muscle) and notochord respectively, or a probe for aT4-globin (Banville and Williams, 1985; figure 3.4 C, D and E) which is a marker for ventral blood islands. The results confirm that dorsal mesodermal tissues are progressively lost with increasing severity of the ventralized phenotype, and show that there is also a slight expansion of the domain of aT4-globin expression on the ventral side of the embryo.

RNase protection analysis (RPA) was also used to look at the expression of a-actin and aT4-globin in tailbud embryos injected with 1.6 ng of either SpAN or ASpAN mRNA. Figure 3.4F shows that SpAN mRNA injected embryos (lane 3) do not express muscle- specific a-actin , and when levels of the loading control, ODC, are taken into account there is a concomitant increase in aT4-globin.

Taken together these results demonstrate that expression of SpAN in Xenopus embryos eliminates dorsal structures in a dose dependent manner, such that dorsoanterior structures are completely absent in the most severely effected tailbud embryos.

80 3.2.4 Injection of mRNA encoding the metalloprotease domain of SpAN disrupts morphogenetic movements.

To test that the phenotype observed after SpAN mRNA injection was not due simply to a 'rampant protease' activity indiscriminately degrading proteins in the embryo, Xenopus embryos were injected at the 1-cell stage with an mRNA encoding a mutated form of SpAN, smp (SpAN metalloprotease domain). In this construct a stop codon was introduced after the metalloprotease domain producing a protein which contains no CUB or EOF repeats, thought to be important in substrate recognition, but which retains an active metalloprotease domain (see materials and methods).

Embryos injected with 1.6 ng of smp mRNA develop normally until mid-gastrulae when development starts to slow in a similar way to SpAN mRNA-injected embryos (see figure 3.ID), in that yolk plug invagination is retarded (figure 3.5, compare A with B). By tailbud stages, smp mRNA-injected embryos appear truncated (figure 3.5D) and have an average DAI score of 1.00 (table 3.3). This effect is concentration dependent since embryos receiving lower doses of smp mRNA (64 pg and 13 pg) have less severe anterior and dorsal axial defects (table 3.3; figure 3.5 E and F). By stage 40, however, most embryos injected with 1.6 ng of smp mRNA, contain melanocytes (figure 3.5G, me), indicating that dorsal neural tissue has differentiated. Fifteen of these embryos were sectioned (figure 3.5H) and the results show that, although disorganized, they contain a full range of dorsal structures, neural tube (100%), notochord (67%) and muscle (100%).

RPA analysis confirmed that these embryos contain dorsal mesoderm and are not ventralized (figure 3.51). smp mRNA-injected embryos (lane 2) do not show a reduction in a-actin expression compared to controls (lane 1), although interestingly these embryos show a decrease in aT4-globin expression. This decrease in aT4-globin expression in smp mRNA-injected embryos may be due to their abnormal morphology which brings organizer tissue in to closer contact with ventral mesoderm than in normal embryos, for instance notochord is often seen winding through the centre of the embryo. Since notochord is able to dorsalize ventral mesoderm (Dale and Slack, 1987b), levels of aT4-globin would be reduced.

These results demonstrate that the metalloprotease domain of SpAN is capable of disrupting morphogenetic movements during gastrulation, however, smp mRNA-injected embryos are not ventralized. This indicates that blocking morphogenetic movements at gastrulation does not cause ventralization per se, and that mesodermal patterning can be separated from morphogenesis. In addition, the results show that DAI scores at tailbud stages are not a reliable indication of ventralization unless backed by additional histological or molecular data.

81 3.2.5 SpAN ventralizes Xenopus embryos during gastrula stages

The phenotype of SpAN mRNA-injected embryos closely resembles that of embryos injected with bmp4 mRNA (Dale et ah, 1992; Jones et ah, 1992); in both cases a dorsal blastopore lip forms at the beginning of gastrulation but dorsal structures do not develop. A recent study (Jones et ah, 1996) has shown that early markers for dorsal mesoderm such as goosecoid {gsc\ Blumberg et ah, 1991) andXnor (von Dassow et ah, 1993) are activated as normal at the start of gastrulation in bmp4 mRNA-injected embryos, but expression is soon down-regulated. To determine if SpAN has a similar effect, Xenopus embryos were injected at the 1-cell stage with 1.6 ng SpAN, ASpAN or smp mRNA and allowed to develop to gastrula stages. The expression of both dorsal and ventral mesodermal markers was then analyzed in whole embryos by WISH and by RPA in explanted dorsal marginal zones (DMZs).

3.2.5.1 Early dorsal markers RPA analysis of explanted DMZs shows that in early gastrulae (stage 10.5) both Xnot and gsc are expressed at normal levels in SpAN mRNA-injected embryos, but that both genes are down-regulated relative to control DMZs at mid-late gastrulae (stage 12-13; figure 3.6A).

Down-regulation of Xnot was confirmed by WISH analysis of SpAN mRNA-injected gastrulae. In control early gastrulae Xnot is expressed throughout the marginal zone with stronger staining on the dorsal side, but by the end of gastrulation expression becomes restricted to the future notochordal cells (figure 3.7A; see also von Dassow et ah, 1993). Xnot is also expressed throughout the marginal zone of SpAN mRNA-injected early gastrulae, although expression is more diffuse than in controls and stronger dorsal staining is not seen (not shown). However, Xnot expression is greatly reduced or absent in SpAN mRNA-injected mid gastrulae (figure 3.7B). Xnof expression is detected by WISH of smp mRNA-injected mid-gastrulae, although expression is frequently spread out along the dorsal marginal zone rather than in a tight domain (figure 3.7C). This abnormal expression pattern is probably due to disrupted morphogenetic movements.

WISH analysis shows that gsc is also switched on as normal at the dorsal blastopore lip of SpAN mRNA-injected early gastrulae, but in contrast to Xnot, gsc expression could still be detected at mid-gastrulae (stage 12). However, whereas in control gastrulae gsc expressing cells move anteriorly with the migrating prechordal mesoderm (figure 3.7D), in SpAN mRNA-injected mid-gastrulae gsc expressing cells remain localized to the dorsal blastopore lip (figure 3.7E).

By RPA gsc is expressed at much lower levels in SpAN mRNA-injected embryos compared to controls at stage 12-13, however, WISH analysis shows strong expression of gsc in SpAN mRNA injected embryos at stage 12. These results are reproducible and it

82 maybe that where gsc expression has spread out along the dorsal axis in control embryos, cells expressing gsc in SpAN mRNA-injected embryos remain concentrated at the dorsal lip and consequently expression appears stronger. Also, at longer exposures (20 hours as compared to 4 hours) some residual gsc expression is seen by RPA in SpAN mRNA- injected embryos at stage 12-14, but it is lower than that observed in control embryos even after taking into account differences in levels of the loading control ODC.

3.2.5.Ü Early ventral markers The expression of two early markers for ventral-lateral mesoderm, XhoxS (Ruiz i Altaba & Melton, 1989) and XwntS (Christian et al., 1991) was also examined. RPA analysis shows that the expression domain of both genes is expanded into the dorsal mesoderm of mid- gastrulae injected with SpAN mRNA, but not early gastrulae (figure 3.6A). XwntS is expressed at a similar level in the dorsal mesoderm of SpAN mRNA-injected embryos as in normal ventral mesodermal explants (figure 3.6B). This expansion of the XwntS expression domain into the dorsal mesoderm of mid-late gastrulae is also seen by WISH in 9/20 SpAN mRNA-injected embryos (compare figure 3.7F and G). In contrast, WISH analysis of smp mRNA-injected embryos shows the XwntS expression domain is not expanded into the dorsal side (figure 3.7H), indicating that smp mRNA-injected embryos are not ventralized at gastrulation.

3.2.5.iii Other mesodermal markers The expression of two additional mesodermal markers, Xbrachyury (Xbra; Smith et al. 1991) and Xpo (Sato and Sargent, 1991), supports the conclusion that the ventralizing effect of SpAN is exerted during mid-gastrulation.

Xbra is a pan-mesodermal marker during early gastrulation but becomes restricted to future notochord cells once mesoderm has invaginated (Smith et al., 1991). Since embryos injected with high concentrations of SpAN mRNA lack notochord (see figure 3.2B), a loss of the notochordal component of Xbra expression was expected in these embryos. WISH analysis shows that at low doses of SpAN mRNA (64 pg/embryo), which results in embryos that are only partially ventralized, Xbra expression in the dorsal axis of late gastrulae is often less strong when compared to controls and in some cases there is no expression at all on the dorsal side (figure 3.SC). At higher doses of SpAN mRNA (1.6 ng/embryo) the domain of Xbra expression is expanded towards the animal pole (figure 3.SB) when compared to control late gastrulae (figure 3.SA). RPA analysis of whole embryos, however, shows that when compared to controls the level of Xbra expression in SpAN mRNA-injected embryos during gastrulation is not significantly changed, and may even be reduced when levels of the loading control, ODC, are taken into account (figure 3.6C).

Finally, the expression of Xpo was examined, which . io late gastrulae and neurulae is expressed in the posterior mesoderm and ectoderm of normal Xenopus embryos (figure

83 3.8D; Sato and Sargent, 1991). SpAN mRNA-injected early neurulae show an expansion of the Xpo expression domain into the animal hemisphere when compared to controls (figure 3.8E), and together with the above data suggests that the character of mesoderm in 5pAA mRNA-injected embryos is posterior-ventral.

These results show that dorsal and ventral genes are initially expressed as normal in SpAN mRNA-injected early gastrulae. Subsequently, dorsal mesodermal genes are down- regulated and ventral mesodermal genes are up-regulated on in the dorsal side of mid-late gastmlae, showing that SpAN ventralizes dorsal mesoderm during gastrulation. Since smp mRNA-injected embryos are not ventralized at these stages, it is unlikely that the disrupted gastrulation movements seen in SpAN mRNA-injected embryos causes ventralization.

3.2.6 SpAN blocks Activin induced morphogenetic movements and dorsal mesoderm formation in isolated animal caps

Overexpression of BMP4, BMPl or Xolloid blocks formation of dorsal mesoderm in animal pole explants (animal caps) treated with Activin (Dale et al., 1992; Jones et al., 1992a; 1996b; Goodman et al., 1998), which is a potent inducer of mesoderm in this system. Since injection of SpAN mRNA ventralizes Xenopus embryos, and disrupts morphogenetic movements (see 3.2.1.), SpAN may also inhibit dorsal mesoderm formation and associated morphogenetic movements induced by Activin in isolated animal caps. To test this animal caps were isolated from mid-blastulae (stage 8) injected with 1.6 ng SpAN, ASpAN or smp mRNA, or from uninjected embryos. The caps were then treated with human Activin A conditioned medium and cultured until sibling control embryos had reached stage 26-27 (tailbud).

While animal caps round up and form atypical epidermis in the absence of Activin (figure 3.9 A, B and C), Activin treated caps from control embryos (uninjected and ASpAN mRNA-injected) undergo extensive morphogenetic movements and elongate, consistent with dorsal mesoderm formation (figure 3.9 D, E and G). In contrast, Activin treated caps from SpAN or smp mRNA-injected embryos do not undergo these morphogenetic movements (figure 3.9 F and H) and those from SpAN mRNA-injected embryos form ovoid structures, often with fluid-filled vesicles, which is consistent with ventral mesoderm formation.

RPA analysis shows that Activin treated caps from control embryos express muscle- specific a-actin, a marker for dorsal mesoderm, but not aT4-globin, a marker for ventral mesoderm (figure 3.10A, lanes 2 and 4). On the other hand, Activin treated caps from SpAN mRNA-injected embryos express little a-actin and large amounts aT4-globin (figure 3.10A, lane 6). Activin-treated animal caps from smp mRNA-injected embryos do not show a marked decrease in a-actin expression when compared to controls (figure

84 3.1 OB, compare lanes 1 and 2), indicating that these caps contain dorsal mesoderm, despite not undergoing the morphogenetic movements associated with dorsal mesoderm formation.

These results demonstrate that both SpAN and Smp block dorsal morphogenetic movements, but only SpAN is able to ventralize dorsal mesoderm induced by Activin in animal caps.

3.2.7 SpAN enhances the activity of BMP4

To test whether SpAN is able to enhance the activity of BMP4 when coexpressed, embryos were injected at the 1-cell stage with SpAN mRNA or xbmp4 mRNA or a combination of both. At tailbud stages embryos injected with 16 pg of SpAN or xbmp4 mRNA have an average DAI score of 3.88 (table 3.4), whereas those injected with \6 SpAN and 16 pg xbmp4 mRNA have an average DAI score of 2.62 (table 3.4). This enhancement equates to a five-fold increase in the concentration of bmp4 mRNA (Dale et al., 1992), or a five-fold increase in the concentration of 5pAAmRNA (table 3.1).

3.3 Discussion

A recent study (Goodman et al., 1998) has shown that both Xenopus BMPl and the closely related metalloprotease, Xolloid, are able to partially ventralize Xenopus embryos when overexpressed, a phenotype consistent with the BMP signalling pathway being regulated by metalloproteases in early development. This chapter investigated the effects of overexpressing SpAN, a sea urchin BMP 1-related metalloprotease, in Xenopus embryos and provided some evidence that SpAN may enhance endogenous BMP activity, suggesting that a BMP pathway may also act in sea urchins.

3.3.1 SpAN disrupts morphogenetic movements

The initial phases of Xenopus gastmlation are not affected by SpANvciRNA injection, with the first signs of abnormality occurring in mid-gastrulae. SpAN mRNA-injected gastrulae do not undergo the normal dorsal gastmlation movements responsible for elongation along the anterior-posterior axis, and this is most clearly seen by the expression of gsc in mid- gastmlae; whereas gsc expressing cells migrate towards the animal pole in control embryos, they remain localized to the dorsal lip in SpAN mRNA-injected embryos. Furthermore, animal caps expressing SpAN and treated with Activin do not undergo the convergent-extension movements and consequent elongation normally induced by Activin in isolated animal caps.

Injection of smp mRNA, which encodes the SpAN metalloprotease domain alone, also disrupts morphogenetic movements in a similar way to SpAN, for instance gastmlation movements are delayed, tadpoles are severely tmncated and Activin-treated caps do not

85 undergo convergent-extension movements. Despite this, smp mRNA-injected embryos and Activin-treated animal caps differentiate dorsal mesoderm and express dorsal marker genes. Gastmlation can also be dismpted by injecting antibodies against ECM components, such as Fibronectin, into the blastocoel of Xenopus embryos (Boucaut et al., 1984; Ramos and DeSimone, 1996; Ramos et al. 1996), and it is likely then that SpAN dismpts morphogenetic movements by acting on the ECM. The Smp constmct does not contain CUB or EOF repeats, which have been implicated in protein recognition and interaction (Bond and Beynon, 1995). Since Smp dismpts morphogenetic movements, this component of SpAN’s phenotype is likely to be due to the non-specific, i.e. non-targeted, effect of SpAN’s metalloprotease domain on ECM components. This possibility is tested in the next chapter (Chapter 4).

These data also indicate that a full range of dorsal mesodermal tissue types can form in the absence of normal morphogenetic movements, another example of which is provided by overexpression Xwnt5A in Xenopus embryos (Moon et al., 1993). XwntSA mRNA- injected animal caps do not elongate when treated with Activin, despite differentiating dorsal axial mesoderm (Moon et al., 1993). Gerhart et al. (1989) suggested that the most anterior structures of the Xenopus embryo will not form if mesoderm is not allowed to migrate as normal. Observations from this study and from Goodman et al. (1998; Appendix 1) support this view. Firstly, embryos injected with the highest concentration of smp mRNA do not contain recognizable head structures such as eyes, so although disruption of morphogenesis does not block dorsal differentiation, it does eliminate the most anterior structures of the embryo. In addition, when Xenopus embryos are injected with XwntSA mRNA, which is known to disrupt morphogenetic movements, they develop with mild anterior truncations (Moon et al., 1993), a phenotype very similar to that seen when xbmpl or xld mRNAs are overexpressed in Xenopus embryos (Goodman et al., 1998). Anterior truncations seen in xbmpl and xld mRNA-injected Xenopus embryos do not extend beyond the hindbrain, as indicated by Krox20 expression (Goodman et al., 1998), and it would be interesting to ascertain how far truncations in smp mRNA-injected embryos extend since this may indicate which anterior structures are dependent on full mesodermal migration.

Another explanation for the retarded gastmlation movements seen in SpAN, xbmpl or xld mRNA-injected embryos may be due to the change in fate of cells from dorsal to ventral (see below). An example of this is that animal caps treated with fibroblast growth factor (FGF), which induces ventral mesoderm (Slack et al., 1987), do not elongate and treated cells do not spread on Fibronectin (Howard and Smith, 1993). In contrast, animal caps treated with Activin, which induces dorsal mesoderm, undergo extensive morphogenetic movements and elongate. Thus, the morphological defects seen in embryos injected with SpAN, xbmpl and xld mRNA may be due, in part, to the ventral character of the mesoderm causing a decrease the motility of cells.

8 6 3.3.2 SpAN ventralizes Xenopus embryos

As well as disrupting morphogenesis, SpAN acts to ventralize dorsal mesoderm when overexpressed in Xenopus embryos and animal caps. At high concentrations SpAN mRNA-injected embryos fail to differentiate dorsal structures, such as notochord, neural tissue and muscle, and exhibit greatly reduced expression of dorsal marker genes. In addition SpAN mRNA injection blocks the formation of dorsal mesoderm in animal caps treated with Activin, causing ventral mesoderm to form instead. Analysis of gastrulae expressing SpAN shows that dorsal markers are activated as normal at the start of gastmlation but are subsequently down-regulated, as the expression of ventral markers is expanded into the dorsal side of the embryo. This indicates that ventralization occurs during gastmlation rather than earlier, during mesoderm induction. This is also supported by the observation that the dorsal blastopore lip forms at the start of gastmlation and extends laterally and ventrally as normal in SpAN mRNA-injected embryos, but that yolk plug invagination is subsequently delayed.

SpAN’s phenotype is much stronger than that seen when XBMPl or Xld are overexpressed in Xenopus embryos. Part of this may be because the metalloprotease domain of SpAN is much more active thanXBM PI or Xld, since injection of smp mRNA causes morphological defects whereas injection of the metalloproteases domain of XBMP1 has no effect on Xenopus development (Goodman et al., 1998). Another explanation is that SpAN has a different arrangement of CUB and EOF repeats in its C-terminal domain when compared to XBMPl and Xld, which may confer a different and perhaps wider substrate specificity. In addition XBMPl and Xld, being Xenopus proteins, may be subject to tighter post-transcriptional, and/or post-translational controls from endogenous factors than SpAN, which is a sea urchin protein.

3.3.3 How does SpAN ventralize dorsal mesoderm?

SpAN's phenotype closely resembles that seen after injection of mRNAs for hmp2, bmp4, bmp7 or the BMP-related anti-dorsalizing morphogenetic protein (Dale et al., 1992; Jones et al., 1992a; Clement et al., 1995; Hawley et al., 1995; Hemmati-Brivanlou and Thomsen, 1995; Moos et al., 1995). It is also similar to, although stronger than, the phenotype seen when xbmpl or xld mRNAs are injected into Xenopus embryos (Goodman et al., 1998). The similarities between the activities of BMP 1-related metalloproteases, including SpAN, and TGPp BMPs when misexpressed in Xenopus embryos, and genetic studies in Drosophila showing the BMP 1-related metalloprotease, Tld, enhances the activity of Dpp (Ferguson and Anderson, 1992b), suggest that SpAN ventralizes Xenopus embryos by enhancing the activity of endogenous Xenopus BMPs. Consistent with this, coexpression of SpAN and XBMP4 gives a stronger phenotype than expression of either molecule alone.

87 Members of the TGFp family form disulphide-linked dimers which must be subsequently cleaved to release the proregion and the mature, bioactive C-terminal domain. One way in which SpAN might enhance BMP activity is to process the protein into a mature form by cleaving its proregion. In TGFp’s the proregion is cleaved at an Arg-X-X-Arg site, which is generally a target for subtilisin-like serine proteases (Barr, 1991). SpAN cleavage sites are not known, and whilst it is possible that SpAN cleaves this Arg-X-X-Arg site and releases mature BMP, it is unlikely since the known cleavage sites for other astacin metalloproteases are very different (see figure 1.6). Although coinjection of SpAN with BMP4 causes more severe anterior truncations than injection of BMP4 alone, this is more likely to be due to SpAN releasing inactive BMPs from inhibitory complexes (see below).

SpAN may release BMP4, or other ventralizing factors, from inhibitory interactions with extracellular matrix (ECM) components. There are many examples of growth factors, including members of the TGFp family, being sequestered by the ECM which require proteolytic cleavage to become active (for reviews see Sarras, 1996; Taipale & Keski-Oja, 1997). For instance BMPs 2-4 bind heparin (Wozney et al., 1988) and BMPs 3 and 7 bind Collagen type IV in basement membranes (Paralkar et al., 1990; Vukicevic et al., 1994). It is possible that SpAN cleaves or degrades ECM components since other astacin metalloproteases are known to do so. Recently, BMPl was shown to be identical to pro- Collagen C-Proteinase (PCP), an enzyme responsible for cleaving the C-terminal peptide from pro-Collagen types I, II and III (Kessler et al., 1996; Li et al., 1996). PCP also processes pro-Lysyl Oxidase (Panchenko et al., 1996), an enzyme required for crosslinking Collagen and Elastin fibres. Moreover, a loss-of-function mutation in the mouse bm pl gene disrupts the normal deposition of Collagen and assembly of basement membrane at the dermal-epithelial junction. These bm pl deficient mice display persistent herniation of the gut, which may be due to disruption of the ECM (Suzuki et al., 1996; Amano et al., 1997). In addition, the crayfish metalloprotease, astacin, degrades mature fibrillar Collagens into small fragments (Stocker and Zwilling, 1995).

A third explanation is that SpAN releases BMPs from other inhibitory interactions, for instance with Chordin, Noggin or Follistatin, which have all been shown to bind BMP4 (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). Recent work by Piccolo et al. (1997) shows that in vitro Xolloid (Xld) is able to specifically cleave Chordin, but not Noggin, in two places and releases bound BMP4/7 heterodimers which were shown to be active. In vivo Xld is able to inhibit the dorsalizing activity of Chordin, but not Noggin, indicating that Xld is cleaving Chordin in the embryo. In addition, Blader et al. (1997) have shown that zebrafish Tolloid is able to cleave Xenopus Chordin, and Marqués et al. (1997) have shown that Drosophila Tolloid cleaves Short gastmlation in at least three places in the presence of Dpp, BMP2 and BMP4. Thus SpAN may also cleave Chordin, and other inhibitory proteins, when expressed in the Xenopus embryo. The possibility that SpAN, XBMPl and Xld release BMPs from inhibitory binding proteins, and act through a BMP signalling pathway is investigated in Chapter 5.

The mutated form of SpAN, ASpAN, in which the glutamic acid within the conserved HEXXH zinc-binding domain of the metalloprotease domain is replaced by an alanine residue, might be expected to act as a dominant inhibitor of endogenous metalloprotease activity by sequestering targets but not processing them. However, the results show that ASpAN has no effect in the embryo and does not act as a dominant-inhibitor. Although it is not clear why this should be, there are several possible explanations. Analysis of mutant tolloid alleles in Drosophila indicate that while most mutations in the metalloprotease domain are antimorphic, this is not always the case and two point mutations which result in amino acid substitutions are null. Thus, mutations within the metalloprotease domain do not always lead to dominant-inhibitory activity. It is possible that the lack of phenotype observed when ASpAN is expressed in Xenopus embryos is due to the protein not being secreted, perhaps because its conformation is changed due to the mutation. While there is evidence that wild type SpAN protein is secreted, it is not known whether this is true of ASpAN (L. Angerer, pers. comm.). A second possibility is that ASpAN is secreted, but its conformation is altered in such a way that the C-terminus is not available to bind any targets. Alternatively, ASpAN may be secreted but be more labile than wild type SpAN and consequently degraded before it is able to exert any effect on development. Determining whether ASpAN is secreted would be the first step in distinguishing between these possibilities.

3.3.4 Regulation of BMP signalling in sea urchins.

SpAN may ventralize Xenopus embryos by activating BMPs, but does SpAN regulate BMP signalling in sea urchin embryos? Injection of Sp A N mRNA has a similar effect to hmp4 mRNA injection in both Xenopus and sea urchin embryos, in both embryos this phenotype is opposite to that of noggin mRNA injection. This indicates that if SpAN is acting in a BMP pathway in Xenopus it is also likely to act in a BMP pathway in sea urchins.

89 Table 3.1 Injection of different amounts of SpAN mRNA causes a graded series of dorsoanterior defects

DAI score SpAN mRNA (pg) 0 1 2 3 4 5 mean n 0 0 0 0 0 6 112 4.95 118 13 6 15 16 7 39 49 3.55 132 64 24 13 17 18 35 16 2.61 123 320 71 23 26 8 24 2 1.33 154 1600 119 15 0 0 0 0 0.11 134

Embryos were injected at the 1-cell stage with the indicated amount of SpA N mRNA and scored at stage 28-30 using the dorso-anterior index (DAI) of Kao and Elinson (1988; see materials and methods). Briefly, a normal embryos scores 5 and a completely ventralized embryo scores 0. Numbers in the table indicate the number of cases at each DAI score, pooled from 4 separate experiments. In the same experiments injection of 1600 pg of ASpAN mRNA gave a mean DAI score of 5.0 (n=81).

90 Table 3.2. Dorsal injection of S p A N mRNA causes more extreme phenotypes than ventral injection

DAI score mRNA Blastomere 1 mean n SpAN Ventral 0 2 7 14 40 4.43 63 SpAN Dorsal 10 15 21 6 5 2.53 60

ASpAN Ventral 50 4.96 52 ASpAN Dorsal 58 4.97 60

64 pg of SpAN or ASpAN mRNA was injected into a single dorsal or ventral blastomere at the 4-cell stage and the resulting embryos were scored at stage 28-30 using the dorso­ anterior index (DAI). Numbers in the table indicate the number of cases at each DAI score, pooled from 2 separate experiments.

91 Table 3.3 Injection of different amounts of smp mRNA causes a graded series of dorsoanterior defects

DAI score smp mRNA (pg) 0 1 2 3 4 5 mean n 0 0 0 0 0 0 22 5.00 22 13 0 0 0 0 1 13 4.93 14 64 1 2 1 1 2 1 2.50 8 1600 12 10 7 1 1 0 1.00 31

Embryos were injected at the 1-cell stage with the indicated amount of smp mRNA and scored at stage 28-30 using the dorso-anterior index (DAI). Numbers in the table indicate the number of cases at each DAI score, pooled from 3 separate experiments.

92 Table 3.4 Coinjection of SpAN mRNA with xbmp4mRNA causes an increase in DAI score

DAI score mRNA 0 1 2 3 4 5 mean n BMP4 0 0 3 1 9 5 3.88 18 SpAN 0 0 2 5 12 6 3.88 25 BMP4 + SpAN 2 2 3 0 4 2 2.62 13 BMP4 + ASpAN 0 0 2 8 9 5 3.71 24

Embryos were injected at the 1-cell stage with the indicated combinations of SpA N , ASpAN and xbmp4 mRNA (16 pg each) and scored at stage 28-30 using the dorso-anterior index (DAI). Numbers in the table indicate the number of cases at each DAI score from 1 experiment, n = total number of embryos scored.

93 Figure 3.1 SpAN exerts its effect during later gastrulation

Injection of 1.6 ng SpAN m R N A at the 1-cell stage does not disrupt the appearance of the dorsal blastopore lip (arrows) or bottle cells at early gastrula stages (stage 10.5), as seen in whole embryos and sections (B, F and H), when compared to uninjected embryos (A, E and G). (G) and (H) picture a close up section of a dorsal blastopore lip from uninjected and mRNA-injected embryos, showing bottle cells, one of which is outlined in each section.

At later stages SpAN mRNA-injected gastrulae (stage 12.5; D and J) exhibit a delay in yolk plug invagination (arrowheads) when compared to controls (C and I). Note the greater size of the vegetal yolk plug in SpA N mRNA injected embryos (compare C with D, I with J). In sections B-I the dorsal side of the embryo is to the right, in section J it is not possible to tell which side is dorsal.

94 Figure 3.1

uninjected SpAN Injected B stage 10.5

D stage 12.5

uninjected SpAN injected E F stage 10.5

. i

stage 12.5

9 5 Figure 3.2 SpAN suppresses differentiation of dorsal structures

Embryos injected with 1.6 ng SpA N mRNA at the 1-cell stage and allowed to develop to tailbud stages (B) appear to lack dorsal stmctures when compared to embryos injected with 1.6 ng ASpAN m R N A (A) and uninjected embryos (not shown).

Histological sections of stage 40 embryos exhibiting varying degrees of ventralization confirms that SpA N mRNA-injected embryos lack dorsal structures. Normal embryos, scoring DAI 5 (C), contain notochord (no), somites (muscle; so) and a neural tube (nt). S p A N mRNA-injected embryos scoring DAI 3 (D) contain muscle, notochord and neural tube which are reduced compared to controls, and have an expanded ventral domain (en = endoderm; compare C with D). Completely ventralized embryos, scoring DAI 0 (E), contain no dorsal structures at all.

Figure 3.3 Dorsal injection of SpAN mRNA causes more severe anterior truncations than ventral injection

Injection of 64 pg Sp A N mRNA into one dorsal blastomere at the 4-cell stage (B) causes more embryos to develop with more severe anterior tmncations (arrows) than does the reciprocal ventral injection (A).

96 Figure 3.2

ASpAN injected SpAN injected Mm

B

DAI 5 DAI 3 DAI 0

Figure 3.3

SpAN: ventral injection SpAN: dorsal injection

97 Figure 3.4 SpAN suppresses expression of dorsal markers

Tailbud embryos previously injected with either 13 pg or 64 pg of SpAN mRNA were hybridized with probes for a-actin (A) or type II collagen (B), which under the conditions used here stain somites (muscle) and notochord respectively, or a probe for blood-specific aT4-globin (C, D and E). The results illustrate the progressive loss of dorsal structures with increasing severity of phenotype, whilst the domain of expression of aT4-globin increases. Sections through embryos stained for aT4-globin, confirm an increase in the expression domain (compare D with E).

These results were confirmed by RNase protection analysis (F). Total RNA isolated from uninjected embryos and embryos injected with either 1.6 ng ASpAN or SpAN mRNA was hybridized with probes for muscle-specific a-actin (m-actin) and blood-specific aT4- globin. Levels of ornithine decarboxylase (ODC) mRNA were determined as a loading control for aT4-globin and levels of cytoskeletal actin (c-actin) for muscle actin. The results demonstrate a significant decrease in expression of a-actin mRNA and a small increase in the expression of aT4-globin in SpAN mRNA-injected embryos (lane 3) compared to controls (lanes 1 and 2)

98 Figure 3.4

muscle actin notochord

T4-globin uninjected SpAN injected D \ E

m-actin

c-actin

T4-globin

9 9 Figure 3.5 Injection of sm p mRNA disrupts morphogenetic movements at gastrulation, but does not affect differentiation of dorsal structures

Embryos injected with 1.6 ng smp mRNA (B) exhibit a delay in yolk plug invagination when compared to control embryos at stage 12 (A), a phenotype similar to SpAN mRNA- injected embryos (compare with figure 3.1 D).

At tailbud stages embryos injected with 1.6 ng smp mRNA exhibit severe anterior truncations (D) when compared to controls (C). At lower doses (E and F) embryos are less severely truncated, such that embryos injected with 13 pg smp mRNA do not have anterior defects (F). Embryos pictured in E and F are older than those pictured in C and D.

At stage 40, many severely affected smp mRNA-injected embryos (G) contain melanocytes (me), and histological sections (F) show that although very disorganized, these embryos contain dorsal structures, such as notochord (no), neural tube (nt) and muscle (mu).

RNase protection analysis confirms these embryos contain dorsal mesoderm (I). Total RNA isolated from uninjected embryos and embryos injected with 1.6 ng smp mRNA was hybridized with probes for muscle-specific a-actin (m-actin) and blood-specific aT4-globin mRNAs. Levels of ornithine decarboxylase (ODC) mRNA was determined as a loading control for aT4-globin and levels of cytoskeletal actin (c-actin) for muscle actin. smp mRNA-injected embryos do not show a decrease in a-actin expression when levels of the loading control are taken into account, although they do express lower levels of aT4-globin compared to controls (compare lanes 1 and 2).

100 Figure 3.5

stage 12 uninjected Smp Injected

B

stage 28-34 uninjected Smp: 1.6ng

&

Smp: 64pg Smp: 13pg

stage 40 I a £ (/) m-actln

c-actin % T4-globin

no ODC

101 Figure 3.6 SpAN ventralizes embryos during gastrulation

Dorsal marginal zones (DMZs) were isolated from early gastrulae of both uninjected embryos and embryos injected with 1.6 ng SpAN mRNA, incubated until sibling control embryos reached the indited stages and then analyzed for the expression of both dorsal- specific, ventral-specific and general mesodermal markers by RNase protection.

The dorsal-specific genes Xnot and goosecoid (gsc) are initially induced in DMZs explanted from SpAN mRNA-injected embryos, but expression is subsequently repressed during gastrulation to levels below those for uninjected DMZs (A). At longer exposures (20 hours as compared to 4 hours) some residual gsc expression is seen by RPA in SpAN mRNA-injected embryos at stage 12-14, but it is lower than that observed in stage 12-14 control embryos even after taking into account differences in levels of the loading control ODC. In contrast to dorsal markers, expression of the ventral-specific genes Xhox3 and XwntS is increased in mRNA-injected embryos as gastrulation proceeds (A). Levels of induced XwntS in the DMZs of 5pAN mRNA-injected embryos reach the same levels as control ventral marginal zone explants (VMZs) from ASpAN mRNA-embryos (B). Although expression of dorsal and ventral mesodermal marker genes is effected by SpAN mRNA injection, the overall level of mesoderm in these embryos stays approximately constant throughout gastrulation, as shown by expression of the pan-mesodermal marker Xbrachyury (Xbm\ C).

102 Figure 3.6

uninjected SpAN injected stage 10.5 12 13 10.5 12 13

Xnot

gsc Xhox3 e# XwntS

ODC

B ASpAN VMZ SpAN DMZ stage 10.5 12 14 10.5 12 14

XwntS

ODC 3 4 5 6

uninjected SpAN injected stage 105 12 13 " 10.5 12 13

ODC # # #

1 2 3 4 5 6

103 Figure 3.7 SpAN alters the expression of dorsal and ventral mesodermal marker genes in late gastrulae

Uninjected embryos (A, D and F), and embryos injected with 1.6 ng SpA N (B, E and G) or smp (C and H) mRNA were analyzed at gastrula stages for expression of Xnot (A, B and C), gsc (D and B) or X w ntS (F, G and H) by WISH. In control mid-gastrulae (stage 12), X not is localized to the dorsal blastopore lip (A), but is absent or very diffuse in mRNA-injected embryos (B), consistent with the loss of dorsal structures. In smp mRNA-injected embryos X not is present, although expression is often spread laterally along the dorsal marginal zone (arrows; C), probably due to abnormal morphogenetic movements in these embryos. At stage 12, cells expressing gsc disperse towards the animal pole in control embryos (D), but remain at the dorsal blastopore lip in Sp A N mRNA- injected embryos (B). This lack of movement may account for the stronger signal in S p A N mRNA-injected embryos when compared to controls. X w ntS is expressed in ventral and lateral mesoderm, but is absent in the dorsal mesoderm, of control mid-gastrulae (arrowhead; F), whereas expression is expanded into the dorsal mesoderm in S p A N mRNA-injected embryos (G), indicating ventralization of the dorsal marginal zone has occurred by this stage. X w ntS expression remains absent from the dorsal side of smp mRNA-injected mid-gastrulae (arrowhead; H). In all photographs, where it was possible to deduce, the dorsal side of the embryo is uppermost.

Figure 3.8 SpAN alters the expression of the mesodermal marker genes, Xbra and Xpo, in late gastrulae and neurulae

Uninjected embryos (A and D) and embryos injected with 1.6 ng or 64 pg S pA N mRNA (B, C and B) were analyzed at gastrula stages for expression of Xbra (A, B and C) or X p o (D and B) by WISH. At stage 12.5 (A) Xbra is normally expressed in mesoderm around the closing blastopore and in the notochord (arrow). In embryos injected with 64 pg S p A N mRNA (C) Xbra expression is seen around the blastopore, except that expression is often absent from the dorsal side, and any notochordal expression is weak (arrow). In embryos injected with 1.6 ng SpAN mRNA the domain of Xbra expression is expanded towards the animal pole and no notochordal expression is seen (B). In control early neumlae (stage 13) X po expression is localized to posterior tissues (D), but is seen throughout the animal hemisphere of Sp A N mRNA-injected embryos (B). In all photographs, where it was possible to deduce, the dorsal side of the embryo is uppermost.

104 Elgure3J

uninjected SpAN injected Smp injected

/ ^ ! Xnot

gsc

% XwntS L,

Figure 3.8

uninjected SpAN injected SpAN injected

Xbra

D E Xpo y w 1.6 ng

105 Figure 3.9 SpAN blocks Activin induction of dorsal morphogenetic movements in isolated animal caps

Embryos were injected at the 1-cell stage with 1.6 ng SpAN, ASpAN or smp mRNA. Animal caps isolated from injected or uninjected mid-biastulae (stage 8) were incubated in the presence or absence of human Activin A conditioned medium until sibling controls had reached stage 26-27. Untreated animal caps round up into balls of epidermis (A, B and C). Uninjected (D) and A SpA N (B) mRNA-injected animal caps treated with Activin undergo extensive morphogenetic movements and elongate. SpAN (F) and smp (G) mRNA-injected animal caps treated with Activin do not elongate, with SpA N mRNA-injected caps forming ovoid structures with fluid-filled vesicles (arrowhead; F).

Figure 3.10 SpAN blocks Activin induction of dorsal mesodermal markers, and elicits expression of ventral markers, in isolated animal caps.

Embryos were injected at the 1-cell stage with 1.6 ng SpAN, ASpAN or smp mRNA. Animal caps isolated from injected or uninjected mid-blastula embryos (stage 8) were incubated in the presence (4-) or absence (-) of human Activin A conditioned medium until sibling control embryos had reached stage 26-27. Total RNA isolated from these caps was hybridized with probes for muscle-specific a-actin (m-actin) and blood-specific aT4-glohin (SpAN only). Levels of ornithine decarboxylase (ODC) mRNA was determined as a loading control for aT4-globin and levels of cytoskeletal actin (c-actin) for muscle actin. The results show that SpAN ventralizes the mesoderm in Activin treated caps, lowering the expression of a-actin and increasing the expression of aT4-globin (A; lane 6) compared to control embryos (lanes 2 and 4). Despite dorsal morphogenetic movements being disrupted in Activin treated caps isolated from smp injected-embryos, levels of a-actin are not significantly effected (B; lane 2), indicating that the caps remain dorsalized.

106 Figure_3^

ASpAN SpAN Smp uninjected injected injected injected untreated

Figure 3.10

■o B o o z "S < .5 ll activin T3 m-actin B B I ■ O Q) c-actin ####» I I activin + + T4-globin m-actin # #

c-actin 1 2 3 4 5 6

107 CHAPTER 4

EFFECT OF BMPl-RELATED METALLOPROTEASES ON EXTRACELLULAR MATRIX AND CELL-CELL INTERACTIONS

4.1 Introduction

Both cell-cell adhesion and cell contact with the extracellular matrix (ECM) are essential for morphogenesis and tissue differentiation during normal embryonic development (reviewed by Adams and Watt, 1993; Hynes, 1996). In addition to being important in mechanical interactions during morphogenesis, cell adhesion molecules and ECM components are also involved in regulating gene expression (reviewed by Hynes, 1992; Boudreau and Bissel, 1996; Gumbiner, 1996; Sastry and Horwitz, 1996; Taipale and Keski-Oja, 1997).

4.1.1 Cell and ECM interactions during embryogenesis

Two main classes of molecule that mediate cell-cell adhesion exist on the surface of cells: (i) the calcium dependent cadherin superfamily, and (ii) the calcium-independent immunoglobulin superfamily. Other molecules expressed on the surface of cells, such as integrins (reviewed by Hynes, 1992), mediate cell-ECM interactions as well as cell-cell adhesion. In addition, many cell surface glycoproteins and glycolipids are known to play a role in cell adhesion and ECM interaction (Rubin et al., 1996). For instance, a calcium- dependent adhesion system involving a group of glycolipids present in blastomere membranes until gastrulation has been described in the early Xenopus embryo (Turner et al. 1992).

Cadherins are required for normal embryonic development from the earliest stages (reviewed by Marrs and Nelson, 1996; Kühl and Wedlich, 1996). In mammals E-cadherin is expressed on all cells of the blastocyst and is required for compaction at the 8 cell stage; early mouse embryos deficient in E-cadherin dissociate, and the blastocyst fails to hatch and implant into the uterine wall (Lame et al., 1994). In sea urchins, two cadherin molecules have been described (Ghersi et al., 1993) which are expressed in the embryo. As might be expected, dismption of cadherin function, using specific antibodies, causes ectodermal cells of the blastula to disaggregate (antibody directed against a 125KDa cadherin) or cause dismption of vegetal plate invagination at gastmlation (anti-E-cadherin antibody). In Xenopus several cadherins have been described which are present during development either maternally (EP-cadherin, U-cadherin, XmN cadherin), or zygotically (E-cadherin, F-cadherin and N-cadherin; reviewed by Kühl and Wedlich, 1996). Using antisense oligonucleotides to deplete the maternal EP-cadherin RNA pool, Heasman et al. ( 1994b) have demonstrated that EP-cadherin is essential in the early X enopus embryo for cell adhesion and normal development. If cadherin levels are depleted then cells dissociate 108 from each other during blastula stages, although zygotic expression of EP-cadherin rescues this early phenotype. Mutant constructs of cadherins, where the cytoplasmic domain has been removed so that interaction with the cytoskeleton is inhibited, have also been used to study the role of cadherins in development (reviewed by Kühl and Wedlich, 1996). For instance, expressing a truncated version of EP-cadherin (also called C-cadherin) disrupts dorsal morphogenetic movements and does not allow the blastopore to close properly so that embryos develop with an opening on the dorsal side, a severe spina bifida phenotype (Lee and Gumbiner, 1995). In addition to cell adhesion, cadherins may regulate cell signalling since they interact with cytoskeletal components and signal transduction molecules, including p-catenin which is involved in dorsal mesoderm induction in Xenopus (reviewed by Miller and Moon, 1996; Gumbiner, 1996).

Cellular interactions with the ECM are mediated in part by members of the integrin family, a large family of heterodimeric transmembrane receptors that function as a mechanical link between ECM and the cytoskeleton and also participate in a variety of adhesion-dependent cell signalling events (reviewed by Sastry and Horwitz, 1993; Clark and Brugge, 1995). In Xenopus several a and P integrin subunits have been described which are expressed during early embryogenesis (Ransom et al., 1993; Whittaker and DeSimone, 1993; DeSimone and Hynes, 1994; Joos et al., 1995; Lallier et al., 1996; Meng et al., 1997). The Pi subunit, at least, is important for gastmlation, since antibodies directed against Pi integrin prevent gastmlation movements subsequent to dorsal lip formation (Howard et al., 1992). The genes for most integrin subunits have now been ‘knocked out’ in mice, leading to many different phenotypes both pre and postnatally (reviewed by Hynes, 1996). For instance, mice carrying a null mutation in the integrin subunit undergo normal gastmlation and embryos develop with normal axial stmctures anteriorly, however, somites do not form posterior to somite 10 and the neural tube is deformed (Yang et al., 1993). The inner cell mass of early pi knockout mice embryos is small and often disorganized compared to controls, and undergoes no further morphogenesis once the trophoblast has implanted (Fassler and Meyer, 1995; Stephens et al., 1995).

One of the best characterized ECM proteins to which integrin binds is Fibronectin. Different integrins recognize distinct regions of the Fibronectin molecule, for instance the

tt5 pi receptor recognizes an Arg-Gly-Asp (RGD) sequence and nearby ‘synergy’ site, which lie close to the centre of the Fibronectin molecule (Aota et al., 1991). In amphibian embryos Fibronectin is localized to the roof of the blastocoel where it comes into contact with involuting marginal zone cells during gastmlation (Boucaut and Darribère, 1983; Lee et al., 1984). Fibronectin and integrin-dependent cell adhesion to Fibronectin is required for normal gastmlation in both Xenopus and the newt, Pluerodeles, since injection of anti- Fibronectin antibodies inhibits gastmlation movements (Boucaut et al., 1984; Ramos and DeSimone, 1996; Howard et al., 1992; Ramos et al. 1996). However, Fibronectin may not be essential for normal gastmlation in all vertebrates, since mesodermal cells migrate as

109 normal through the primitive streak in Fibronectin-deficient mice embryos. Yet Fibronectin is required for the development of mesodermal structures, since notochord and somites do not form in these Fibronectin-deficient mice (George et al., 1993). In sea urchins it is not clear whether Fibronectin is present in the embryo, although primary mesenchyme cell (PMC) migration is inhibited by injection of anti-Fibronectin antibodies or Fibronectin- related synthetic peptides (Katow, 1990; Katow et al., 1990). Recently, a p integrin subunit has been isolated from sea urchins which is expressed throughout development, with a peak in expression during gastmlation, and may mediate cell interactions with Fibronectin and other ECM components (Marsden and Burke, 1997).

4.1.2 ECM and growth factor interactions

As well as acting as physical support for cells, the ECM can also regulate cell proliferation and differentiation by presenting ligands to their receptors, or by acting as a store for growth factors, which may require proteolytic activity for their release (reviewed by Ruoslahti and Yamaguchi, 1991; Lin and Bissell, 1993; Taipale and Keski-Oja, 1997). For instance, prebinding to Heparin-sulphate is required for bFGF to activate its receptor (Yayon et al., 1991), whilst the third ‘receptor’ for TGFp, also a Heparin-sulphate proteoglycan, Betaglycan, is implicated in presenting TGFpl to the type II receptor (Lin and Lodish, 1993). BMPs 2-4 bind Heparin (Wozney et al., 1988) and BMP2 has an increased activity when its N-terminal Heparin binding domain is removed (Ruppert et al., 1996). Other ECM components, including Fibronectin, Decorin, Thrombosporin and Collagen, have all been implicated in binding active and inactive forms of TGF-p (Ruoslahti et al., 1992; Schultz-Cherry and Murphy-Ullrich, 1993), and BMPs 3 and 7 bind Collagen type IV in basement membranes (Paralkar et al., 1990; Vukicevic et al., 1994). In addition BMP activity, as defined by the ability to induce ectopic bone formation, is tightly associated with bone matrix (Wozney, 1989), which consists mainly of Collagen (type I) but also proteoglycans and other proteins. TGFp-related growth factors can also bind to ECM components via an intermediate protein. For instance, latent TGFpl is bound to the ECM by LTBP (Taipale et al., 1994), and Activin binds Heparin-sulphate via follistatin (Nakamura et al., 1991). BMP7 and BMP4 also bind Follistatin (Yamashita et al., 1995; Fainsod et al. 1997), and so may also associate with the ECM via this molecule. It is possible that these growth factors are released from inhibitory ECM interactions by BMP 1-related metalloproteases (reviewed by Sarras, 1996).

This chapter aims to investigate whether the morphological defects caused in Xenopus embryos by overexpression SpAN, Smp, XBMPl and Xld, are a consequence of disruption of cell adhesion and/or ECM degradation. To this end, injected embryos were analyzed for the presence of Fibronectin protein during gastmlation, and the reaggregation of dissociated cells was tested.

110 4.2 Results

4.2.1 SpAN and SMP decrease Fibronectin levels in Xenopus embryos

Gastmlation movements are severely dismpted in amphibian embryos injected with anti- Fibronectin antibodies, often resulting in tmncated embryos resembling those injected with SpAN or smp mRNAs (Boucaut et al., 1984; Howard et al., 1992; Ramos and DeSimone, 1996; Ramos et al. 1996). This suggests that SpAN and Smp may dismpt gastmlation movements by degrading this important component of the ECM. To test this embryos were injected at the 1-cell stage with 1.6 ng of either SpAN mRNA or smp mRNA and collected as early gastmlae (stage 10.5), late gastmlae (stage 12) and neumlae (stage 15). Embryos were homogenized and proteins analyzed by Western blot with an anti-Fibronectin monoclonal antibody, 4H2 (Ramos and DeSimone, 1996; Ramos et al. 1996), that recognizes an epitope near the centre of the protein, upstream of the RGD cell-binding site.

Uninjected, control embryos contain Fibronectin at all stages tested, with levels increasing as development proceeds (figure 4.1 A; see also Lee et al., 1984). In contrast, embryos injected with 1.6 ng of SpAN or smp mRNA contain no detectable Fibronectin at the start of gastmlation, although Fibronectin levels start to recover in late gastmlae (figure 4.1 A). This decrease in Fibronectin levels is not a result of there being less total protein on the SpAN and Smp blots, since each sample contains a similar amount of total protein, as detected by Coomassie Blue staining of samples mn on a polyacrylamide gel (data not shown, but see example in Figure 4. ID). Since SpAN may activate endogenous BMPs, Fibronectin levels were also examined in embryos injected with 1.6 ng of xbmp4 mRNA. The results show that xbmp4 mRNA-injected embryos contain similar levels of Fibronectin as uninjected controls, indicating the lack of Fibronectin in SpAN and smp mRNA-injected embryos is not an indirect result of activating endogenous BMP4 in the Xenopus embryo.

To investigate whether degradation of Fibronectin in injected embryos correlates with the severity of morphogenetic defects, embryos were injected with low concentrations of SpA N or smp mRNA and analyzed as described above. Embryos injected with 13 pg SpAN mRNA typically have a DAI score of 3.55 (see table 3.1), while embryos injected with 13 pg of smp mRNA have no obvious anterior defects (see figure 3.4). However, Fibronectin levels in both SpA N and smp mRNA-injected embryos are greatly reduced at the start of gastmlation, and are not significantly different from embryos injected with 1.6 ng of these mRNAs, although levels of Fibronectin recover earlier in embryos injected with the lower dose (figure 4.IB). These results demonstrate that there is no correlation between the severity of the phenotype and the extent of Fibronectin degradation. Furthermore, they indicate that little, if any, Fibronectin is required for normal gastmlation movements in the Xenopus embryo.

Ill 4.2.2 XBMPl decreases Fibronectin levels slightly, whilst Xolloid has no effect

To test whether XBMPl and Xld also degrade Fibronectin in early Xenopus embryos, 2 ng xld or xbm pl mRNA was injected into each blastomere at the two cell stage. In previous experiments this concentration of xbm pl and xld mRNA gave a mean DAI score of 3.8 and 3.7 respectively (Goodman et al., 1998), similar to the DAI score seen following injection of 13 pg SpA N mRNA (DAI 3.55). Injected embryos were collected at gastrulae and neurulae stages, and analyzed for the presence of Fibronectin by Western blot. In contrast to SpAN and smp mRNA-injected embryos, no reduction in the levels of Fibronectin were detected in xld mRNA injected embryos at all stages tested, or in xb m p l mRNA-injected embryos at late gastrulae or neurulae stages (figure 4.1C). There is, however, a slight reduction in Fibronectin levels in xbmpl mRNA-injected early gastmlae, although the level of Fibronectin is greater at stage 12, when compared to controls. This is not due to differences in total protein in this sample compared to others, as indicated by Coomassie Blue staining of the total protein in each sample mn on a polyacrylamide gel (figure 4. ID), although may be a consequence of uneven protein transfer during blotting.

The results confirm the absence of any correlation between the extent of anterior defects and the levels of Fibronectin in Xenopus gastmlae. Of course these results do not exclude the possibility that a correlation exists for other ECM proteins that were not tested.

4.2.3 SpAN does not affect calcium dependent cell-cell adhesion

The possibility that SpAN may inhibit gastmlation by dismpting cell-cell contacts was investigated by testing the calcium-dependent reaggregation of dissociated cells, although SpAN mRNA-injected embryos have no obvious sign of adhesion problems and explants do not fall apart upon isolation.

Marginal zone cells from uninjected embryos, or embryos injected with 1.6 ng of S p A N mRNA were isolated from early gastmlae (stage 10.5) and dissociated in calcium-, magnesium-free medium. The dissociated cells were then transferred to medium containing calcium and magnesium, and the time course of re-aggregation was observed. Cells from both SpA N mRNA-injected and uninjected embryos had loosely re-aggregated after 30 mins - 1 hour, and by 2 hours had formed compact clumps of cells (figure 4.2). This demonstrates that SpAN has no discernible effect on cell-cell adhesion and aggregation in vitro, and that it is unlikely to have a significant effect on calcium-dependent adhesion in the embryo.

112 4.3 Discussion

The results presented in this chapter show that Fibronectin levels are greatly reduced when either SpAN or smp mRNAs are injected into Xenopus embryos. The results also indicate that SpAN and smp are stably expressed in the embryo until at least the end of gastmlation since normal Fibronectin levels do not begin to recover until late gastmlation/neumlation. In contrast, overexpression of xld mRNA has no effect on Fibronectin levels in X enopus embryos, although overexpression of xb m p l mRNA may cause a slight decrease in Fibronectin levels at early gastmla stages. In addition, the results demonstrate that SpAN has no effect on calcium-dependent cell adhesion.

There are two explanations for the effect SpAN and Smp have on Fibronectin levels. They may either indirectly cause Fibronectin levels to fall, for instance by activating growth factors which in turn modulate ECM deposition rates, or they may directly degrade Fibronectin in the extracellular space.

4.3.1 Regulation of the ECM composition by growth factors

There is a good deal of evidence that growth factors can modulate the deposition of the ECM. For instance, TGFp up-regulates expression of ECM components such as Fibronectin, and down-regulates the transcription of ECM degrading enzymes (matrix metalloproteases; MMPs) such as stromelysins (reviewed by Massagué, 1990); other growth factors such as EDGE and FGF are able to induce expression of MMPs (reviewed by Birkedal-Hansen, 1995) and hence cause degradation of certain ECM components. It is possible then, that SpAN and Smp are able to release inactive growth factors from the ECM, through proteolytic cleavage. In turn these factors may then down-regulate transcription of Fibronectin, or up-regulate Fibronectin degrading enzymes, such as gelatinases or stromelysins. However, several pieces of evidence argue against this. Firstly, overexpression of the growth factor, BMP4, has no effect on the levels of Fibronectin in the Xenopus embryo, suggesting that SpAN and Smp are not acting through BMPs to alter Fibronectin levels, although of course they may effect other growth factors. Secondly, most of the Fibronectin in the early embryo is transcribed from maternal stores (Lee et al., 1984); thus SpAN and Smp cannot have an indirect effect, via growth factor signalling, on transcriptional control of Fibronectin in the early embryo, although they may in some way affect post-translational control of Fibronectin secretion. Finally, overexpression of MMPs such as MMPl or XMMP (Yang et al., 1997) or their inhibitors (TIMPs), by mRNA injection into the early Xenopus embryo, has no effect on development (Goodman et al., 1998; M. Kurkinen pers. comm.; my own observations).

4.3.2 Regulation of ECM composition by proteases

It is more likely that SpAN and Smp act directly in the extracellular space to degrade Fibronectin, and maybe other ECM proteins that were not analyzed in this study. There are 113 many examples of zinc metalloproteases acting on ECM components. MMPs, of which there are three main groups - stromelysins, gelatinases and collagenases - are able to degrade many components of the ECM (reviewed by Overall, 1991; Docherty et al., 1992; Birkedal-Hansen, 1995); in particular, stromelysins and gelatinases degrade Fibronectin. The metalloprotease domain of SpAN is related to the crayfish zinc metalloprotease, astacin, which completely degrades the native triple helix of Collagen type I (Stocker and Zwilling, 1995). In the Xenopus embryo, fibrillar Collagens (types I, II and III) are not expressed in the embryo until late gastmlation/neumlation (Bieker and Yazdani-Buicky, 1992; Su et al., 1991), however Collagen type VI has been detected in the early embryo and anti-Collagen VI antibodies block gastmlation movements (Otte et al., 1990). Astacin contains only a metalloprotease domain with no CUB or EGF-like repeats, similar to the tmncated version of SpAN, Smp. Thus, Smp may also be able to degrade Collagen, and consequently the activity of SpAN may be due to a non-specific activity that does not depend on its putative protein interaction domains in the C-terminal. In addition, BMPl is known to cleave the C-terminal peptide from pro-Collagens I, II and III (Kessler et al., 1996; Li et al., 1996) and process pro-Laminin y2 (a component of Laminin 5) into its tissue form (Amano et al., 1997). Although the only Laminin (Laminin (3) to described in Xenopus as yet is not expressed until mid-gastmla stages, around the somites and notochord (Fey and Hausen, 1990), laminin expression is detected in the oocytes and early embryos of the newt, Pluerodeles (Riou et al. 1987), indicting that other Laminins may be present in the early Xenopus embryo and may be important in gastmlation.

4.3.3 Is Fibronectin required for gastrulation movements?

Embryos injected with low concentrations (13 pg) of S p A N or smp mRNA both contain very low levels of Fibronectin during gastmlation. However, S p A N mRNA-injected embryos exhibit anterior truncations whereas smp mRNA-injected embryos are almost normal and do not exhibit anterior defects, despite containing similar amounts of Fibronectin. This implies that little, if any, of the Fibronectin normally present in the embryo is needed for development of anterior stmctures. In addition embryos injected with higher doses of SpA N or Smp mRNA, which also contain residual amounts of Fibronectin, exhibit delayed blastopore closure during gastmlation, implying that Fibronectin may not be required for normal gastmlation movements. This contrasts with experiments that use specific antibodies to block Fibronectin function, which show Fibronectin is required for normal gastmlation movements in the Xenopus embryo (Howard et al., 1992; Ramos and DeSimone, 1996; Ramos et al. 1996). This discrepancy may be explained if the residual amounts of Fibronectin detected in SpAN and smp mRNA- injected embryos are sufficient for normal gastmlation. Alternatively, SpAN/Smp may cleave Fibronectin in such a way that it is not detected with the 4H2 antibody, but leaves intact the RGD and ‘synergy’ site, which are required for integrin binding to Fibronectin and normal gastmlation movements (Ramos et al., 1996). Fibronectin does not, however, 114 seem to be necessary for normal gastmlation in mice embryos, since cells migrate as normal through the primitive streak of Fibronectin-deficient mice (George et al., 1993), although subsequently mesodermal stmctures, somites and notochord, do not differentiate. 4.3.4 Cell adhesion is not affected by SpAN

The results presented in this chapter show that SpAN has no detectable effect on cell adhesion, since cells dissociated from early gastmlae injected with S p A N mRNA, reaggregate in a calcium-dependent manner. This observation indicates that SpAN is unlikely to be degrading cadherins, which are essential for adhesion during early development and are dependent on calcium for their function. There may be other classes of adhesion proteins (such as integrins) or surface glycoproteins, which appear during blastula stages and are involved in adhesion. It is possible then, that SpAN degrades some classes of adhesion molecules, but that other classes are left intact so that cells are able to reaggregate.

4.3.5 Does SpAN regulate ECM composition in sea urchins?

In Xenopus, injection of both SpAN and smp mRNAs into the embryo results in a decrease in Fibronectin levels. Although it is not clear whether Fibronectin is present in sea urchin embryos, injection of anti-Fibronectin antibodies dismpts some aspects of gastmlation (Katow, 1990). The sea urchin has a complicated ECM network in the blastocoel (Cherr et al., 1992) that cells migrate along during gastmlation, and most classes of ECM and cell adhesion molecules have been described in the sea urchin embryo (e.g. Ghersi et al., 1993; Tomita et al., 1994; Gambino et al., 1996; Marsden and Burke, 1997). There is no reason why SpAN may not be responsible for correct processing and deposition of ECM components (e.g. Collagen) in a similar way to BMPl, or for degradation/remodelling of ECM during gastmlation, although as yet there is no evidence to support this. In fact, sea urchin BMPl (Hwang et al., 1994) is a better candidate for a metalloprotease affecting ECM deposition or degradation, since it is maximally expressed in the mesenchyme blastula just before PMC migration, whereas SpAN is maximally expressed during very early blastula stages when there is no cell movement.

115 Figure 4.1 SpAN, Smp and XBMPl cause a decrease in Fibronectin levels

Embryos were injected at the 1-cell stage with the indicated amount of SpA N , smp or xbm p4 mRNA, or at the 2-cell stage with a total of 4 ng of xb m p l or xld mRNA. Uninjected and injected embryos were collected at early gastrula (stage 10.5), late gastrula (stage 12) and neurula (stage 15) stages. Proteins from the equivalent of one embryo were run on a 5% SDS-acrylamide gel, and total protein was stained with Coomassie Blue (D) or Fibronectin was detected by Western blot with 4H2 anti-Fibronectin antibody (A-C).

(A) Fibronectin (-200 KDa) is present in uninjected embryos at the start of gastmlation and levels increase as development proceeds. Embryos injected with 1.6 ng SpA N or smp mRNA contain no detectable Fibronectin at the start of gastmlation, but levels begin to recover by late-gastmlae (SpAN) or mid-neumlae (Smp). xbmp4 mRNA injection has no effect on Fibronectin levels. (B) At low concentration (13 pg) of SpA N and smp mRNA, Fibronectin levels are greatly reduced during gastmlation when compared to control, uninjected embryos, although residual amounts of Fibronectin are present. (C) xb m p l mRNA injection causes a slight decrease in Fibronectin levels at the start of gastmlation, although levels have completely recovered by late gastmla stages. Injection of xld mRNA has no effect on Fibronectin levels in Xenopus embryos when compared to controls. (D) Coomassie Blue staining of total protein from uninjected embryos, or embryos injected with xld or xbm pl mRNA; only the major bands are shown. The gel shows that observed changes in Fibronectin levels are not due to a lower amount of total protein being loaded for each sample. Similar staining of SpAN, Smp and BMP4 samples also showed no detectable difference in the total amount of protein in each sample (not shown).

Figure 4.2 SpAN does not affect calcium-dependent cell adhesion.

Marginal zones were isolated from uninjected or SpA N mRNA-injected (1.6 ng/embryo) early gastmlae (stage 10.5), and dissociated in calcium-, magnesium-free medium. Dissociated cells were placed back into medium containing calcium and magnesium and allowed to reaggregate. After 2 hours cells from both control and SpA N mRNA-injected embryos had tightly reaggregated to the same degree.

116 Figure 4.1

B Stage 10.5 12 15 stage 10.5 12 15 -205 uninjected uninjected % % 205 » SpAN SpAN 1.6ng I3pg

SMP SMP 1.6ng 13pg m # BMP4 _ 1.6ng

Stage 10.5 12 15 stage 10.5 12 15 205 uninjected ^ ^ uninjected

Xolloid Xolloid

XBMPl XBMPl

Figure 4.2

control SpAN

10 mins

1 hour %

2 hours

117 Chapter 5

DO BMPl-RELATED METALLOPROTEASES ACT IN A BMP SIGNALLING PATHWAY?

5.1 Introduction

The results presented in Chapter 3 and recent work from this lab (Goodman et al., 1998; Appendix 1), show that BMPl-related metalloproteases are able to ventralize dorsal mesoderm in X enopus embryos or Activin treated animal caps. In many ways this ventralization resembles that seen when BMP4 is overexpressed in Xenopus em bryos (Dale et al., 1992; Jones et al., 1992a; 1996b). Thus an explanation for the ventralizing activity of SpAN, XBMPl and Xolloid may be that they enhance endogenous BMP ventralizing activity in the Xenopus embryo, similar to the way Tolloid enhances Dpp activity in the Drosophila embryo (Ferguson and Anderson, 1992b).

5.1.1 Regulation of ventral signalling during Xenopus development

In the Xenopus embryo BMP4 is a potent ventralizing molecule that is able to overcome the dorsalizing effect of the organizer (Dale et al., 1992; Jones et al., 1992a; Jones et al., 1996b). At the onset of gastmlation, bmp4 transcripts are expressed uniformly in the animal cap and ventral-lateral marginal zone, but are excluded from the organizer (Fainsod et al., 1994; Hemmati-Brivanlou and Thomsen, 1995). When BMP4 signalling is blocked by a dominant-negative BMP receptor, ventral mesoderm is converted to dorsal mesoderm, indicating that the ventral state must be actively maintained by BMPs (Graff et al., 1994; Suzuki et al., 1994). Similarly, inhibition of BMP signalling by dominant-negative ligands or antisense bmp4 RNA results in dorsalization of ventral mesoderm (Hawley et al., 1995; Steinbeisser et al., 1995). More recently three proteins secreted by the Xenopus organizer, Chordin, Noggin and Follistatin, have been shown to bind BMP4 and prevent interaction with its receptor (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997), and as a consequence ventral mesoderm is dorsalized when these proteins are overexpressed in the Xenopus embryo.

Other TGFp-related BMPs expressed in the Xenopus gastrulae may also play a role in ventral mesoderm formation. For example, xbm pl, which is expressed throughout the embryo, will ventralize dorsal mesoderm when overexpressed in X enopus em bryos (Suzuki et al., 1997). XBMP7 is able to form heterodimers with XBMP4 when expressed in insect cells (Hazama et al., 1995), and this heterodimer is a more potent ventralizing signal than homodimers for either XBMP4 or XBMP7 (Suzuki et al., 1997). In addition, BMP4/7 heterodimers are bound and inactivated by Xenopus Chordin (Piccolo et al., 1997). XBMP2 is also present in Xenopus gastrulae, being expressed throughout the 118 animal cap and marginal zone, and like BMP4 and BMP7, will promote the development of ventral mesoderm when overexpressed in Xenopus embryos or isolated animal caps (Hemmati-Brivanlou and Thomsen, 1995; Clement et al., 1995). Another related factor, anti-dorsalizing morphogenetic protein (ADMP) is expressed in the organizer but excluded from the rest of the marginal zone, despite this injection of admp mRNA ventralizes Xenopus embryos (Moos et al., 1995).

BMPs 2, 4 and 7 are bound, and inactivated, to differing degrees by the secreted proteins Chordin, Noggin and Follistatin (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). There is now clear evidence that BMPl-related metalloproteases are able to inactivate Chordin and its Drosophila homologue, Sog, by cleaving these proteins (Piccolo et al., 1997; Marqués et al., 1997). For instance, incubation of the X enopus metalloprotease, Xolloid, with BMP4/7 heterodimers bound to Chordin, led to Chordin being cleaved at two locations and active BMP4/7 heterodimers released (Piccolo et al., 1997). Xolloid will cleave Chordin in the presence or absence of BMPs, however its Drosophila homologue, Tolloid, will only cleave Sog in the presence of Dpp, or related BMPs such as BMP4 and BMP2 (Marqués et al., 1997). Other BMPl-related metalloproteases, such as XBMPl and XTolloid (Goodman et al., 1998; Lin et al., 1997), are present in the Xenopus embryo at the time Chordin is active, thus it is possible that these metalloproteases also cleave Chordin, or other inhibitory proteins such as Noggin and Follistatin.

This chapter investigates whether SpAN, XBMPl and Xolloid act in a BMP signalling pathway to cause ventralization, and if so at what point in the pathway they act. To test this, BMP signalling was blocked at two different points in the pathway. First, these metalloproteases were coexpressed with a dominant-negative receptor for BMP2 and BMP4 (tBR; Suzuki et al., 1994). This receptor lacks the intracellular serine/threonine kinase domain, and blocks propagation of a BMP signal. Second, these metalloproteases were coexpressed with Chordin and Noggin (Sasai et al., 1994; Smith and Harland, 1992), which block signalling by preventing BMP4 and BMP2 from interacting with their receptors (Piccolo et al., 1996; Zimmerman et al., 1996).

5.2 Results

5.2.1 SpAN requires a functional BMP signalling pathway to ventralize Xenopus embryos

To test whether the ventralizing activity of SpAN requires a functional BMP signalling pathway, embryos were injected at the 2-cell stage with a total of 1 ng tBr mRNA, or 150 pg Sp A N mRNA, or both mRNAs together. Ventral marginal zones (VMZs) were explanted from early gastrulae and cultured until sibling controls had reached stage 26-

119 27, then analyzed by RPA for the expression of muscle-specific a-actin and the blood- specific marker, aT4-globin.

Whereas control VMZs express significant amounts of aT4-globin and no a-actin (figure 5.1 A, lanes 1 and 2), rRr mRNA-injected VMZs express significant quantities of a-actin, but very little aT4-globin (figure 5.1 A, lane 7), consistent with the dorsalizing activity of this dominant negative receptor. Coinjection of 150 pg SpA N mRNA, an amount sufficient to markedly ventralize embryos (DAI score < 2.6; see table 3.1), has no effect on levels of a-actin in VMZs injected with 1 ng tBr mRNA (figure 5.1 A, lane 8), indicating that a functional BMP2/4 signalling pathway is required for the ventralizing activity of SpAN.

5.2.2 SpAN suppresses dorsalization by Chordin and Noggin

These results suggest that SpAN ventralizes the early Xenopus embryo by causing the BMP2/4 receptor to be activated. One possible mechanism is that SpAN functions, directly or indirectly, to release BMP2 or BMP4 from inactive complexes with inhibitors such as Chordin or Noggin. To test this possibility embryos were coinjected with a total of 150 pg of SpAN mRNA and either (i) 1 ng of chordin or (ii) 50 pg of noggin mRNA, and VMZs explanted and analyzed as described above. These amounts of chordin and noggin mRNA were chosen because they induce similar levels of a-actin in VMZs as 1 ng of tBr mRNA, with Noggin being a more potent inducer of dorsal mesoderm than Chordin or tBr (figure 5.1 A, compare lanes 3, 5 and 7).

Whereas VMZs from embryos injected with noggin or chordin mRNA strongly express a-actin (figure 5.1 A, lanes 3 and 5), those coinjected with SpAN mRNA express greatly reduced amounts of a-actin {chordin", figure 5.1 A, lane 4), or none at all {noggin", figure 5.1 A, lane 6). In addition, aT4-globin expression in coinjected VMZs increases (figure 5.1 A, lanes 4 and 6) compared to VMZs injected with chordin or noggin mRNA alone (figure 5.1 A, lanes 3 and 5), indicating that the mesoderm in these explants is ventralized. These results suggest that SpAN acts to release the inhibitory effects of Chordin and Noggin on BMP2 and/or BMP4.

5.2.3 XBMPl and Xolloid require a functional BMP signalling pathway to ventralize Xenopus embryos

Previously, Goodman et al. (1998; Appendix 1) have shown that overexpression of XBMPl and Xolloid causes Xenopus embryos to become weakly ventralized. In addition, XBMPl and Xolloid are able to ventralize the dorsal mesoderm induced by Activin in animal caps. To investigate whether this ventralizing activity requires a functional BMP signalling pathway, XBMPl and Xolloid were coexpressed with tBr either in Activin- induced animal caps, or in VMZs.

120 Embryos were injected at the 2-celI stage with 1.5 ng of xbm pl or xolloid mRNA, with or without 1 ng of tBr mRNA. Animal caps were then isolated from mid-blastulae, treated with human Activin A conditioned medium and cultured until sibling controls had reached stage 26-27 (tailbud), when they were analyzed by RPA for the expression of a- actin. Activin-treated animal caps injected with xbmpl or xolloid mRNA alone express very low levels of a-actin (figure 5.IB lanes 2 and 4; see also figure 8A and B in Goodman et al., 1998, Appendix 1). In contrast, those Activin-treated animal caps coinjected with tBr mRNA and xolloid or xbmpl mRNA, express a-actin at similar levels to controls, when levels of the loading control, ODC, are taken into account (figure 5.1C, compare lane 1 with lanes 3 and 4).

In addition, embryos were injected at the 2-cell stage with a total of 1 ng tBr mRNA with either 1 ng xb m p l or 1 ng xld mRNA. VMZs were explanted from early gastrulae and cultured until sibling controls had reached stage 26-27, then analyzed for the expression of a-actin. The results show that tBr mRNA-injected VMZs express significant quantities of a-actin (figure 5.1C, lane 6), and that coinjection of xbm pl or xolloid mRNA has no effect on these levels of a-actin (figure 5.1C, lanes 9 and 12).

Taken together these results demonstrate that the ventralizing activity of XBMPl and Xolloid require a functional BMP signalling pathway.

5.2.4 Xolloid and XBMPl suppress dorsalization by Chordin but not Noggin

To investigate whether XBMPl and Xolloid are able to overcome the dorsalizing affects of Chordin and Noggin, embryos were injected with a total of 1 ng of xolloid or xbm pl mRNA together with 1 ng of chordin or 50 pg of noggin mRNA at the 2-cell stage. VMZs were explanted from early gastrulae, cultured until sibling control embryos reached stage 26-27 and then analyzed for expression of a-actin. Both Xolloid and BMPl are able to overcome the effects of Chordin, as shown by a decrease in the levels of a- actin in VMZs coinjected with chordin mRNA (figure 5.1C, lanes 7 and 10), but can not overcome the dorsalizing effects of Noggin (figure 5. IB, lanes 8 and 11). As the levels of aT4-globin expression were not analyzed in this experiment, it is not clear whether these VMZs are ventralized, although this seems likely since both Xolloid and XBMPl are able to ventralize dorsal mesoderm in other situations (Goodman et al., 1998; Piccolo et al., 1997). These results corroborate and extend the result obtained by Piccolo et al. (1997) showing that Xolloid is able to overcome the dorsalizing effect of Chordin, but not Noggin, in the whole embryo, and cleaves Chordin but not Noggin in vitro.

121 5.3 Discussion

5.3.1 Requirement for a functional BMP signalling pathway

The results presented in this chapter show that SpAN, XBMPl and Xolloid do not ventralize dorsal mesoderm in VMZs coinjected with a dominant-negative version of a type I BMP2/4 receptor, and that this dominant-negative receptor (tBr; Suzuki et al., 1994) blocks the ventralization of dorsal mesoderm by XBMPl and Xolloid in Activin- treated animal caps. tBr lacks the cytoplasmic serine/threonine kinase domain essential for intracellular signalling, and blocks mesoderm induction in animal caps by BMP2 and BMP4, but not Activin (Suzuki et al., 1994). This receptor therefore exhibits a degree of specificity towards BMPs not observed with the related type IIB receptor for Activin, which blocks mesoderm induction by Activin, Vgl and BMP4 (Hemmati-Brivanlou and Melton, 1992; Schulte-Merker et al., 1994; Kessler and Melton, 1995). Because this receptor appears to specifically block BMP signalling, the results presented here demonstrate that SpAN, XBMPl and Xolloid act upstream of the BMP receptor and that a functional BMP signalling pathway is required for the ventralizing activity of these metalloproteases.

5.3.2 A different specificity of SpAN and XBMPl/Xolloid for Chordin and Noggin

Recent studies have shown that Chordin, Noggin and Follistatin, bind BMPs 2, 4 and 7 to varying degrees, preventing them from interacting with their receptors (Zimmerman et al., 1996; Piccolo et al., 1996; Fainsod et al., 1997). As a consequence all three proteins dorsalize ventral mesoderm when expressed in Xenopus embryos (Smith and Harland, 1992; Sasai et al., 1994; Sasai et al., 1995). However, when coexpressed with either SpAN, XBMPl or Xolloid the dorsalizing activity of Chordin is blocked, while only SpAN blocks the dorsalizing activity of Noggin. These results are consistent with a role for SpAN, XBMPl and Xolloid in regulating interactions between BMPl-like growth factors and their inhibitory binding proteins, thereby controlling the availability of free ligand in the extracellular space. Since SpAN, XBMPl and Xolloid act upstream of the BMP receptor and are secreted (see Chapter 6), it is likely that they act directly on Chordin, and Noggin in the case of SpAN, in the extracellular space to inhibit their activity. Although the results presented here do not provide evidence as to the nature of the interaction between these metalloproteases and Chordin or Noggin, recent in vitro studies have shown that Xolloid is able to cleave Chordin at two sites. Cleaved Chordin is unable to bind BMPs and prebound BMP4/7 heterodimers are released by this cleavage (Piccolo et al., 1997).

SpAN has a stronger ventralizing activity than XBMPl and Xolloid (see Chapter 3 and Goodman et al., 1998; Appendix 1). This may be explained by the fact that SpAN inhibits

122 the dorsalizing effects of both Chordin and Noggin, whereas XBMPl and Xolloid can only inhibit the effects of Chordin. In the Xenopus Chordin, Noggin, and Follistatin, bind BMPs and prevent them ventralizing dorsal mesoderm in the organizer. If SpAN is able to take away the inhibition of two of these binding proteins, whereas XBMPl and Xolloid can only remove one, then SpAN will have a stronger effect. Follistatin was not tested in this study, but Piccolo et al. (1997) have shown that Xolloid will not inhibit Follistatin in the whole embryo. It would be interesting to see if this is also the case with XBMPl and SpAN.

5.3.3 Noggin and Chordin are very different proteins

Although the actual 3-D structure of Noggin and Chordin is not known, their deduced amino acid sequences indicate that they are very different types of protein. This may explain why XBMPl and Xolloid do not recognize both Chordin and Noggin. SpAN, on the other hand, is less specific in its substrate recognition, being able to inhibit the activity of both Chordin and Noggin. This difference in the substrate recognition of SpAN and XBMPl/Xolloid may reside in the structure of their C-terminal domains; SpAN has a very different arrangement of EGF and CUB repeats to XBMPl and Xolloid, and in addition has an extra threonine-rich domain not present in XBMPl or Xolloid (Reynolds et al., 1992). However, XBMPl and Xolloid also have a different arrangement of CUB and EGF repeats (Goodman et al., 1998), despite having a similar substrate specificity when overexpressed in the Xenopus embryo. This may be because XBMPl and Xolloid are more related to each other in the C-terminal domain than to SpAN, which may allow them to recognize similar targets. Additionally, overexpression may relax the substrate specificity of XBMPl and Xolloid, masking more specific interactions that may exist at physiological concentrations.

Noggin is a small protein of 222 amino acids (Smith and Harland, 1992) which may be distantly related, in its C-terminal domain, to the Kunitz family of proteins (McDonald and Kwong, 1993), although it is not clear whether this putative Kunitz domain can adopt the 3D structure common to other Kunitz domains (Creighton and Kemmink, 1993). Kunitz-domain proteins are a large family of inhibitor proteins which fall into two main classes; the serine protease inhibitors and the toxin class which inhibit ion channel function. The Kunitz domain of both these types of inhibitor is implicated in protein- protein interactions, either with serine proteases or with ion channels (McDonald and Kwong, 1993). Another Kunitz-domain protein, Aprotinin, has been implicated in the control of TGFp activity (Wells and Strickland, 1994). Active TGFp prevents differentiation of myoblasts, but when Aprotinin is added to myoblasts in culture, they differentiate into myotubes (Wells and Strickland, 1994). Wells and Strickland explain their results in terms of Aprotinin inhibiting (serine) proteases which might activate latent TGFp, since addition of active TGFp with Aprotinin does not lead to differentiation of myoblasts, indicating that Aprotinin acts downstream of active TGFp. However, since 123 Noggin is now known to bind BMP4 and BMP2, it is also possible that Aprotinin is able to bind (latent) TGPp.

Chordin is a large protein of 941 amino acids, which contains four cysteine-rich repeats, one at its N-terminal and three at its C-terminal. BLAST searches show the cysteine-rich repeats, which have 10 cysteines with conserved spacing, share homology with repeats found in the N-terminal peptide of pro-Collagens a l (I, II, III) and a2(V), and near the N-terminus of Thromospondins 1 and 2 (see also Sasai et al., 1994; Francois et al, 1994). These repeats also show a more limited homology with cysteine-rich domains found in other extracellular proteins, von Willebrand factor and Laminin chains a5, pi, yl. These cysteine rich repeats may also be involved in protein-protein interactions since these extracellular proteins all form multimeric complexes (e.g. Shelton-Inloes et al, 1986; Bornstein, 1992; Maurer and Engel, 1996; Kadler et al, 1996), and in the case of Thromospondin 1 the cysteine-rich repeats may mediate interaction with Collagen (reviewed by Frazier, 1991).

If Xld and XBMPl are able to cleave Chordin but not Noggin, this raises the question as to what, if anything, regulates the activity of Noggin in the Xenopus embryo. SpAN is able to overcome Noggin’s dorsalizing activity when expressed in the Xenopus embryo, and so perhaps other BMP 1-related metalloproteases, homologous to SpAN, are expressed in the Xenopus gastrula which cleave Noggin. Alternatively, Noggin may be cleaved by another class of protease, such as a serine protease, or be regulated by some other mechanism.

124 Figure 5.1 SpAN, XBMPl and Xolloid act in the BMP signalling pathway to ventralize dorsal mesoderm

Embryos were injected with the indicated mRNAs (see text for amounts). Animal caps cut from mid-blastulae and treated with human Activin A conditioned media (B), or VMZs cut from early gastrulae (A andC ) were cultured until sibling controls had reached stage 26-27, then analyzed for expression of muscle-specific a-actin (m-actin; A, B and C) and blood-specific aT4-globin (A only). Levels of cytoskeletal actin (c-actin) were determined as a loading control for muscle actin, and ODC as a loading control for aT4- globin.

(A) Chordin (lane 3), Noggin (lane 5) and a dominant-negative BMP receptor (tBr; lane 7) dorsalizes ventral mesoderm as shown by the high levels of a-actin expression and lower levels of aT4-globin expression when compared to control VMZs (lane 1). SpAN was not able to ventralize dorsal mesoderm induced by tBr (lane 8) indicating SpAN acts upstream of the BMP receptor and requires a functional BMP signalling pathway to act. SpAN, however, is able to overcome the dorsalizing activity of both Chordin (lane 4) and Noggin (lane 6).

(B) XBMPl and Xolloid ventralize dorsal mesoderm induced by Activin in animal caps, as indicated by a decrease in a-actin expression (lanes 2 and 4). However, when coexpressed with tBr, which blocks BMP signalling, XBMPl and Xolloid are unable to ventralize the mesoderm, as indicated by the remaining high levels of a-actin expression (lanes 3 and 5) when compared to controls (lane 1).

(C) XBMPl and Xolloid are unable to ventralize dorsal mesoderm induced by tBr (compare lane 6 with lanes 9 and 12), as shown by the remaining high levels of a-actin expression, indicating XBMPl and Xolloid act upstream of the BMP receptor. XBMPl and Xolloid are able to overcome the dorsalizing activity of Chordin (compare lane 4 with lanes 7 and 10), but not the dorsalizing activity of Noggin (compare lane 5 with lanes 8 and 11).

125 0 1 o o 03 a T1 Q) Q) (O (Û O O O Q) Q) 11 Ô c 1.1 —T d ’ 3 3 s Î I I (D cn # None None None ro Xld ro Xld SpAN w XBMP1 me# tBr Chordin w + Xld Chordin t ## Chordin N3 0> XBMP1 + SpAN cn Noggin tBr cn Noggin cn + XBMP1 O) tBr Noggin O) Chordin + SpAN + Xld tBr 00 Noggin +Xld tBr CO + SpAN

THE ROLE OF THE C-TERMINAL DOMAIN OF BMPl-RELATED METALLOPROTEASES

6.1 Introduction

6.1.1 Domain structure of BMPl-related metalloproteases

The BMPl-related subgroup of astacin metalloproteases share a common domain structure; the N-terminus is composed of a signal sequence, which direct proteins into the secretory pathway, and a proregion that must be removed for activation (Bode et al., 1992; Stocker et al., 1993). Following this is the metalloprotease domain, containing the conserved zinc- binding motif, and the C-terminal domain which contains varying arrangements of EGF- like and CUB repeats which are implicated in protein-protein interaction (reviewed by Bond and Beynon, 1995).

BMPl, which has a CUB-CUB-EGF-CUB (C-C-E-C) arrangement in its C-terminal domain, is a short splice variant of a single gene found in humans, mice, Xenopus and sea urchins (Wozney et al., 1988; Maeno et al., 1993; Takahara et al., 1994; Hwang et al., 1994; Goodman et al., 1998). A longer splice variant has an additional EGF and two CUB repeats (C-C-E-C-E-C-C) at the C-terminal end (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997). This longer domain structure is also found in the products of the Drosophila genes, tolloid (Shimell et al., 1991) and tolloid-related (Nguyen et al., 1994; Finelli et al., 1995), the mouse gene mammalian tolloid-like (Takahara et al., 1996), the Xenopus gene xolloid (Goodman et al., 1998) and the Aplysia gene apTBL-1 (Lui et al., 1997). The sea urchin proteins SpAN and BPIO are more distantly related to BMPl in their C-terminal domain, which consists of an EGF-like repeat followed by two CUB repeats sandwiching a threonine-rich domain (Reynolds et al., 1992; LePage et al., 1992).

6.1.2 The role of EGF-like and CUB repeats in protein-protein interactions

Both EGF-like and CUB repeats have been implicated in protein-protein recognition and interaction. EGF-like repeats are found in many different proteins, all of which are involved in, or may be expected to be involved in, protein-protein interactions (reviewed by Appel la et al., 1988). For instance, EGF-like repeats are found in growth factors (e.g. TGF-a), receptors (e.g. Drosophila notch, C.elegans lin 12), extracellular matrix proteins (e.g. Laminins) and proteins of the blood coagulation pathway (e.g. coagulation factors, VII, IX, X and XII). Two CUB repeats and an EGF-like repeat, containing a high affinity site for calcium, and are present in the N-terminal interaction domain of C lr and Cls, two 127 serine proteases which together with Clq comprise the Cl complex of complement (reviewed by Arlaud and Thielens, 1993). Clr self-associates through its C-terminal catalytic domain to form a dimer, which then associates in the presence of calcium with two Cls subunits to form a tetramer. This Clr-Cls interaction is mediated by the N-terminal EGF/CUB repeats in C lr and Cls, since when this region is isolated by proteolysis, it will self-associate in the presence of calcium (Busby and Ingham, 1987). It is also likely that these EGF/CUB repeats mediate interaction of C lr 2-C ls 2 with Clq to form the mature Cl complex (Arlaud and Thielens, 1993).

Evidence that BMPl-like metalloproteases are able to recognize and interact with substrates has recently come from Piccolo et al. (1997), who showed that the Xenopus BMPl-related metalloprotease, Xolloid, is able to recognize and cleave Chordin. In addition. Marqués et al. (1997) have shown that Drosophila Tolloid recognizes and binds complexes of the Chordin homologue, Sog, when bound to Dpp or related BMPs, cleaving Sog in at least three different places. In Drosophila, loss of function alleles for dpp cause strongly ventralized embryos, whereas loss-of-function mutations of tolloid only weakly ventralize embryos. Genetic studies in Drosophila (Childs and O'Connor, 1994; Finelli et al., 1994) have identified antimorphic mutations of tolloid, which have a stronger phenotype than loss-of-function mutations of tolloid. These generally contain mutations in the protease domain which presumably renders it inactive. Antimorphic Tolloid is thus able to recognize a Dpp/Sog complex but is unable to proteolytically release Dpp, which has the effect of sequestering Dpp and results in a phenotype resembling dpp loss-of-function.

Piccolo et al. (1997) made a mutated form of Xolloid, based on an antimorphic allele of tld (6P41; Childs and O’Connor, 1994), where a missense mutation (Tyr^^^—>Asn) in the metalloprotease domain renders the protease inactive. When mRNA for this mutated form of xolloid is injected into Xenopus embryos they become weakly hyperdorsalized, with enlarged heads and cement glands, suggesting that this construct can interfere with endogenous Xolloid activity and block the ventralizing activity of BMPs. This chapter also aims to block endogenous metalloprotease function in Xenopus embryos, however, a different approach is taken. Instead of making a single point mutation in the metalloprotease domain, truncated versions of SpAN, XBMPl and Xolloid were made and injected into Xenopus embryos. These constructs lack the metalloprotease and proregion, but have an intact C-terminal domain, fused to a signal sequence to allow secretion.

128 6.2 Results

6.2.1 Secretion and activity of myc-tagged XBMPl and Xolloid

XBMPl and Xolloid, like all members of the astacin family of metalloproteases, contain an N-terminal signal sequence, and to date all members of the family that have been examined are secreted (Bond and Beynon, 1995). To test whether XBMPl and Xolloid are also secreted, PCR was used to attach 6 myc-tags to their C-termini (XBMPl.MT and Xld.MT). Mature Xenopus oocytes were injected with 30 ng x b m p l.M T or xld.M T mRNA and incubated in IX MBS for up to 48 hours, with a change of medium after 24 hours. Oocytes were collected after 24 or 48 hours and homogenized. Proteins from the oocyte homogenate and the corresponding culture medium were analyzed by Western blot probed with 9E10 anti-myc antibody and detected by ECL. Figure 6.1 A and B shows that both XBMPl.MT and Xld.MT are secreted from Xenopus oocytes with similar efficiency (lanes 4 and 6), although it is clear that a large proportion of the protein is retained in the oocyte and not secreted even after 48 hours incubation (lanes 3 and 5).

It is possible that the relatively poor secretion of these constructs is a result of the additional myc-tag sequences. If they were blocking secretion we would expect a lower activity for these myc-tagged constructs relative to the wild-type metalloproteases. To test this, Xenopus embryos were injected with 2 ng of xb m p l.M T or xld .M T mRNA into each dorsal blastomere at the 4-cell stage. Embryos injected with xbm pl.M T mRNA had a mean DAI score of 4.31, and those injected with xld.M T mRNA had a mean DAI score of 3.22 (table 6.1), which are similar to the DAI scores previously obtained with wild-type XBMPl and Xolloid (Goodman et al., 1998). Figure 6.2 C and D shows examples of these embryos exhibiting anterior truncations. A small number of extra tail-like structures and displaced patches of blood were also seen in xld.M T and x b m p l.M T mRNA-injected embryos (12-20%), as previously seen for XBMPl and Xolloid (Goodman et al., 1998). 56% of xb m p l.M T mRNA-injected embryos also had gastrulation defects, where the dorsal axis did not fuse in the abdomen; however, many of these embryos also had anterior truncations, and were included in the DAI scores for this experiment. These results demonstrate that the myc tags do not significantly change the activity of XBMPl or Xld when overexpressed in Xenopus embryos, suggesting that they do not affect the secretion of the protein.

6.2.2 Secretion of XBMPl and Xolloid C-terminal domains

In order to block BMPl-related metalloprotease activity, putative dominant-negative constructs of XBMPl, Xld and SpAN were made (XBMPlCub, XldCub, and SpANCub). These constructs have the proregion and metalloprotease domain removed, leaving the C-terminal domain intact. In each construct, the C-terminal domains are fused to the XBMPl signal sequence to allow secretion (see materials and methods).

129 The results presented above demonstrate that XBMPl and Xolloid are secreted, but to check that XBMPl Cub and XldCub are also secreted, additional constructs were made which have 6 myc epitopes attached at the C-terminal end of each protein (XBMPlCub.MT and XldCub.MT). Mature oocytes were injected with 30 ng xbmplcub.MT or xldcub.M T mRNA and incubated for up to 48 hours, with a change of medium after 24 hours. Proteins from oocytes and culture medium were analyzed by Western blot as described above. Although most of the protein is retained by the oocyte both XBMPlCub.MT and XldCub.MT are secreted into the culture medium (figure 6.2), with most XBMPl Cub.MT being secreted within the first 24 hours (compare figure 6.2 A and B). In addition, XBMPl Cub.MT is more efficiently secreted than XldCub.MT, which is clearly seen in figure 6.2C. The samples on this blot (6.2C) are from the same injection experiment as the other blots, but the equivalent of 2 oocytes homogenate and culture medium were run on this polyacrylamide gel in a separate blotting experiment. Translation of the injected message is also more efficient for XBMPl Cub.MT than XldCub.MT; whereas XBMPl Cub.MT protein is clearly seen on the blot after short exposure times (1 min; figure 6.2), longer exposures (up to 1 hour) are often needed to see significant amounts of XldCub.MT protein (figure 6.2 B and C). The blots also show a degradation product in the XldCub lane after long exposures, that is not seen with full-length Xolloid or XBMPl Cub. This must be a result of degradation from the N-terminus since the myc-tag used to detect the protein is at its C-terminus.

Comparison of blots 6.1 A and 6.2A, which are from the same blotting experiment, shows that XBMPl Cub.MT is more efficiently secreted than full-length XBMPl.MT during the first 24 hours. At low exposures (1 min; not shown) hardly any full-length XBMPl.MT protein is seen in the culture medium, whereas XBMPl Cub.MT protein is seen in the culture medium when the blot is exposed for only 1 min (figure 6.2A).

These results demonstrate that both XBMPl Cub and XldCub are secreted, demonstrating that neither the proregion nor the metalloprotease domain is essential for secretion.

6.2.3 BMPlCub, XolloidCub and SpANCub dorsalize ventral mesoderm

To test the effect of overexpressing the C-terminal constructs, X enopus embryos were injected with 1.5 ng xbmplcub, xldcub or SpANcub mRNA into each blastomere at the 2- cell stage. These embryos develop normally until late neurulae when a small number of embryos (5-10%) exhibit enlarged cement glands and hatching glands, indicating a degree of hyperdorsalization (figure 6.3A and B). However, the embryos seem to regulate and at later stages appear morphologically normal.

Since the C-terminal domain constructs have a weak effect on whole embryos, their effect on isolated ventral mesoderm was tested. Embryos were injected with 1.5 ng of

130 xbmplcub, xldcub or SpANcub mRNA into each ventral blastomere at the four-cell stage and VMZ expiants were explanted from early gastrulae. The VMZs were incubated until sibling controls had reached tailbud stages when they were collected for RPA analysis of a- actin and aT4-globin expression, or were incubated until stage 40 and then analyzed by histology. VMZs injected with xbmplcub, xldcub and SpANcub mRNA are elongated compared to control VMZs, consistent with dorsal mesoderm having formed. When SpAN cub and xbm plcub mRNA-injected VMZs are allowed to develop to tadpole stages (as judged by sibling control embryos) melanocytes are clearly visible, indicating dorsal neural tissue is present in these explants. Histological sections of these VMZs were scored for dorsal mesodermal and neural tissue content (table 6.2). All xbm plcub mRNA-injected VMZs (figure 6.3F) that were scored contained either dorsal mesoderm or neural tissue; 62% contained neural tissue, 54% contained notochord and 54% contained muscle. In contrast, only a small percentage of SpANcub mRNA-injected VMZs (figure 6.3E) contained neural tissue (10%) and notochord (5%), but all VMZs contained muscle, xldcub mRNA-injected VMZs (figure 6.3G) contained no notochord or neural tissue, and only 36% contained muscle, retaining large amounts of ventral-type mesoderm (mesenchyme and blood).

RPA analysis confirmed that xbmplcub, xldcub or SpANcub mRNA-injected VMZs are dorsalized, since they all express muscle-specific a-actin (figure 6.3H). When levels of the loading control, ODC, are taken into account, SpANcub mRNA-injected VMZs contain higher levels a-actin than xbmplcub or xldcub mRNA-injected VMZs, which is consistent with the observation that all SpANcub mRNA-injected VMZs contain muscle in histological sections. In addition, xbm plcub and SpANcub mRNA-injected VMZs show a concomitant decrease in aT4-globin expression (figure 6.3H, lanes 2 and 4), but xldcub mRNA- injected VMZs express similar amounts of aT4-globin as controls (figure 6.3H, compare lane 1 with lane 3), consistent with the weakly ventralized phenotype seen in histological sections.

Taken together these results show that the C-terminal domains of SpAN, XBMPl and Xolloid are able to dorsalize ventral mesoderm, with XBMPl Cub having the strongest dorsalizing activity, and XldCub only a weak activity.

6.2.4 XBMPlCub and SpANCub cause ectopic expression of dorsal genes

XBMPl Cub and SpANCub are able to down-regulate expression of the late ventral marker aT4-globin and all the C-terminal domain constructs induce expression of a-actin, a dorsal marker, in VMZs. To test whether these constructs are able to induce expression of dorsal markers or down-regulate ventral markers at earlier stages, embryos were injected with 1.5 ng xbmplcub, xldcub or SpANcub mRNA into a single ventral or dorsal blastomere at the 4-cell stage. At mid-gastrulation (stage 11.5-12) uninjected and injected embryos were

131 analyzed by WISH for the expression of the ventral marker, XwntS, or the dorsal markers, chordin (chd) and goosecoid {gsc\ figure 6.4).

Injection of SpANcub or xldcub mRNA into ventral blastomeres had no effect on X w n tS expression in mid-gastrulae, which is normally expressed in the ventral and lateral marginal zone (figure 6.4A). In contrast, injection of xbm plcub mRNA into ventral blastomeres caused an apparent drop in the levels of Xw ntS expressed throughout the ventral-lateral marginal zone (figure 6.4B). Injection of SpANcub, xbmplcub or xldcub mRNA into the dorsal side of the embryo did not cause an expansion in the normal expression domain of either gsc or chd on the dorsal side of the embryo. However xbm plcub mRNA injection into the ventral side of the embryo caused ectopic chd expression in 2 out of 14 cases (figure 6.4D). Also, in 1 case out of 11 SpANcub mRNA injection into the ventral side of the embryo caused ectopic gsc expression on the ventral side of the embryo (figure 6.4F).

These results demonstrate that SpANCub and XBMPl Cub are able to induce dorsal gene expression in the ventral marginal zone of gastrula stage embryos, albeit at a low frequency, and that XBMPl Cub can inhibit ventral gene expression. XldCub does not have any effect on early marker gene expression at gastrulation. These early phenotypes are much weaker than in isolated VMZs with later dorsal and ventral markers, although a lower concentration of mRNA (1.5 ng compared to 3 ng) was used in this experiment.

6.2.5 XBMPlCub induces expression of neural marker genes in isolated animal caps

Since inhibition of BMP signalling causes isolated animal caps to differentiate neural tissue (for review see Bier, 1997; Hemmati-Brivanlou and Melton, 1997), neuralization of animal caps by the C-terminal domain would strongly suggest that they act to block BMP signalling. To test this embryos were injected with 1.5 ng of xbmplcub, xldcub or SpANcub mRNA into each blastomere at the 2-cell stage. Animal caps were cut from mid- blastulae and cultured until sibling controls had reached tailbud stages, when they were analyzed for expression of both cement gland and neural-specific markers.

Animal caps were analyzed by WISH for expression of X A l (figure 6.5 A-D), an anterior marker normally expressed in the cement gland, hatching gland, and other non- ectoderm (Sive et al., 1989; Hemmati-Brivanlou et al., 1990). X A l is expressed in animal caps isolated from {\?>l\l), xldcub (4/24) and SpANcub (6/24) mRNA-injected animal caps, although expression in xldcub and SpANcub mRNA-injected caps is weak and patchy.

RPA analysis confirmed that XBMPl Cub neuralizes animal caps, since expression of both nrpl (Richter et al., 1988), a pan-neural marker, and XAG (Sive et al., 1989), an anterior marker expressed at high levels in the cement gland and lower levels in the hatching gland, is induced in xbm plcub mRNA-injected animal caps (figure 6.5E, lane 3). SpANcub and 132 xldcub mRNA injection does not induce expression of nrpl or XAG in animal caps (figure 6.5E, lanes 2 and 4), consistent with the observation that they are only very weak inducers o f X A l in animal caps.

These results demonstrate that the C-terminal domain of XBMPl induces neural tissue in isolated animal caps. The presence of dorsal mesoderm, which is able to induce neural tissue in ectoderm, was not tested for in this study, but recent results from this lab show that the dorsal mesodermal marker, a-actin, is not induced in these caps (L. Dale and J . Welch, pers. comm.). This indicates that the induction of neural tissue by XBMPl Cub in animal caps is not the result of dorsal mesoderm induction and subsequent neural induction, but of direct neural induction consistent with the idea that XBMPl Cub inhibits BMP activity.

6.2.6 XBMPlCub rescues anterior structures in x b m p l mRNA- injected embryos

If SpANCub, XBMPl Cub and XldCub are blocking endogenous metalloprotease activity by acting as dominant inhibitors, we would expect injection of these constructs to also block the activity of coinjected full-length metalloprotease. To test this, embryos were injected with 0.75 ng of xbm pl mRNA, with or without 0.75 ng xbm plcub mRNA and allowed to develop to tailbud stages.

Embryos injected with xbm pl mRNA have a range of anterior defects, and some gastrulation defects as previously described (see section 6.2.1; Goodman et al., 1998). Table 6.3 shows that 13 out of 18 xbm pl mRNA-injected embryos had reduced head structures (figure 6.6A), such as a small cement gland or reduced foreheads, whereas only 3 out of 51 embryos that were coinjected with xbm plcub mRNA had anterior truncations. When injected embryos were scored for cement glands, the most anterior structure, it was found that 16 out of 18 xbm pl mRNA-injected embryos exhibited cement glands, of which 8 were reduced compared to controls, and 3 were enlarged compared to controls. All 51 embryos coinjected xbmpl and xbmplcub mRNAs had cements glands, of which 39 were enlarged compared to controls (figure 6.6B), and none were reduced. However, despite rescuing anterior truncations induced by xbm pl mRNA injection, XBMPl Cub did not rescue the spina bifida-like phenotype often seen in xb m p l mRNA-injected embryos (figure 6.6 C and D).

The results suggest that XBMPl Cub blocks the ventralizing activity of coinjected XBMPl and that the phenotypes observed following injection of XBMPl Cub alone may result from a dominant-inhibitory effect on endogenous XBMPl.

133 6.3 Discussion

6.3.1 Full-length and C-terminal domain constructs of XBMPl and Xolloid are secreted

The results presented here show that both XBMPl and Xolloid are secreted from Xenopus oocytes. This supports the results of Piccolo et al. (1997) who used a human kidney cell line (293T cells), transfected with a Xolloid construct, to produce Xolloid conditioned media. Since both oocytes and mammalian cells secrete Xolloid, it is likely that cells of the Xenopus embryo also secrete Xolloid, and XBMPl, into the extracellular space. However, XBMPl and Xolloid are not efficiently secreted, which may explain why large amounts of mRNA must be injected into the embryo to produce a weak ventralizing phenotype (see Goodman et al., 1998; Appendix 1). The results also show that the C-terminal myc-tags have little, if any, effect on the activity of these metalloproteases, since embryos injected with myc-tagged constructs are weakly ventralized and have DAI scores similar to embryos injected with the untagged proteases. In addition, defects seen with the wild-type XBMPl and Xolloid constructs, such as misplaced blood islands and a spina bifida-like phenotype were also seen with the myc-tagged constructs. These results suggest that the wild-type proteins may also be secreted inefficiently.

The main aim of this chapter was to analyze the effects on embryonic development of XBMPl, Xolloid and SpAN constructs lacking the metalloprotease domain. This was achieved by using PCR to generate constructs that fused the C-terminal EGF and CUB repeats to the signal sequence of XBMPl (see materials and methods). Since the C-terminal domain is believed to be responsible for protein-protein interactions, it was hoped that these constructs would act as dominant-inhibitors of endogenous BMPl-like metalloproteases. To check that these C-terminal domains are secreted, constructs of XBMPl Cub and XldCub were also made in which 6 myc epitopes were added to the C-terminus. Injection of mRNA into oocytes confirmed that both XBMPl Cub.MT and XldCub.MT are secreted, although XBMPl Cub.MT appears to be more efficiently translated and secreted than XldCub.MT. In addition XBMPl Cub is more efficiently secreted than full-length XBMPl, suggesting that the proregion and metalloprotease domain may in fact hinder secretion of the full-length protein. SpAN contains a single C-terminal myc tag, but unfortunately this protein could not be detected by Western blotting after injection of the mRNA into either Xenopus oocytes or embryos.

6.3.2 The activity of the C-terminal domain constructs is consistent with BMP signalling being inhibited

Injection of xbmplcub, xldcub or SpANcub mRNAs into early Xenopus embryos usually has little effect on development, with most embryos forming morphologically normal tadpoles. However, in a few cases (5-10%) injected embryos have enlarged cement glands,

134 which is indicative of mild hyperdorsalization. Consistent with this, XBMPl Cub down- regulates expression of ventrally expressed and in 2 out of 14 cases induced ectopic expression of dorsally expressed chordin at gastrula stages. Ectopic expression of the dorsal marker, goosecoid, was also observed in 1 out of 11 SpANcub mRNA-injected gastrulae. In contrast, dorsalization of ventral mesoderm is consistently seen when VMZs are isolated from early gastmlae injected with xbmplcub, xldcub or SpANcub mRNAs. Whereas VMZs isolated from control embryos differentiate ventral mesoderm, i.e. mesenchyme and blood, VMZs from xbm plcub mRNA-injected embryos differentiate dorsal mesoderm, notochord and muscle, as well as neural tissue. VMZs from SpANcub mRNA-injected embryos typically differentiate muscle, although in a few cases (5-10%) notochord and neural tissue are also seen. Finally, VMZs from xldcub mRNA-injected embryos differentiate ventral mesoderm although a significant minority (36%) also differentiate muscle. This is consistent with the observation that XldCub does not cause a decrease in aT4-globin expression, despite up-regulating a-actin expression in isolated VMZs, whereas both XBMPl Cub and SpANCub cause aT4-globin levels to drop dramatically. Considering the extent of dorsalization of xbm plcub mRNA-injected VMZs, it is perhaps surprising that injection of this mRNA has such little effect on whole embryos. This may reflect the ability of the embryo to regulate back to normality, as seen in both xb m p l and xld mRNA-injected Xenopus embryos (Piccolo et al., 1997; Goodman et al., 1998) and in the chordino (chordin) mutant in zebrafish embryos (Schulte-Merker et al., 1997).

The dorsalization of ventral mesoderm by XBMPl Cub, XldCub and SpANCub is consistent with these constructs inhibiting endogenous BMP signalling in the embryo. Since previous work has shown that blocking BMP signalling causes direct neuralization of animal cap ectoderm (for reviews see Bier, 1997; Hemmati-Brivanlou and Melton, 1997), the effect of the C-terminal domain constructs on isolated animal caps was also tested. In the Xenopus embryo, signals from the organizer which dorsalize ventral mesoderm also reach the animal cap ectoderm and cause neural differentiation. Factors such as Noggin and Chordin, which bind BMPs and block ventralization of mesoderm (Piccolo et al., 1996; Zimmerman et al., 1996), are also able to directly neuralize animal cap ectoderm, without first inducing dorsal mesoderm (Lamb et al., 1993; Sasai et al., 1995). In addition, dissociating animal caps cells, which has the effect of diluting out secreted proteins including BMPs, causes neuralization (Godsave and Slack, 1989; Wilson and Hemmati- Brivanlou, 1995). A sequence of cell fates - epidermis, cement gland then neural tissue - can be produced by progressively inhibiting endogenous BMP signalling in animal caps, so that high doses of a BMP antagonist induce neural fate and lower doses induce cement gland (Wilson et al., 1997). Animal caps injected with xbmplcub, xldcub or SpANcub mRNA all express XA1, an anterior marker expressed in the cement gland and hatching gland, to varying degrees, with XBMPl Cub being the more potent inducer. RPA analysis demonstrates that XBMPl Cub induces neural tissue as well as cement gland in animal

135 caps, whereas XldCub and SpANCub do not. Since the induction of neural tissue requires stronger inhibition of BMP signalling than induction of cement gland markers, these results are consistent with the phenotype seen in VMZs, where XBMPl Cub has a stronger dorsalizing effect than SpANCub or XldCub on ventral mesoderm.

6.3.3 XBMPlCub has greater activity than XldCuh or SpANCuh

There are a number of possible reasons why XBMPl Cub has a greater potency than either XldCub or SpANCub. Firstly, these proteins have a different arrangement of EGF and CUB repeats, which may affect substrate specificity; the version of XBMPl used here has a short CUB-CUB-EGF-CUB (C-C-E-C) motif, Xolloid has a longer C-C-E-C-E-C-C motif, and SpAN has an E-C-T-C motif, in which a threonine rich domain (T) is sandwiched between two CUB repeats. In addition, the actual amino acid sequences of the EGF and CUB repeats in each protein differ outside certain conserved residues, and these differences may too influence the substrate specificity of XBMPl Cub, XldCub and SpAN Cub, as has been suggested for Tolloid and Tolloid-related l(Bond and Beynon, 1995). Tolloid (Tld; Shimell et al., 1991) and Tolloid-related 1 (Tlrl, also called tolkin; Nguyen et al., 1994; Finelli et al., 1995), which are both expressed in the early Drosophila embryo, have a C-C-E-C-E-C-C motif and an additional insert in the first CUB domain that is not found in other BMPl-related metalloproteases. However, the insert in Tlrl is much shorter than that in Tld. Loss-of-function mutations in tld disrupt patterning in the dorsal half of the blastoderm stage embryo, transforming dorsal tissue towards ventral fates (Shimell et al., 1991; Ferguson and Anderson, 1992b). In contrast, loss-of-function mutations in tlrl cause lethality during larval and pupal stages (Nguyen et al., 1994; Finelli et al., 1995). Expression of tlrl from the tld promoter fails to rescue tld mutations and extra copies of tld fail to rescue tlrl mutations (Nguyen et al., 1994), demonstrating that Tld and Tlrl are not interchangeable. In addition, whereas tld mutations can be partially rescued by increasing the copy number of dpp (Ferguson and Anderson, 1992b), tlrl mutations cannot (Nguyen et al., 1994). These results suggest then that Tlrl does not regulate the activity of Dpp, at least in the early embryo, in the way that Tld does, rather it must have different substrates, probably as a consequence of the differences in the C-terminal domain.

Secondly, the difference in potency of the C-terminal domain constructs may be due to differences in the amount of proteins translated and secreted in the embryo. Although we have no data on the secretion of SpAN, the results presented in this chapter show that XBMPl Cub is more efficiently secreted than XldCub from Xenopus oocytes, consistent with XBMPICub having a stronger activity than XldCub. Since the full-length proteins of XBMPl and Xolloid are secreted with similar efficiency, and have similar activities in the Xenopus embryo despite having different C-terminal domain motifs, differences in the efficiency of secretion of the C-terminal domains may be the most likely explanation. This could be tested by injecting decreasing concentrations of xbm plcub mRNA into Xenopus embryos and isolating VMZs. The prediction would be that lower doses of xbm plcub 136 mRNA would lead firstly to more muscle formation in VMZs at the expense of more dorsal structures, notochord and neural tissue, and eventually at very low doses to only small induction of muscle, with large amount blood and ventral mesoderm remaining.

If the C-terminal domain constructs were acting as dominant-inhibitors of metalloproteases, then injection of these constructs should block the activity of coinjected full-length metalloprotease. Initial experiments, where SpA N and SpANcub mRNAs were coinjected into Xenopus embryos at the one cell stage (data not shown), did not provide good evidence that these constructs act as dominant inhibitors. This may be because SpAN has too many ‘non-specific’ effects, independent of the C-terminal domain (see Chapter 3), making the experiment difficult to interpret and the results inconclusive. These experiments were repeated by coinjecting xbm pl mRNA with xbm plcub mRNA into the whole embryo. XBMPICub was able to rescue the mild anterior truncations induced by full- length XBMPl, although the other morphological defects, such as spina bifida, exhibited by xbm pl mRNA-injected embryos were not affected. This may be because XBMPl has some degree of non-specific protease activity, which is independent of the C-terminal domain, although injection experiments with an XBMPl metalloprotease domain construct indicate otherwise (Goodman et al., 1998).

While the results presented here suggest that the C-terminal domain constructs block endogenous metalloprotease activity, and consequently BMP activity, further work is clearly needed to demonstrate that these constructs act in a dominant-negative manner.

137 Table 6.1 Injection of myc-tagged xbmpl and xolloid mRNAs causes anterior truncations

DAI score

mRNA 0 1 2 3 4 5 mean n

None 0 0 0 0 0 30 5.00 30

xbmpl.MT 0 0 0 1 9 6 4.31 16

xld.MT 0 2 4 5 9 2 3.22 22

Embryos were injected with 2 ng of the indicated mRNA into each dorsal blastomere at the 4-cell stage and the resulting embryos scored at stage 30 using the dorso-anterior index (DAI) of Kao and Elinson (1988; see materials and methods). Numbers in table indicate the number of cases at each DAI score.

138 Table 6.2 SpANCub, XBMPICub and XldCub induce dorsal differentiation in ventral marginal zones

mRNA NT No Mu n

None 0 0 0 15

SpA NC UB 2 1 20 20

BMP1CUB 8 7 7 13

XldCUB 0 0 5 14

Embryos were injected with 1.5 ng of xbmplcub, xldcub or SpANcub mRNA into each ventral blastomere at the 4-cell stage and VMZs explanted from early gastrulae. These were incubated until stage 40 and then sectioned and scored for the presence of dorsal structures. Numbers in table indicate the number of sections containing neural tube (NT), notochord (No) or muscle (Mu). All SpANcub and xbm plcub mRNA-injected VMZs contained at least one type of dorsal structure, n = total number of VMZ sections scored.

139 Table 6.3 Injection of xbmplcub mRNA rescues anterior structures in xbmpl mRNA-injected embryos

RNA______AD______CG______reduced enlarged ______n_ xbm pl 13 16 8 3 18 xbmpl + xbmplcub 3 51 0 39 51

Embryos were injected with 0.75 ng of xbm pl mRNA or 0.75 ng of xbm pl plus 0.75 ng of xbm plcub mRNA into a single ventral blastomere at the 4-cell stage and scored for anterior defects (AD); these included reduced cement gland, small eyes and tmncated forehead. Additionally, cement glands (CG), were scored according to their size (reduced or enlarged) compared to normal control embryos. Numbers in table indicate the number of cases at each DAI score; n = total number of embryos scored from 1 experiment.

140 Figure 6.1 XBMPl and Xolloid are secreted proteins

(A) and (B) Western blet analysis of myc-tagged XBMPl (lanes 3 and 4) and Xolloid (lanes 5 and 6) in Xenopus oocytes. Oocytes were injected with 30 ng xb m p l.M T or xld.M T mRNA and incubated for up to 48 hours with a change of medium after 24 hours. ‘48 hour’ culture medium (B) is thus medium collected in the second 24 hours of incubation. The equivalent of half an oocyte of cell extract (o) and the culture medium (m) of 4 oocytes were Western blotted and probed with 9E10 anti-myc antibody. Proteins were detected by ECL; the exposure shown is for 1 hour. Both XBMPl and Xolloid are secreted into the culture medium (A and B, lanes 4 and 6), although most of the protein is retained by the oocyte (A and B, lanes 3 and 5), even after 48 hours. No signal could be detected in uninjected oocytes and culture medium (A and B, lanes 1 and 2).

(C) and (D) Embryos were injected with 2 ng mRNA for xb m p l.M T (C) and xld.M T (D) into each dorsal blastomere at the 4-cell stage. The embryos pictured exhibit anterior truncations, similar to those caused by injection of wild-type xbm pl or xld mRNA. Note the misplaced blood islands (arrow) in the top Xld.MT mRNA-injected embryo.

Figure 6.2 XBMPICub and XldCub are secreted

Western blot analysis of myc-tagged XBMPICub (tracks 1 and 2) and XLDCub (tracks 3 and 4) in Xenopus oocytes. Oocytes were injected with 30 ng xbmplcub.MT or xldcub.M T mRNA and incubated for up to 48 hours with a change of medium after 24 hours. For blots (A) and (B) the equivalent of half an oocyte of cell extract and the culture medium of 4 oocytes were Western blotted, for blot C the equivalent of 2 oocytes of cell extract and the culture medium from 2 oocytes were Western blotted. Proteins were detected by ECL; the exposures shown are for 1 min, except the 48 hour XldCub.MT blot (B) which is a 1 hour exposure and the second 24 hour XldCub.MT blot (C) which is a 45 min exposure. Both XBMPlCub.MT and XldCub.MT are secreted into the culture medium (tracks 2 and 4), although a large proportion of the protein is retained by the oocyte (tracks 1 and 3), with XBMPlCub.MT being more efficiently translated and secreted than XldCub.MT

141 Figure 6.1

24 hrs B 48 hrs control XBMP1 Xolloid control XBMP1 Xolloid omomom omomom

m "

123456 123456

-

XBMP1.MT Xolloid.MT

Figure 6.2

A 24 hrs ® 48 hrs XBMPICUB XldCUB XBMPICUB XldCUB omom omom

1 2 3 4 1 2 3 4

24 hrs XBMPICUB XldCUB O m O m

1 2 3 4

142 Figure 6.3 XBMPICub, XldCub and SpANCub ventralize dorsal mesoderm

(A)-(C) Embryos were injected with 1.5 ng of xbm plcub or xldcub mRNA into each blastomere at the 2-cell stage. Both xbmplcub (B) and xldcub (C) mRNA-injected embryos appear weakly hyperdorsalized at early tailbud stages, with enlarged cement glands (arrowheads) and hatching glands when compared to controls (A). SpANcub mRNA injection has also been observed to cause hyperdorsalization of Xenopus embryos (not shown).

(D)-(G) Embryos were injected with 1.5 ng of SpANcub, xbmplcub or xldcub mRNA into each ventral blastomere at the 4-cell stage and incubated until stage 10.5 when ventral marginal zones (VMZs) were isolated. VMZs were cultured until sibling controls had reached stage 26, and analyzed by RPA, or stage 40, and processed for histology. Control VMZs (D) contain mesoderm of a ventral type (mesenchyme and blood), whereas SpANcub mRNA-injected VMZs (E) contain muscle (mu), and occasionally neural tissue (nt) and/or notochord, xbm plcub (F) mRNA-injected VMZs frequently contain notochord (no) as well as muscle and neural tissue, xldcub (G) mRNA-injected VMZs contain muscle but never neural tissue or notochord.

RPA analysis (H) confirms that injected VMZs are dorsalized. Total RNA isolated from injected and uninjected embryos was hybridized with probes for muscle-specific a-actin (m-actin) and blood-specific aT4-globin mRNAs. Levels of ornithine decarboxylase {ODC) were determined as a loading control for aT4-globin and cytoskeletal actin (c-actin) as a loading control for muscle actin. xbmplcub, xldcub and SpANcub mRNA-injected VMZs all express a-actin. In addition, levels of aT4-globin are down-regulated in xbm plcub and SpANcub mRNA-injected VMZs (lanes 2 and 4) but not xldcub mRNA- injected VMZs (lane 3), confirming that XldCub is a less potent than either XBMPICub or SpANCub.

143 Figure 6.3

control XBMPICUB XldCUB

control SpANCUB

CD 3 ü Z nt mu < RNA (/)O. m-actin ## c-actin globin

000 f l XBMPICUB XldCUB 1 2 3

Figure 6.4

control injected A B u XwntS xbmplcub

chd ^ ► xbmplcub F

gsc

^A N cub

145 Figure 6.4 XBMPICub and SpANCub cause ectopic expression of dorsal genes

Embryos were injected with 1.5 ng xbm pl cub or SpANcub mRNA into a single ventral or dorsal blastomere at the 4-cell stage, and cultured until stage 11.5-12. Uninjected and injected embryos were analyzed by WISH for the expression of the dorsal markers chordin {chd) or goosecoid {gsc), or the ventral marker XwntS.

Both uninjected (A) and injected embryos (B) express XwntS in the ventral-lateral marginal zone, although levels of expression of Xw ntS appeared lower in embryos injected with xbm plcub mRNA on the ventral side (B). Both uninjected (C and B) and injected (D and F) embryos express chordin or goosecoid on the dorsal side (arrows). Ectopic expression of chordin or goosecoid was rarely induced in injected embryos, however in 2 out of 14 cases ventral injection of xbm plcub mRNA induced expression of chordin on the ventral side of the embryo (D; arrowheads), and in 1 case out of 11 ectopic goosecoid expression was induced by ventral injection of SpANcub mRNA injection (F; arrowheads).

145 This page intentionally left blank

146 Figure 6.5 XBMPICub induces neural tissue in isolated animal caps

Embryos were injected with 1.5 ng of the indicated mRNAs into each blastomere at the 2- cell stage, caps were cut from mid-blastula stage embryos and cultured until stage 27 when they were analyzed for expression of cement gland or neural markers.

WISH analysis shows that SpANcub (B), xbm plcub (C) and xldcub (D) mRNA-injected caps express XAl, a cement gland marker, although expression is weak and patchy in SpANcub and xldcub mRNA-injected VMZs.

(E) RPA in which total RNA isolated from injected and uninjected caps was hybridized with probes for XAG, another cement gland marker, and n rp l, a pan-neural marker. Levels of ornithine decarboxylase {ODC) mRNA were determined as a loading control. XBMPICub (lane 3) induces both XAG and nrpl expression in animal caps, confirming that XBMPICub is a more potent inducer of neural tissue in animal caps than SpANCub or XldCub which do not induce the expression of XAG or nrpl (lanes 2 and 4).

Figure 6.6 XBMPICub rescues anterior truncations caused by injection of xbmpl mRNA

Embryos were injected with 0.75 ng xb m p l mRNA with or without 0.75 ng xbm plcub mRNA and incubated until sibling control embryos had reached late tailbud stage. Embryos injected with xb m p l mRNA display mild anterior truncations such as reduced head size (A). In addition, many embryos exhibit morphological defects resulting in curled embryos with spina bifida, exposing yolky endodermal tissue on the dorsal side. These embryos often display reduced heads (arrow; C) and cement glands. Coinjection with xbm plcub mRNA generally rescues these anterior truncations and induces enlarged cement glands (B). However, XBMPICub does not rescue other morphological defects such as the spina bifida-like phenotype and curled axis.

147 Figure 6.5

control SpANCUB ffl 3 U CO z 3 < u Q. m 2 C/) X X nrp1

XAG ODC #e# XBMPICUB XldCUB

Figure 6.6

XBMP1 + XBMP1 XBMPICUB

148 Chapter 7

CHARACTERIZATION OF UNIVIN AND suBMP2/4, TWO TGFp FAMILY MEMBERS FROM SEA URCHINS

7.1 Introduction

Previous chapters have provided evidence that SpAN acts to regulate BMP activity in the Xenopus embryo, but SpAN, of course, is a sea urchin protein. This raises the question as to whether SpAN regulates BMP activity in sea urchins and if so what are its endogenous targets?

7.1.1 TGFp-related genes in sea urchins

Results from the Angerer lab (University of Rochester, USA) suggest that a BMP signalling pathway may function during axis determination in the early sea urchin embryo (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted). Injection of bmp4 mRNA into sea urchin embryos causes an expansion of ectodermal structures at the expense of vegetal- derived structures, whereas injection of mRNA for noggin, a BMP antagonist, gives the opposite phenotype - an expansion of vegetal-derived structures at the expense of ectoderm. Injection of Sp A N mRNA into sea urchin embryos elicits a phenotype very similar to bmp4 mRNA injection, whereas injection of antibodies which block the activity of BPIO (LePage et al., 1992), a SpAN homologue, causes a vegetalized phenotype similar to that seen with noggin mRNA injection. These results indicate that SpAN may act in a BMP pathway in the early sea urchin embryo.

At the start of this study, only one member of the TGFp superfamily, Univin (Stenzel et al., 1994), had been described in sea urchins, univin was isolated from Strongylocentrotus purpumtus embryos in a screen for members of the TGFp family. It is expressed at high levels in the embryo until mesenchyme bias tula stages when levels begin to fall. During early cleavage stages univin is expressed throughout the embryo, but during blastula stages transcripts are first lost from the vegetal and then the animal pole, so that by hatching blastula stages transcripts are localized to a circumequatorial band (Stenzel et al., 1994). Sequence analysis places Univin in the Vgl subgroup of the TGFp superfamily (see figure 1.1), being most closely related to zDVRl, a zebrafish homologue of Xenopus Vgl, with 70% amino acid identity in the mature C-terminal domain (figure 7.1).

In Xenopus Vgl is a candidate mesoderm inducing factor. During early development, Vgl mRNA and precursor protein are abundantly expressed in the vegetal hemisphere, which is the source of the endogenous mesoderm inducing signal (Weeks and Melton, 1987; Dale et al., 1989; Tannahill and Melton, 1989). Members of the TGFp family form disulphide-

149 linked dimers which are subsequently cleaved to release the N-terminal proregion and the mature, bioactive C-terminal domain. In Xenopus oocytes and embryos, however, all detectable Vgl is in the larger, inactive precursor form and the mature protein is not seen. Forms of Vgl in which the proregion has been replaced by the proregion of BMP2, BMP4 (BVgl) or Activin (AVgl) are processed by the embryo and oocytes, and have potent dorsal mesoderm inducing activity in animal caps. In addition injection of bvgl mRNA into Xenopus embryos will induce a secondary axis, and rescue the dorsal axis in UV irradiated embryos (Dale et al., 1993; Thomsen and Melton, 1993). This suggests a role for Vgl in early mesoderm induction and patterning. The zebrafish homologue of Vgl, zDVRl, is expressed during early development, but as with Xenopus Vgl the mature bioactive protein is not detected in the zebrafish embryo (Helde and Grunwald, 1993; Dorhmann et al., 1996). Chimeric zDVRl, in which the proregion is replaced by the BMP2 proregion, is efficiently processed in the zebrafish embryo and induces ectopic gsc expression, an early dorsal marker (Dorhmann et al., 1996). In contrast, wild-type zDVRl is processed into the mature C-terminal peptide when injected into Xenopus embryos and induces dorsal mesoderm in isolated animal caps (Dorhmann et al., 1996).

Since Univin is closely related to zDVRl and falls within the Vgl subgroup of the TGFp superfamily in pile up analysis, it may be a sea urchin Vgl homologue (see figure 1.1). However, Univin is more closely related to BMP2 (66% identity in the C-terminal domain) than Xenopus Vgl (64% identity) and may therefore be a BMP2 homologue. In the Xenopus embryo, these two molecules have quite different activities, with mature Vgl inducing dorsal mesoderm and BMPs inducing ventral mesoderm, so it should be possible to distinguish between these two possibilities by expressing Univin in the Xenopus embryo. In addition SpAN and Univin have overlapping expression patterns raising the possibility that SpAN regulates Univin activity in the sea urchin embryo. To test this I began this study by injecting univin mRNA into Xenopus embryos. s^per During the course of the study a second TGFP ^family member was isolated from sea urchins and made available for testing in the Xenopus system (C. Logan and D. McClay, unpublished). This TGFp ^family member, named suBMP2/4, was isolated from Lytechinus variegatus embryos in a screen for BMP4 homologues. Sequence analysis places suBMP2/4 in the Dpp subgroup (see figure 1.1), and shows it is more closely related to hBMP2 (88% identity in the C-terminal domain), hBMP4 (88% identity) and Drosophila Dpp (75% identity) than to Univin (68%; figure 7.1), making suBMP2/4 a good candidate for the sea urchin homologue of BMP2 and/or BMP4. suBMP2/4 has also been isolated from the sea urchin Strongylocentrotus purpuratus (Lynne Angerer, pers. comm.), and is expressed throughout the embryo from the end of cleavage stages with a peak of expression at mesenchyme blastula stages. Thus suBMP2/4 has an overlapping domain of expression with SpAN in the early S.purpuratus embryo, once again raising the possibility that SpAN may regulate suBMP2/4 activity in the sea urchin embryo.

150 The role of Univin and suBMP2/4 in sea urchin development is not clear. This chapter aims to characterize the activities of Univin and suBMP2/4 in early Xenopus embryo to establish which class of TGFp family members these proteins belong: the BMP class, which in the Xenopus embryo induces ventral mesoderm (Dale et ah, 1992; Jones et ah, 1992a; Clement et ah, 1995; Hemmati-Brivanlou and Thomsen, 1995 ; Holley et ah, 1995), or the Vgl class which induces dorsal mesoderm (Thomsen and Melton, 1993; Dale et ah, 1993; Dorhmann et ah, 1996; Seleiro et ah, 1996).

7.2 Results

7.2.1 Sea urchin BMP2/4 ventralizes Xenopus embryos

To test the activity of suBMP2/4 and Univin, Xenopus embryos were injected at the 1-cell stage with 1.6 ng of submp2/4 or univin mRNA and their development observed. Embryos injected with submp2/4 mRNA develop normally until mid-gastrulae, when gastrulation movements begin to slow in some embryos, similar to the phenotype seen in SpA N or bmp4 mRNA-injected embryos (see 3.2.1). In contrast, univin mRNA-injected embryos show no gastrulation defects. At tailbud stages embryos were scored using the DAI scoring system of Kao and Elinson (1988; see materials and methods). Embryos injected with 1.6 ng submp2/4 mRNA exhibit anterior truncations at tailbud stages (figure 7.2/1), with an average DAI score of 4.00 (table 7.1 figure 7.2A). This effect is concentration dependent, since embryos injected with a five-fold lower concentration of submp2/4 mRNA (320 pg) have a mean DAI score of 4.70 (table 7.1). Embryos injected with univin mRNA develop to tailbud stages normally, having a mean DAI score of 5.00 (table 7.1; figure 7.2C). Histological sections of submp2/4 mRNA-injected embryos show they have reduced dorsal structures such as notochord and neural tube, and occasionally somites fuse across the midline of the embryo due to a complete lack of notochord (figure 7.2B). univin mRNA- injected embryos, on the other hand, show no abnormalities when sectioned (figure 7.2D).

These results demonstrate that suBMP2/4 is able to weakly ventralize Xenopus embryos, while Univin has no effect, suggesting that suBMP2/4 is most likely the sea urchin homologue of vertebrate BMP2 and/or BMP4.

7.2.2 Bunivin dorsalizes Xenopus embryos

As described above, Univin is closely related to zebrafish zDVRl and its Xenopus homologue, Vgl. When Vgl is expressed in Xenopus embryos it has no observable effect on development, probably due to it not being processed into a bioactive form. The same is true of Chick Vgl, which has no effect on development when overexpressed in Xenopus embryos (Seleiro et al., 1996). Similarly when zdvrl mRNA is injected into zebrafish embryos it is not processed into a mature bioactive form and has no effect on development,

151 although expression in the Xenopus embryo leads to proper processing of zDVRl and induction of dorsal mesoderm. This suggests that Univin may not be active in the X enopus system because it is not properly processed into a bioactive form. To test this a chimeric constmct of Univin (Bunivin) was made in which the bioactive C-terminal domain of Univin was fused to the preproregion of Xenopus BMP4 (see materials and methods).

Xenopus embryos were injected with 320 pg bunivin mRNA at the 1-cell stage, or into a single ventral blastomere at the 4-cell stage and allowed to develop to tailbud stages. A fairly low concentration of bunivin mRNA was injected since BVgl, BzDVRl and chick BVgl are able to induce dorsal mesoderm at picogram concentrations (Dale et al., 1993; Thomsen and Melton, 1993; Dorhmann et al., 1996; Seleiro et al., 1996). Of those embryos injected with univin mRNA, many die during gastrulae stages (45%), due to exogastmlation (GD; table 7.2). This phenotype is frequently seen when b vg l or chick bvgl mRNA is injected into Xenopus embryos (L. Dale, pers. comm.; Seleiro et al., 1996). Of those that survive most develop normally and show no external signs of hyperdorsalization (table 7.2). However, a small number (7%) of those embryos injected with bvgl mRNA into a single ventral blastomere had a bent axis, indicating weak dorsalization (DAI = 6; Kao and Elinson, 1988; see materials and methods). When these embryos were sectioned, they were found to contain enlarged somites on one side of the embryo (figure 7.3B) which frequently extend laterally down the body wall. In addition, 2% of ventrally-injected embryos had a partial secondary axis (figure 7 .3 0 , which when sectioned were found to contain muscle and small amounts of neural tissue, but not notochord (figure 7.2E ).

These results indicate that Bunivin is able to weakly dorsalize Xenopus embryos, and although BVgl is a more potent dorsalizing factor, suggests that Univin may be a sea urchin homologue of Vgl.

7.2.3 Bunivin induces dorsal mesoderm in isolated animal caps

To further test the functional homology between BVgl and Bunivin, suBMP2/4 and vertebrate BMP2 and 4, Xenopus embryos were injected at the 1-cell stage with 1.6 ng univin, bunivin, bvgl, submp2/4 or xbmp2 mRNA and animal caps explanted from mid- blastulae. These were incubated until sibling controls had reached stage 27 when they were analyzed for the expression of the dorsal mesodermal marker, a-actin, and aT4-globin, a ventral mesodermal marker.

Animal caps from uninjected and univin mRNA-injected embryos round up into balls of epidermis (figure 7.4 A and B), whereas caps from embryos injected with b vg l or bunivin mRNAs elongate, consistent with dorsal mesoderm having formed (figure 7.4 C and D). Animal caps from embryos injected with xbmp2 or submp2/4 mRNA do not elongate, and

152 instead form ovoid structures with fluid-filled vesicles, consistent with ventral mesoderm having formed (figure 7.4 E and F).

RPA analysis confirms that animal caps from uninjected and univin mRNA-injected embryos do not express mesodermal markers (figure 7.4G, lanes 1 and 2), whereas b vg l or bunivin mRNA-injected animal caps express high levels of a-actin, but not aT4-globin, indicating that both BVgl and Bunivin induce dorsal mesoderm (figure 7.4G, lanes 3 and 4; see also Dale et al. 1993; Thomsen and Melton, 1993). Animal caps injected with xbm p2 mRNA express aT4-globin (figure 7.4G, lane 6; see also Clement et al. 1995; Hemmati- Brivanlou and Thomsen, 1995). However, animal caps injected with submp2/4 do not express aT4-globin (figure 7.4G, lane 5), despite the morphology of submp2/4 mRNA- injected caps being similar to xbmp2 mRNA-injected caps. This may be because XBMP2 is a stronger ventralizer than suBMP2/4 (Clement et al., 1995; section 7.2.i), consequently if more submp2/4 mRNA had been injected in this experiment aT4-globin expression may have been induced. In addition, RPA analysis shows animal caps injected with 1.6 ng bunivin mRNA weakly express endodermin, an endodermal marker (Sasai et al., 1996; figure 7.4H). This provides further evidence that Univin is homologous to Vgl, since BVgl also induces endoderm in animal cap explants (Henry et al., 1996).

These results suggest that Univin is a functional homologue of Vgl, and that suBMP2/4 is a functional homologue of vertebrate BMP2 or BMP4.

7.3 Discussion

7.3.1 suBMP2/4 is a vertebrate BMP2/4 homologue and Univin is a Vgl homologue

The results presented in this chapter demonstrate that the sea urchin protein suBMP2/4 has similar activity to vertebrate BMPs when overexpressed in Xenopus embryos. In addition a processed, mature version of the sea urchin protein Univin acts in a similar way to mature V gl in the Xenopus embryo. submp2/4 mRNA injection in Xenopus embryos produces a phenotype similar to that seen when bmp4 or bmp2 mRNAs are injected, causing anterior truncations. This and the fact that suBMP2/4 is most closely related to BMP2 and BMP4 indicates that suBMP2/4 is a sea urchin homologue of vertebrate BMP2 and 4.

Univin is most closely related to the zebrafish Vgl homologue, zDVRl (70% identity in the C-terminal bioactive domain), and to suBMP2/4 (68% identity), and pile up analysis places Univin in the same group as zDVRl, chick Vgl and Xenopus V gl (figure 1.1). At the start of this study it was not clear whether Univin was a Vgl or BMP2/4 homologue. Using the Xenopus system to test functional homology has indicated that Univin has a Vgl-like 153 activity and provided evidence that Univin may be a sea urchin homologue of Vgl. Firstly, the wild-type version of Univin is not active in Ûvt Xenopus embryo, like Vgl, presumably because it is not processed to a mature bioactive protein by removal of the proregion. Vgl can be made active by replacing its proregion with the proregion of BMP2, BMP4 or Activin, which are readily processed in the Xenopus embryo and oocyte (Thomsen and Melton, 1993; Dale et al., 1993; Kessler and Melton, 1995). A similar construct, consisting of the BMP4 proregion and Univin bioactive domain, causes mild hyperdorsalization when injected into Xenopus embryos and occasional secondary axis formation, consistent with Bunivin dorsalizing ventral mesoderm in a similar way to BVgl. In addition, Bunivin induces dorsal mesoderm and endoderm in isolated animal caps, again consistent with a BVgl-like activity.

7.3.2 What is the role of Univin and suBMP2/4 in sea urchin development?

In X enopus, Vgl is a candidate mesoderm inducing factor, but little is known about the role of Univin in the sea urchin embryo, univin transcripts are present maternally and are initially expressed uniformly throughout the embryo. After early blastula stages univin expression becomes restricted to the equatorial cells and is undetectable in vegetal and animal regions. Levels of univin then begin to drop during mesenchyme blastulae stages to extremely low levels in pluteus larvae (Stenzel et al., 1994).

By blastula stages in the sea urchin five separate territories can be distinguished, the aboral ectoderm, the oral ectoderm, the vegetal plate, the primary skeletal mesenchyme and the small micromeres (reviewed by Cameron and Davidson, 1991). Until recently it was thought that these territories were specified as early as the 60 cell stage by a propagated signal originating from the vegetal pole, as discussed in Chapter 1 (Davidson, 1989). Although this model is probably unnecessarily complicated, there is clearly a signal(s) released by the micromeres and veg tiers which is able to induce a range of tissues in the animal pole tissue (reviewed by Davidson, 1989; Livingston and Wilt, 1990). This signal may be analogous to the mesoderm inducing signal(s) acting in early Xenopus development. As yet, the molecules responsible for early signalling events in sea urchins have not been identified. Since Univin is present throughout early development it may be a candidate for an early signalling molecule, and even though univin mRNA is not localized at these early stages (Stenzel et al., 1994), its protein or protein activity maybe.

In Xenopus BMP4 is implicated in dorsal-ventral mesodermal patterning, but in sea urchins the role of suBMP2/4 is uncertain. Injection of suBMP2/4 into sea urchin embryos causes them to develop with a reduced vegetal domain and an expanded ectodermal domain which has mainly aboral character, whilst injection of the BMP antagonist, noggin, causes the opposite phenotype (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted). These results suggest that BMPs may block the vegetal signal, which itself could be a BMP

154 antagonist such as noggin. As yet no inhibitory BMP binding proteins have been isolated in sea urchins, although the phenotype of noggin mRNA injected-embryos suggests it is recognizing an endogenous target in the sea urchin, possibly suBMP2/4. These observations may indicate that there are two opposing signals emanating from each pole of the embryo, which act to pattern the early embryo. Runnstrdm originally proposed a two gradient model of early sea urchin development in 1929 (reviewed in Horstadius, 1973), based on classical isolation experiments. However, this model has fallen out of favour (Davidson, 1989; Livingston and Wilt, 1990) and a propagated signal emanating from the vegetal pole is now the preferred model. In light of the results in sea urchins described above, and recent studies in Xenopus, zebrafish and Drosophila, which have found that two opposing activities, BMPs and their antagonists, act to pattern the early embryo (reviewed by De Robertis and Sasai, 1996; Holley and Ferguson, 1997), the two signal model of sea urchin development should perhaps be reconsidered. Alternatively, BMPs may act later in development to pattern the oral-aboral axis, once this has been established by a vegetal signal, similar to the way that BMPs and their antagonists pattern mesoderm after it has been induced in the Xenopus embryo.

7.3.3 Does SpAN regulate suBMP2/4 and Univin activity?

Both univin and submp2/4 have expression domains that at some time spatially and temporally overlap with SpA N expression. Since suBMP2/4 acts as a BMP it is a likely candidate for regulation by SpAN, through cleavage of BMP inhibitory binding proteins. SpAN transcripts, however, are expressed earlier during cleavage than submp2/4, disappearing during mesenchyme blastula stages when submp2/4 expression peaks, and so perhaps Univin or another, unidentified BMP-like factor, expressed earlier in the embryo is an endogenous target for SpAN. A better candidate for suBMP2/4 regulation later in development may be suBMPl (Hwang et al., 1994), which is more closely related to XBMPl and Xolloid than to SpAN. XBMPl and Xolloid have been shown to regulate BMP activity in the Xenopus embryo (Chapter 5; Goodman et al., 1998; Piccolo et al., 1997), and suBMPI is maximally expressed from hatching blastulae stages through to mesenchyme blastulae stages, when suBMP2/4 expression peaks.

155 Table 7.1 Injection of submp2/4 mRNA into Xenopus embryos causes anterior truncations

DAI Score RNA 0 1 2 3 4 5 mean DAI n submp2/4 (1.6 ng) 1 0 7 7 24 25 4.00 64 submp2/4 (320 pg) 0 0 1 1 3 22 4.70 27 univin (1.6ng) 0 0 0 0 0 58 5.00 58

Embryos were injected with indicated amounts of submp2/4 or univin mRNA and scored at stage 26-30 for dorsoanterior defects using the DAI scoring system of Kao and Elinson (1988; see materials and methods). submp2/4 mRNA injected embryos exhibit moderate anterior truncations, and this effect is concentration dependent. Numbers in table indicate the number of cases at each DAI score, pooled from a total of 2 experiments, n = total number of embryos scored.

156 Table 7.2 Injection of bunivin mRNA weakly dorsalizes Xenopus embryos

RNA GD bent axis 2° axis mean DAI Bunivin 46 5.13 102

Embryos were injected with 320 pg of bunivin or univin mRNA at the 1-cell stage or into a single ventral blastomere at the 4-cell stage. Embryos were scored for exogastmlation (GD) and those that survived to tailbud stages were scored for dorsoanterior defects using the DAI scoring system of Kao and Elinson (1988; see materials and methods). Embryos injected with 320 pg of bunivin mRNA occasionally have a bent axis, which is indicative of moderate dorsalization and score DAI 6. Two embryos also had a partial secondary axis. Numbers in table indicate the number of cases at each DAI score, pooled from a total of 4 experiments; n = total number of embryos scored.

157 Figure 7.1 Alignment of the C-terminal (CT) regions of Univin and suBMP2/4 with other closely related TGF-p family members.

10 20 30 ZDVR1.CT 0 K P R R Y D F KDVGWQDWI I A P Q.CY L A N Y C H G E 0 P VgI.CT |C K K R H^l7y E f; K :|XV G W Q N W V I A'‘PîQ V y m;A’N Y c Y^GE'C P Univin.CT Q R H R’L F S %■ R j x T V c r w E NiWfi. i ivi Q?À; Y Y,0,D:G Ë CT SUBMP2/4.CT O R Ri.H PfLÿY D F S D-V, H .W N b W^lyV'VPI A rG'ïY Q A Y Y C H,GEO P hBMP4.CT C)R R H SKjY SDVGWNDWIV AJJ>,P ^ Y Q -AI F Y C H 'Gl OOP hBMP2.CT O K R^H P^C S. D V G W N D W I V, A pi P ?G Y H,A] F'Y O HTQ"ë>0'P' DPP.CT C R r 'h S (L V" D F, S/D V.G W D D W"l V A P L G Y D Y / 0 H G'kVc P

40 50 60 70 ZDVR1.CT S 1^ S c Ni G T^N H A I L Q f'L' V;H Si F T DP.K P G.0 O V> I Vgl.CT T E : I IL'n Ig s N"H A iH Q T LIV^H S I E »P! E DI P L P 0':%^V p T Univin.CT G:E R*L NiG T N HfPf'P I Q T L V'N S I D N R A p ,/ T SUBMP2/4.CT H^L N]T T n HW j VQTLVNSVNPA L V p K A c OiGjp^T hBMP4.CT DT L N S T N H A I VQTLVNSVN - SS I pIk A c C V P T hBMP2.CT D|H L^N T^N H'A rv Q T«L V N 'S*V‘N - S K I P K A C O'V P T DPP.CT DÏH\ F^N S J N H ATV'V Q T L V Nj N M^N P G K V P K A ^'C O V P T

80 90 100 ZÜVR1.CT K L S S y d n n d n v . v A ' r h y ed m V,V D E c G d R Vgl.CT K M S S M L fp Y D N N D N V w "l ‘ r H Y E N M, A V D E C G C ""R' Univin.CT K'L S S M L Y. FI Dd I]nn ;n N E N V V R Q Y E D M V^V E A 0 G C R; SUBMP2/4.CT E L S, s M L Y U'd 'e y E K N Y Q DM V'v J e‘g_0 G O R hBMP4.CT E L si S^M L Y Y D K V .V .L K N' Y Q E M V V E C G C R hBMP2.CT E ;L\s] eIMjL y |E ?N:E K V V L K N Y Q D M V G C G O R DPP.CT Q . L ; D A M L Y L N D Q S T V V L K N .Y Q E M T V: V G C G C R

Alignment begins with the first conserved cysteine of each protein, and numbering is from this cysteine residue. Amino acid residues which are identical in 4 or more of the proteins are shaded in dark grey, conserved changes are shaded in light grey.

References: zDVRl (Helde and Grunwald, 1993); Vgl (Weeks and Melton, 1987); Univin (Stenzel et al., 1994); suBMP2/4 (C. Logan and D. McClay, unpublished); hBMP4, hBMP2 (Wozney et al., 1988); Dpp (Padgett et al., 1987).

158 Figuré 7.2 Injection of submp2/4 mRNA causes anterior truncations in Xenopus embryos

Embryos were injected at the 1-cell stage with either 1.6 ng submp2/4 mRNA or univin mRNA. (A) shows submp2/4 mRNA-injected embryos at tailbud stages with increasingly severe anterior truncations. When allowed to develop to stage 40 and sectioned these embryos contain reduced neural tube and notochord, and occasionally somites (so) fused across the midline, due a complete lack of notochord (B). (C) shows univin mRNA- injected embryos at tailbud stages which have no axial or anterior defects. When sectioned these embryos also have no obvious axial defects (D).

Figure 7.3 Bunivin weakly dorsalizes Xenopus embryos

Embryos were injected with 320 pg bunivin mRNA into a single ventral blastomere at the 4-cell stage and allowed to develop to tadpole stages. The majority of these embryos are normal, however a small number develop with a bent axis and/or slightly enlarged head (A), indicating weak dorsalization. When sectioned these embryos contain enlarged somites on one side of the embryo (arrowhead; B) which often extend laterally down the side of the embryo. Very rarely bunivin mRNA-injection induces a partial secondary axis (2°; C, D and E), which when sectioned contains muscle and some neural tissue, but never notochord (E).

159 Figure 7.2

SUBMP2/4

univin

Figure 7.3

Bunivin

O

160 Figure 7.4 Bunivin induces dorsal mesoderm and endoderm in isolated animal caps

Embryos were injected with 1.6 ng of either univin, bunivin, bvgl, xbmp2 or submp2/4 mRNA at the 1-cell stage. Animal caps were isolated from mid-blastulae and incubated until sibling control embryos had reached tailbud stages (stage 27), when they were analyzed for expression of the dorsal mesodermal marker a-actin, or the ventral mesodermal marker aT4-globin. bunivin mRNA-injected caps were also analyzed for the expression of endodermin, an endodermal marker. Cytoskeletal actin was determined as a loading control for muscle actin and aT4-globin, and ODC as a loading control for endodermin.

Un injected or univin mRNA-injected animal caps round up into balls of epithelium (A and B), whereas bunivin and bvgl mRNA-injected animal caps undergo extensive morphogenetic movements and elongate (C and D), consistent with dorsal mesoderm having formed. In contrast, xbmp2 or submp2/4 mRNA-injected animal caps form ovoid structures with fluid-filled vesicles (B and F), consistent with ventral mesoderm having formed.

These results were confirmed by RPA analysis (G). Uninjected and univin mRNA-injected animal caps do not express mesodermal markers (lanes 1 and 2), whereas bunivin and bvgl mRNA-injected animal caps express the dorsal marker, a-actin, but not the ventral marker oT4-globin (lanes 3 and 4). xbmp2 mRNA-injected animal caps express the ventral mesodermal marker aT4-globin (lane 5), however, submp2/4 mRNA-injected caps do not (lane 6). At the concentration used suBMP2/4 is unable to induce aT4-globin expression in animal caps, although both animal caps from XBMP2 and suBMP2/4 embryos have a similar morphology, and it is possible at higher concentrations suBMP2/4 would induce ventral markers.

In addition RPA analysis (H) shows that like BVgl (Henry et al., 1996), Bunivin induces endoderm in animal caps, as shown by the expression of endodermin.

161 Figure 7.4

uninjected univin ÜV. BVg1 Bunivin

XBMP2 SUBMP2/4 FT

H ti m-actin endodermin #

T4-globin ODC f§ I 1 2 c-actin

162 Chapter 8

GENERAL DISCUSSION

BMP signalling is important during dorsal-ventral patterning in the early embryos of Drosophila, Xenopus and the zebrafish, Danio rerio, and this conserved pathway is likely to be important during development of other animal species as well. The activity of BMPs is regulated by inhibitory binding proteins, such as Chordin, Noggin and Follistatin, which control the availability of free BMP ligand for its receptor (Piccolo et al., 1996; Zimmerman et al., 1996; Holley et al., 1996; Fainsod et al., 1997). Recent studies have indicated that an additional layer of regulation exists, where BMP 1-related metalloproteases inhibit the activity of these binding proteins through proteolytic cleavage, thereby enhancing BMP activity (Piccolo et al., 1997; Marqués et al., 1997; Blader et al., 1997). This thesis aimed to further characterize the role of BMP 1-related metalloproteases in regulating the BMP signalling pathway.

Since dorsal-ventral patterning in Xenopus embryos is well characterized, and involves a BMP signalling pathway, the early Xenopus embryo was used as a test system to look at the effects of overexpressing XBMPl, Xolloid, and the related sea urchin metalloprotease SpAN. The results indicate that BMP 1-related metalloproteases act to enhance activity of TFGp-related BMPs. As well as overexpression studies, putative dominant-negative constructs were expressed in Xenopus embryos, in order to block endogenous metalloprotease activity, providing further evidence that BMP 1-related metalloproteases enhance TFG(3-related BMP activity. In addition, two sea urchin BMP-like genes, univin and submp2/4, which are potential candidates for regulation by SpAN in the sea urchin embryo, were overexpressed in the Xenopus embryo and found to be functionally homologous to Vgl and vertebrate BMP2 and 4 respectively. The possible conservation of the BMP signalling pathway and patterning mechanisms in Xenopus and sea urchin embryos is discussed, and a model for the role of inhibitory binding proteins and BM Pl- related metalloproteases in patterning Xenopus mesoderm is proposed.

8.1 Effects of overexpressing metalloproteases in the X enopus embryo.

The results presented in Chapter 3 demonstrate that injection of SpAN mRNA into Xenopus embryos causes ventralization of dorsal mesoderm, both in the whole embryo and Activin-treated animal caps. Recent results from this lab have also shown that XBMPl and Xolloid are able to weakly ventralize Xenopus embryos, causing anterior truncations and dorsal axial defects (Goodman et al., 1998; Appendix 1). In addition, XBMPl and Xolloid

163 ventralize the dorsal mesoderm induced by Activin in animal caps. The similarity of phenotypes between injection of mRNAs for these metalloproteases and mRNAs for xbmp2, xbmp4 or xbmpV (Dale et al., 1992; Jones et al., 1992a; Clement et al., 1995; Hawley et al., 1995; Hemmati-Brivanlou and Thomsen, 1995; Jones et al., 1996b), suggests that SpAN, XBMPl and Xolloid may act to enhance endogenous Xenopus BMP activity. Chapter 5 provides evidence for this, demonstrating that SpAN, XBMPl and Xolloid act upstream of a BMP receptor and require a functional BMP signalling pathway for their ventralizing activity.

The results presented in Chapter 5 also suggest that SpAN, XBMPl and Xolloid enhance BMP activity by inhibiting Chordin, and in the case of SpAN by also inhibiting Noggin. While no evidence is provided as to how these metalloproteases may inhibit Chordin or Noggin activity, recent experiments have shown that Chordin is cleaved by both Xolloid and zebrafish Tolloid (Piccolo et al., 1997; Blader et al., 1997), while Drosophila Tolloid cleaves Sog, a Chordin homologue (Marqués et al., 1997). Taken together, these results suggest that SpAN and XBMPl may also be able to cleave Chordin and that SpAN may cleave Noggin. This could be tested by using epitope-tagged versions of Chordin and Noggin proteins and incubating them with XBMPl or SpAN. For instance, using myc- tagged constructs would allow any cleavage products to be easily detected by Western blot using the 9E10 anti-myc antibody. One way to obtain SpAN, XBMPl, Chordin-myc and Noggin-myc proteins would be to transfect cells with the appropriate constructs and collect conditioned media. This method has been used successfully by Marqués et al. (1997), using insect cell conditioned media, and by Blader et al. (1997), using COS cell conditioned media, to demonstrate that Tolloid and zebrafish Tolloid cleave Sog and Chordin respectively. Piccolo et al. (1997) also used conditioned media to demonstrate that Xolloid cleaves Chordin, but took the additional step of partially purifying each protein. It may also be possible to obtain conditioned media from Xenopus oocytes injected with mRNA for the required proteins, or use a Xenopus cell-free extract to manufacture the proteins from synthetic mRNA (Matthews and Colman, 1991). This would bypass the need to make additional constructs for transfection into insect or mammalian cells, allowing the use of existing Xenopus expression constructs. In addition, these experiments could be extended to Follistatin, which also binds and inhibits BMP4 (Fainsod et al., 1997).

Piccolo et al. (1997) utilized specific antibodies that recognize the N-terminus of Chordin, an internal site in Chordin, and a C-terminal myc epitope, which allowed these workers to identify approximately where Xolloid cuts in the Chordin protein. It would be interesting to see whether BMPl and SpAN, if they are also able to cleave Chordin, generate the same size fragments, which would indicate that they cut at, or near, the same sites as Xolloid. Following on from this, anti-Chordin antibodies could be used to immunoprecipitate each fragment generated by cleavage, and the N-terminal amino acid sequences determined using Edman degradation. Similar methods have been used previously to determine the cleavage 164 sites of BMPl (PCP) in pro-Collagens I, II and III and in pro-Lysyl Oxidase (Hojima et al., 1985; Kessler et al., 1986; Panchenko et al., 1996). Identifying the sequence around the cleavage site would not only show whether XBMPl, Xolloid and SpAN all recognize the same sites in Chordin, but also tell us whether these sites are similar to, or the same as, the sites recognized by BMPl in pro-Collagen and pro-Lysyl Oxidase.

As yet, only SpAN, a sea urchin protein, has been shown to inhibit Noggin activity, raising the question as to how Noggin activity is regulated in the Xenopus embryo. Noggin has some homology with proteins from the Kunitz domain family, which includes serine protease inhibitors (McDonald and Kwong, 1993), perhaps indicating that Noggin’s activity is regulated by a serine protease. Another possibility is that a Xenopus homologue of SpAN exists to regulate Noggin activity. The C-terminal domain of SpAN is distinct from most other BMPl-related metalloproteases, having an additional threonine rich domain, with sea urchin BPIO and C. elegans Hchl being the only other metalloproteases so far isolated to have such a domain. This may mean that SpAN’s function is to regulate the activity of proteins other than Chordin that are not recognized by other metalloproteases more closely related to BMPl. Two preliminary screens to isolate a Xenopus SpAN homologue were begun as part of this study. Firstly, a neurula stage cDNA phage library was screened, using a random primed probe for SpAN. In addition, PCR was used in an attempt to isolate clones with a C-terminal domain similar to SpAN from a gastrula stage cDNA plasmid library, however, neither screen yielded any positive results. While these preliminary screens do not rule out the possibility that a SpAN homologue exists in Xenopus, more extensive investigations are needed.

In addition to ventralizing embryos, expression of SpAN disrupts morphogenetic movements during gastrulation. This activity, unlike SpAN’s ventralizing activity, is independent of the C-terminal domain of SpAN. This was shown by injection of smp mRNA, consisting of the SpAN metalloproteases domain alone, which disrupts morphogenesis but has no effect on differentiation of dorsal mesodermal structures. Since metalloproteases are known to degrade ECM components (reviewed by Sarras, 1996), and previous experiments using specific antibodies to block Fibronectin function have indicated that Fibronectin is essential for normal gastrulation in the Xenopus embryo (Howard et al., 1992; Ramos and DeSimone, 1996; Ramos et al. 1996), the levels of Fibronectin in SpAN and smp mRNA-injected embryos were analyzed. Chapter 4 shows that injection of doses as low as 13 pg of SpAN or smp mRNA into Xenopus embryos causes a dramatic decrease in Fibronectin levels by the start of gastrulation. However, while embryos injected with 13 pg SpAN mRNA have severe anterior defects, those injected with 13 pg smp mRNA are normal. This implies that little, if any, of the Fibronectin present in the embryo is needed for normal anterior development. In addition, embryos receiving high doses of SpAN mRNA, which contain little or no Fibronectin, exhibit delayed blastopore closure, suggesting that gastrulation is perturbed in these embryos by disruption of another ECM 165 component which is essential for gastrulation, or by another, unknown mechanism. To investigate this further, future experiments would examine the effects of SpAN, XBMPl and Xolloid on the levels of other ECM molecules expressed in the early Xenopus embryo. BMPl is known to cleave the C-terminal peptide of pro-Collagens I, II and III (Hojima et al., 1985; Kessler et al., 1986), and also cleaves pro-Laminin y2 into its tissue form (Amano et al., 1997). While pro-Collagens I-III are not expressed in the Xenopus embryo until late gastrulation/neurulation (Bieker and Yazdani-Buicky, 1992; Su et al., 1991), Collagen type VI has been detected in the early embryo and anti-Collagen VI antibodies delay gastrulation movements (Otte et al., 1990), thus Collagen VI may be a target for these metalloproteases. As yet the only Laminin to be described in Xenopus (Laminin p) is expressed from mid-gastrulae around the developing somites and notochord, and so is not involved in early gastrulation movements (Fey and Hausen, 1990). However, Laminin- related peptides have been described in the newt, Pluerodeles waltii, where Laminin is detected along the blastocoel roof in blastulae and gastrulae using heterospecific antibodies, suggesting that Laminin may be important for normal gastrulation (Darribère et al., 1986; Riou et al., 1987) and that other Laminins may be present in the early Xenopus embryo.

The results presented in Chapter 4 also demonstrate that a full range of dorsal structures can form in the absence of morphogenetic movements, smp mRNA-injected animal caps treated with Activin express a-actin, despite not undergoing elongation movements, and smp mRNA-injected embryos differentiate neural tissue, notochord and muscle. However, these embryos do not contain anterior structures, such as eyes, indicating that the most anterior structures do not form in the absence of normal gastrulation movements, an idea proposed by Gerhart et al. (1989). Further evidence for this comes from experiments with Xwnt5A, which also disrupts morphogenetic movements, but not the development of dorsal mesoderm (Moon et al., 1993). When Xwnt5A is overexpressed in Xenopus, embryos develop with mild anterior truncations (Moon et al., 1993), similar to the phenotype seen when xbmpl or xld mRNAs are overexpressed in Xenopus embryos (Goodman et al., 1998). The anterior truncations seen in xbmpl and xld mRNA-injected Xenopus embryos do not extend beyond the hindbrain (Goodman et al., 1998) and it would be interesting to ascertain how far truncations in smp mRNA-injected embryos extend, since this may indicate which anterior structures are dependent on full mesodermal migration. This could be determined by analyzing the expression of molecular markers such as krox 20, which is expressed in rhombomeres 3 and 5 in the hindbrain (Bradley et al., 1993), and en2, which is expressed at the midbrain-hindbrain boundary (Hemmati-Brivanlou et al., 1991).

8.2 Effects of inhibiting metalloprotease activity in the Xenopus embryo.

In order to further test the function of metalloproteases in BMP signalling, putative dominant-negative constructs of SpAN, XBMPl and Xolloid, where the metalloprotease domain was removed but the C-terminal domain was left intact, were injected into Xenopus 166 embryos. The results presented in Chapter 6 show that these constructs induce dorsal mesoderm in isolated ventral marginal zones and that XBMPlCub induces neural tissue in animal caps, consistent with endogenous BMP signalling being blocked. To show that these constmcts block BMP signalling by inhibiting the activity of metalloproteases, full length SpAN or XBMPl was coinjected with the corresponding C-terminal domain construct. The results of coinjecting SpANCub with SpAN were inconclusive since a large part of SpAN’s phenotype is independent of its C-terminal domain and is not therefore inhibited by the C-terminal domain constmct. Coinjection of XBMPlCub with XBMPl rescued the anterior tmncations caused by XBMPl, although other morphological defects, such as spina bifida, were not affected. Clearly, further experiments are required to resolve whether the C-terminal domain constmcts are indeed acting in a dominant-negative manner and there are several ways that this could be approached. To overcome the difficulties experienced in whole embryo experiments, one approach may be to coexpress the C- terminal domain constmcts with Chordin and full-length metalloprotease in ventral marginal zone (VMZ) explants. Injection of chordin mRNA alone induces dorsal mesoderm in VMZs, while coexpression with a metalloprotease, say Xolloid, blocks this activity. If the C-terminal domain constmcts are acting as dominant-inhibitors then coinjection of chordin with full-length xolloid and xldcub mRNAs should restore dorsal mesoderm in the VMZ, since Xolloid activity would be inhibited by XldCub. Alternatively, similar experiments could be done in vitro using conditioned media. For example, conditioned media for myc- tagged Chordin could be mixed with conditioned media for Xolloid, with or without XldCub conditioned media. Without XldCub, Chordin should be cleaved by Xolloid, which could be detected by Western blot using anti-myc antibodies. If XldCub were acting as a dominant-inhibitor then it should block cleavage of Chordin by Xolloid when coexpressed. Similar experiments could also be carried out with SpAN/SpANCub and XBMPl/XBMPlCub conditioned media.

8.3 The role of BMPl-like metalloproteases and BMP signalling in sea urchin embryos

Two TGFp BMP-related genes have been isolated from sea urchin embryos to date, univin (Stenzel et al., 1994) and suBMP2/4 (C. Logan and D. McClay, unpublished; L. Angerer, pers. comm.). While Univin falls within the Vgl subgroup of the TGFp superfamily, and may therefore be a sea urchin homologue of Vgl (see figure 1.1). it is also closely related to BMP2 than to Xenopus Vgl and may therefore be a BMP2 homologue. suBMP2/4 is also closely related to BMP2 and to BMP4. In Chapter 7, the Xenopus embryo was used as a test system to investigate the functional homology between these sea urchin proteins and Xenopus Vgl and BMP2. The results indicate that suBMP2/4 is a sea urchin homologue of vertebrate BMP2 and BMP4 and that Univin is homologous to Vgl.

167 Like Vgl, Univin has no activity when expressed in the Xenopus embryo, perhaps because it is not processed into its mature, bioactive form. Vgl can be made active by replacing its prodomain with the prodomain of BMP2 or BMP4 or Activin, which are readily processed in the Xenopus embryo and oocyte (Thomsen and Melton, 1993; Dale et al., 1993; Kessler and Melton, 1996). Chapter 7 demonstrates that a similar construct, Bunivin, consisting of the BMP4 prodomain and Univin bioactive domain, is active in the Xenopus embryo, inducing dorsal mesoderm in the animal cap assay. This supports the argument that wild- type Univin is not processed in the Xenopus embryo, rather than the alternative explanation that receptors are not present. Further experiments could demonstrate this by injecting mRNA for epitope tagged (e.g. myc) versions of Univin and Bunivin into oocytes and using anti-myc antibodies to detect protein by Western blot. The prediction would be that all Univin protein manufactured in the oocyte would be of the larger, precursor form (approximately 46KDa under reducing conditions), whereas a proportion of Bunivin protein would be processed into the smaller, mature form of Univin (predicted size of approximately 30KDa) and the BMP4 proregion.

The mechanism and molecular signals involved in cell fate specification in the sea urchin embryo are not known. However, similar signalling pathways to those used in the Xenopus embryo and the Drosophila embryo may be involved. In the sea urchin embryo a signal(s) is released by the micromeres and veg tiers that induces a range of tissues in the animal pole (reviewed by Davidson, 1989; Livingston and Wilt, 1990), and this signal may be analogous to the mesoderm inducing signal, e.g. Vgl, acting in early Xenopus development. Since Univin is present throughout early development, and is functionally homologous to Vgl when expressed in the Xenopus embryo, it may be a candidate for the vegetal signalling molecule in sea urchins.

Alternatively, a BMP signalling pathway may be responsible for patterning the animal- vegetal axis in the early sea urchin embryo. When expressed in sea urchin embryos both suBMP2/4 and XBMP4 elicit the same phenotype, an expansion of the ectodermal territory at the expense of vegetally derived structures (L. Angerer, D. Oleskyn, L. Dale, R. Angerer, submitted), which is further evidence that suBMP2/4 is functionally homologous to BMP4. A similar phenotype is produced when SpAN is overexpressed in sea urchin embryos, consistent with SpAN activating BMP signalling, whereas injection of noggin mRNA into sea urchin embryos causes an opposite, vegetalized phenotype. These results suggest that BMPs may block the vegetal signal, which itself could be a BMP antagonist such as Noggin. As yet no inhibitory BMP binding proteins have been isolated in sea urchins, although the phenotype of noggin mRNA injected-embryos indicates that inhibitory binding proteins are able to function and recognize endogenous targets in the sea urchin embryo, possibly suBMP2/4. These observations may indicate that there are two opposing signals emanating from each pole of the embryo, acting to pattern the early embryo, an idea originally proposed by Runnstrom (1929; cited in Horstadius, 1973). 168 BMP signalling may also be involved in specifying dorsal-ventral (aboral-oral) fate in the ectoderm. Overexpression of BMP4 in sea urchin embryos leads to ectoderm of an aboral character, while overexpression of Noggin leads to oral markers being expressed throughout the ectoderm, indicating that BMPs may specify aboral fate while BMP antagonists lead to oral fate. In this case patterning in the ectoderm may be analogous to dorsal-ventral patterning of the mesoderm in the Xenopus embryo; once an oral-aboral difference in the ectoderm has been established by a vegetal signal, the BMP signalling pathway maintains this polarity and further patterns the ectoderm along the oral-aboral axis.

If a BMP signalling pathway is conserved in sea urchins, then BMPl-like metalloproteases are likely to act in this pathway, making suBMP2/4 a candidate for regulation by metalloproteases such as SpAN, through cleavage of inhibitory proteins. SpAN transcripts, however, are expressed earlier during development than submp2/4 transcripts and disappear during mesenchyme blastula stages, when submp2/4 expression peaks. A better candidate for regulating suBMP2/4 may be suBMPI (Hwang et al., 1994), which is more closely related to XBMPl and Xolloid than to SpAN. suBMP2/4 is maximally expressed from hatching blastulae stages through to mesenchyme blastulae stages, when suBMP2/4 expression peaks. Since SpAN is predominantly expressed at very early blastula stages it may instead regulate Univin, which is also expressed in the early embryo. Univin is a candidate for the vegetal signal, although univin transcripts are not localized to the vegetal tiers of the sea urchin embryo. It is possible, however, that Univin activity becomes localized in the embryo through post-translational regulation. For instance, in the Xenopus embryo it has been suggested that Vgl dorsal mesoderm inducing activity^localized to dorsal-vegetal cells through localized proteolytic cleavage of the Vgl precursor into the mature, bioactive form (Thomsen and Melton, 1993). Alternatively, Vgl/Univin activity may be regulated through interaction with an inhibitory binding protein, although as yet, inhibitory binding proteins which recognize proteins of the Vgl subgroup have not been identified. However, given that other TGPp binding proteins exist (e.g. Follistatin, Decorin), it is possible that Vgl-like proteins are also regulated in this way.

8.4 A model for the action of metalloproteases in the BMP signalling pathway

The results presented in this thesis, and other recent studies, indicate that the availability of BMP ligand for its receptor is regulated by inhibitory binding proteins and the action of BMPl-related metalloproteases (figure 8.1). BMPs are bound in the extracellular space by inhibitory proteins, such as Chordin, thereby preventing BMP interacting with and activating its receptor. This inhibition is relieved by BMPl-like metalloproteases, which cleave Chordin and release bound BMPs.

169 Regulating the activity of BMPs is important, since in both the Xenopus embryo and the Drosophila embryo BMPs are able to act as morphogens (Ferguson and Anderson, 1992a; Dosch et al., 1997). For instance, a sequence of cell fates - notochord, muscle, pronephros and blood - can be produced by expressing progressively increasing concentrations of BMP4 in the marginal zone of Xenopus embryos, while progressively inhibiting endogenous BMP signalling, by expressing Noggin or a dominant-negative BMP receptor, produces the same sequence in reverse order (Dosch et al., 1997). These results suggest that a gradient of BMP activity exists in the marginal zone of the Xenopus embryo, with high ventral activity and low dorsal activity. hmp4 transcripts, however, are uniformly expressed throughout the ventral and lateral marginal zone, indicating that BMP activity must be regulated at a post-transcriptional or post-translational level. The following model is proposed to explain how BMP activity may be regulated in the marginal zone of the Xenopus embryo. BMP4 is produced and secreted by all the cells of the ventral and lateral marginal zone and is removed by interactions with its receptors and inhibitory binding proteins such as Chordin and Noggin. In order to establish a gradient of BMP4 activity, from high ventrally to low dorsally, more BMP4 must be removed from the dorsal side of the embryo than the ventral side. This could be achieved by localizing the synthesis of inhibitory binding proteins to the dorsal side, as observed for Chordin, Noggin and Follistatin. Diffusion would establish a dorsal-ventral concentration gradient (high dorsally, low ventrally) of one or more of these molecules, which in turn would establish the required gradient of BMP4 activity. At present there is no direct evidence that these molecules can diffuse throughout the marginal zone, but Noggin has been shown to act at a distance when overexpressed in Xenopus embryos (Dosch et al., 1997), as has the Chordin homologue Sog in the Drosophila embryo (discussed in Holley et al., 1996). However, computer modelling has shown that diffusion from a localized source alone is not sufficient to establish a stable concentration gradient, rather, similar levels are ultimately attained throughout the embryo (Slack, 1991). A stable gradient can only be achieved if a ‘sink’ is included that removes the diffusing substance(s), and most models utilize a dispersed sink that is present throughout the tissue. This, then, could provide an explanation for the role of BMPl-related metalloproteases in mesodermal patterning; these molecules could act as a dispersed sink, removing Chordin activity from the marginal zone through proteolytic activity. Uniform distribution of these proteases, combined with a localized source of Chordin should lead to a stable concentration gradient of Chordin. This in turn should generate an opposing gradient of BMP4 activity. Similarly, proteases may also act as cxsink for Noggin and/or Follistatin, thereby allowing stable gradients of these molecules to be established.

170 Figure 8.1 Inhibition and activation of BMP signalling

BMPs are bound as inactive complexes in the extracellular space by inhibitory proteins, such as Chordin or Noggin. The inhibitory proteins are proteolytically cleaved by BMPl- related metalloproteases, releasing active BMPs that interact with their receptors, initiating signal transduction.

171 Figure 8.1

Inactive BMP

BMP 1-related I metalloproteases ♦ Active BMP 1 ♦

t SIGNAL TRANSDUCTION

172 Figure 8.2

(A) Chordin establishes a gradient of BMP4 activity in the dorsal marginal zone Decreasing levels of BMP4 activity specify increasingly more dorsal mesodermal structures - high BMP4 activity leads to formation of blood, intermediate levels lead to formation of pronephros and muscle, low levels lead to notochord formation. A gradient of BMP4 is established through inhibitory interactions with Chordin which has its highest activity on the dorsal side.

(B) Xolloid establishes a gradient of Chordin activity in the marginal zone Chordin is secreted by cells of the organizer, on the dorsal side of the embryo. Xolloid may act as a dispersed sink, removing Chordin from the marginal zone through proteolytic cleavage, thereby establishing a gradient of Chordin with high levels on the dorsal side and low levels on the ventral side.

173 Figure 8.2

A Chordin establishes a gradient of BMP4 activity in the marginal zone

BMP4 Chordin

Blood Pronephros Muscle Notochord

VENTRAL DORSAL

B Xolloid establishes a gradient of Chordin activity in the marginal zone

Chordin

Chordin + Xolloid _

VENTRAL DORSAL

174 REFERENCES

Adams, J.C. and Watt, F.M. (1993). Regulation of development and differentiation by the extracellular matrix. Development 117, 1183-1198.

Amano, S., Takahara, K., Gerecke, D R., Nishiyama, T., Lee, S., Greenspan, D.S., Hogan, B., Birk, D.B., Burgeson, R.E. (1997). The i l chain of laminin 5 is processed by BMPl and processing is essential to basement membrane assembly in vivo. J. Invest. Derm. 108 (Suppl.), no.29. Meeting abstract.

Amaya, E., Stein, P.A., Musci, T.J. and Kirchner, M.W. (1993). FGF signalling in the early specification of mesoderm in Xenopus. Development 118, 477-487.

Angerer, L.M., Oleskyn, D., Dale, L. and Angerer, R.C. (1998). A BMP4 pathway regulates cell fate allocation along sea urchin embryonic axes. Submitted.

Aota, S., Nagai, T and Yamada, K.M. (1991). Characterization of regions of fibronectin besides the Arginine-Glycine-Aspartic acid sequence required for adhesive function of the cell binding domain using site directed mutagenesis. J. Biol. Chem. 266, 15938-15943.

Appella. E., Weber, I T. and Blasi, F. (1988). Structure and function of epidermal growth factor-like regions in proteins. FEBS Letts 231, 1-4.

Arlaud, G.J., Colomb, M.G. and Gagnon, J. (1987). A functional model of the human Cl compXtx. Immunol. Today S, 106-111.

Arlaud, G.J. and Thielens, N.M. (1993). Human complement serine proteases C lr and Cls and their proenzymes. Meth. Enzymol. 223, 61-82.

Arora, K., Levine, M.S. and O'Connor, M.B. (1994). The screw gene encodes a ubiquitously expressed member of the TGF-P family required for specification of dorsal cell fates in the Drosophila embryo. Genes Dev. 8, 2588-2601

Azar, D.T., Hahn, T.W., Jain, S., Yeh, Y.C. and Stetler-Stevenson, W.G. (1996). Matrix metalloproteinases are expressed during wound healing after excimer laser keratectomy. Cornea 15, 18-24.

Baker, J.C. and Harland, R.M. (1996). A novel mesoderm inducer, MADR2 functions in the activin signal transduction pathway. Genes Dev. 10, 1880-1889.

175 Banville, D. and Williams, J.G. (1985). The pattern of expression of the Xenopus laevis tadpole a-globin polypeptides. Nue. Acid. Res. 13, 5407-5421.

Barr, P.J. (1991). Mammalian subtilisins: The long-sought dibasic processing endopeptidases. Cell 66, 1-3.

Basler, K., Edlund, T., Jessell, T.M. and Yamada, T. (1993). Control of cell pattern in the neural tube: regulation of cell differentiation by dorsalin-1, a novel TGF-p family member.Ce// 73, 687-702.

Biehs, B., Francois, V. and Bier, F. (1996). The Drosophila short gastrulation gene prevents Dpp from autoactivating and suppressing neurogenesis in the neurectoderm. Genes Dev. 10, 2922-2934.

Bieker, J.J. and Yazdani-Buicky, M. (1992). Distribution of type II collagen mRNA in Xenopus embryos visulaized by whole mount in situ hybridization. J. Histochem. Cytochem. 40, 1117-1120.

Bier, F. (1997). Anti-neural-inhibition: a conserved mechanism for neural induction. Cell 89, 681-684.

Birkedal-Hansen, H. (1995). Proteolytic remodelling of extracellular matrix. Curr. Opin. Cell Biol. 7, 728-735.

Blader, P., Rastegar, S., Fischer, N., Strahie, U. (1997). Cleavage of the BMP-4 agonist Chordin by zebrafish Tolloid. Science 278, 1937-1940.

Blumberg, B., Wright, C.V.F., De Robertis, F.M., Cho, K.W.Y. (1991). Organizer- specific homeobox genes in Xenopus laevis embryos. Science 253, 194-196.

Bode, W., Gomis-Rüth, F.X., Huber, R., Zwilling, R. and Stocker, W. (1992). Structure of astacin and implications for activation of astacins and zinc-ligation of collagenases. Nature 35S, 164-167.

Bode, W., Gomis-Rüth, F.X. and Stocker, W. (1993). Astacins, serralysins, snake venom and matrix metalloproteinases exhibit identical zinc-binding environments (HFXXHXXGXXH and Met-turn) and topologies and should be grouped into a common family, the 'metzincins'. FEBS Lett. 331, 134-140.

Bond, J.S. and Beynon, R.J. (1995). The astacin family of . Prot. Sci. 4, 1247-1261. 176 Bonyadi, M., Rusholme, S.A.B., Cousins, P.M., Su, H.C., Biron, C.A., Farrall, M. and Akhurst, RJ. (1997). Mapping of a major genetic modifier of embryonic lethality in TGFpl knockout mice. Nature Genet. 15, 207-211.

Bork, P. and Beckmann, G. (1993). The CUB domain. A widespread module in developmentally regulated proteins. J. M o l B iol 231, 539-545.

Bornstein, P. (1992). Thrombospondins: structure and regulation of expression. FASEB J. 6, 3290-3299.

Boucaut, J-C. and Darribère, T. (1983). Fibronectin in early amphibian embryos: migrating mesodermal cells conatct fibronectin established prior to gastrulation. Cell Tissue Res. 234, 135-145.

Boucaut, J-C., Darribere, T, Boulekbache, H. and Thiery, J-P. (1984). Prevention of gastrulation but not neurulation by antibody to fibronectin in amphibian embryos. Nature 307, 364-367.

Boudreau, N. and Bissel, M.J. (1996). In ‘Extracellular Matrix’ Vol 2. (ed. W .D. Comper.). pp 246-261. Harwood, Amsterdam.

Bradley, L.C., Snape, A., Bhatt, S. and Wilkinson, D.G. (1993). The structure and expression of the Xenopus Krox-20 gene: conserved and divergent patterns of expression in rhombomeres and neural crest. Mech. Dev. 40, 73-84.

Busby, T.F. and Ingham, K.C. (1987). Calcium-sensitive thermal transitions and domain structure of human complement subcomponent Clr. Biochem. 26, 5564-5571.

Cameron, R.A. and Davidson, E.H. (1991). Cell type specification during sea urchin development. Trends Genet. 7, 212-218

Cameron, R.A., Hough-Evans, B.R., Britten, R.J. and Davidson, E.H. (1987). Lineage and fate of each blastomere of eight-cell sea urchin embryo. Genes Dev. 1, 75-84.

Celeste, A.J., lannazzi, J.A., Taylor, R.C., Hewick, R.M., Rosen, V., Wang, E.A. and Wozney, J.M. (1990). Identification of Transforming Growth-Factor-p family members present in bone-inductive protein purified from bovine bone. Proa. Natl Acad. Sci. USA 87, 9843-9847.

177 Chang, G., Wilson, P.A., Mathews, L.S. and Hemmati-Brivanlou, A. (1997). a Xenopus type 1 activin receptor mediates the mesodermal but not the neural specification during embryogenesis. Development 124, 827-837.

Chen, X., Rubock, M.J. and Whitman, M. (1996). A transcriptional partner for MAD proteins in TGFp signalling. Nature 383, 691-696.

Cherr, G.N., Summers, R.G., Baldwin, J.D. and Morrill, J.B. (1992). Preservation and visualization of sea urchin embryo blastocoelic extracellular matrix. Micros. Res. Tech. 22,

11- 22.

Chen, X., Weisberg, E., Fridmacher, V., Watanabe, M., Naco, G. and Whitman, M. (1997). Smad4 and FAST-1 in the assembly of activin-responsive factor. Nature 3 8 9 , 85- 89.

Childs, S.R. and O'Connor, M.B. (1994). The two domains of tolloid protein contribute to its unusual genetic interaction with decepentaplegic. Dev. Biol. 162, 209-220.

Cho, K.W., Blumberg, B., Steinbeisser, H. and De Robertis, B.M. (1991). Molecular nature of Spemann's organizer: the role of the Xenopus homeobox gene goosecoid. Cell 67, 1111-1120.

Christian, J.L., McMahon, J.A., McMahon, A.P. and Moon, R.T. (1991). X w nt-8, a Xenopus Wh?7/m?7-related gene responding to mesoderm inducing growth factors may play a role in ventral mesodermal patterning during embryogenesis. Development 1 11, 1045-1055.

Clark, E.A. and Brugge, J.S. (1995). Integrins and signal transduction pathways: the road taken. Science 268, 233-239.

Clement, J.H., Fettes, P., Knochel, S., Lef, J., Knbchel, W (1995). Bone morphogenetic protein 2 in the early development of Xenopus laevis. Mech. Dev. 52, 357-370.

Coffman, J.A., Nelson, S. and McClay, D R. (1985). A cell surface protein that identifies the ventral surface of the ectoderm of sea urchin gastrulae. J. Cell Biol. 101, 469a.

Conlon, F.L., Barth, K.S. and Robertson, E.J. (1991). A novel retrovirally induced embryonic lethal mutation in the mouse: assessment of the developmental fate of embryonic stem cells homozygous for the 413d proviral integration. Development 111, 969-981.

178 Conlon. F.L., Lyons, K.M., Takaesu, N., Barth, K.S., Kispert, A., Hermann, B. and Robertson, E.J. (1994). A primary requirement for nodal in the formation of and maintenance of the primitive streak in the mouse. Development 120, 1919-1928.

Cox, K.H., Angerer, L.M., Lee, J.J., Davidson, E.H. and Angerer, R.C. (1986). Cell lineage specific programs of expression of multiple actin genes during sea urchin embryogenesis. J. M o l B iol 188, 159-172.

Creighton, T.E. and Kemmink, J. (1993). Noggin is unlikely to be homologous to the Kunitz protease-inhibitor family. Trends Biochem. Sci. 18, 424-425.

Cunningham, N.S., Jenkins, N.A., Gilbert, D.J., Copeland, N.G., Reddi, A.H. and Lee, S.J. (1995). Growth/differentiation factor-10: a new member of the transforming growth factor-p superfamily related to bone morphogenetic protein-3. Growth Factors 12, 99-109.

Dale, L., Howes, G., Price, B.M.J. and Smith, J. C. (1992). Bone morphogenetic protein 4: A ventralising factor in early Xenopus development. Development 115, 573-585.

Dale, L., Matthews, G. and Coleman, A. (1993). Secretion and mesoderm-inducing activity of the TGF-p-related domain of Xenopus V gl. EMBO J 12, 4471-4480.

Dale, L., Matthews, G., Tabe, L. and Coleman, A. (1989). Developmental expression of the protein product of Vgl, a localized maternal mRNA in the frog Xenopus laevis. EMBO J. 8, 1057-1065.

Dale, L. and Slack, J.M.W. (1987a). Fate map for the 32 cell stage of Xenopus laevis. Development 99, 527-551.

Dale, L. and Slack, J.M.W. (1987b). Regional specificity within the mesoderm of early embryos of Xenopus laevis. Development 100, 279-295.

Dale, L., Smith, J.C. and Slack, J.M.W. (1985). Mesoderm induction in Xenopus laevis: a quantitative study using cell lineage label and tissue-specific antibodies. J. Embryol. exp. Morph. 89, 289-312.

Daopin, S., Fiez, K.A., Ogawa, Y. and Davies, DR. (1992). Crystal structure of transforming growth factor p2: An unusual fold for the superfamily. Science 257, 369- 373.

179 Darribère, T., Riou, J-F, Shi, D.L., Delarue, M. and Boucaut, J-C. (1986). Synthesis and distribution of laminin-related polypeptides in early amphibian embryos. Cell Tissue Res. 246, 45-51.

Davidson, E.H. (1989). Lineage-specific gene expression and the regulative capacities of the sea urchin embryo: a proposed mechanism. Development 105, 421-445.

Delgadillo-Reynoso, M.G., Rollo, D R., Hursh, D.A. and Raff, R.A. (1989). Structural analysis of the uEGF gene in the sea urchin Strongylocentrotus purpuratus reveals more similarity to vertebrate than to invertebrate genes with EGF-like repeats. J. Mol Evol 2 9 , 314-327.

De Robertis, E.M. and Sasai, Y. (1996). A common plan for dorsoventral patterning in Bilateria. Nature 380, 37-40.

Derynck, R. (1994). TGF-P receptor-mediated signaling. Trends Biochem. Sci. 19, 548- 553.

Derynck, R. and Zhang, Y. (1996). Intracellular signaling - the MAD way to do it. Curr. B io l 6, 1226-1229.

DeSimone, D.W. and Hynes, R.O. (1994). Xenopus integrins: structural conservation and evolutionary divergence of integrin p subunits. J. Biol. Chem. 263, 5333-340.

Dickson, M.C., Martin, J.S., Cousins, F.M., Kulkarni, A.B., Karlsson, S. and Akhurst, R.J. (1995). Defective haematopoiesis and vasculogenesis in transforming growth factor-p 1 knock out mice. Development 121, 1845-1854.

Dochetry, A.J.P., O’Connell, J., Crabbe, T., Angal, S. and Murphy, G. (1992). The matrix metalloproteinases and their natural inhibitors: propects for treating degenerative tissue diseases. Trends Biotech. 10, 200-207.

Doctor, J.S., Jackson, D., Rashka, K.E., Visalli, M. and Hoffmann M. (1992). Sequence, biochemical characterization, and developmental expression of a new member of the TGF-p superfamily in Drosophila melanogaster. Dev. Biol 151, 491-505.

Dorhmann, C.E., Kessler, D.S. and Melton, D.A. (1996). Induction of axial mesoderm by zDVRl, the zebrafish ortholog of Xenopus V gl. Dev. Biol. 175, 108-117.

180 Dohrmann, C E., Hemmati-Brivanlou, A., Thomsen, G.H., Fields, A., Woolf, T.M. and Melton, D.A. (1993). Expression of activin mRNA during early development in Xenopus laevis. Dev. Biol. 157, 474-483.

Dominguez, I., Itoh, K. and Sokol, S.Y. (1995). Role of glycogen synthase kinase 3(3 as a negative regulator of dorsoventral axis formation in Xenopus embryos. Proc. Natl. Acad. Sci. USA 92, 8498-8502.

Dosch, R., Gawantha, V., Delius, H., Blumenstock, C. and Niehrs, C. (1997). Bmp-4 acts as a morphogen in dorsoventral patterning in Xenopus. Development 124, 2325- 2334.

Dumermuth, E., Sterchi, E.E., Jiang, W., Wolz, R.L., Bond, J.S., Flannery, A.V. and Beynon, R.J. (1991). The astacin family of metalloendopeptidases. J. Biol. Chem. 2 6 6 , 21381-21385.

Einspanier, R., Krause, I., Calvete, J.J., Tofperpetersen, E., Klostermeyer, H. and Karg, H. (1994). Bovine seminal plasma ASFP - localization of disulfide bridges and detection of 3 different isoelectric forms. FEBS Letts. 344, 61-64.

Evan, G.I., Lewis, G.K., Ramsay, G. and Bishop, J.M. (1985). Isolation of monoclonal antibodies specific for human c-myc proto-oncogene product. Mol. Cell. Biol. 5, 3610- 3616.

Fainsod, A., Deissler, K., Yelin, R., Marom, K., Epstein, M., Pillemer, G., Steinbeisser, H. and Blum, M. (1997). The dorsalizing and neural inducing gene follistatin is an antagonist of BMP4. Mech. Dev. 63, 39-50.

Fainsod, A., Steinbeisser, H. and De Robertis, E.M. (1994). On the function of BMP4 in patterning the marginal zone of the Xenopus embryo. EMBO J. 13, 5015-5025.

Fang, J.M., Yin, W.S., Smiley, E., Wang, S.Q. and Bonadio, J. (1996). Molecular cloning of the mouse Activin beta (E) subunit gene. Biochem. Biophys. Res. Comm. 2 2 8 , 669-674.

Fassler, R. and Meyer, M. (1995). Consequences of lack of (3i integrin gene expression in mice. Genes Dev. 9, 1896-1908.

Ferguson, E.L. and Anderson, K.V. (1992a). Decapentaplegic acts as a morphogen to organize dorsal-ventral pattern in the Drosophila embryo. Cell 71, 451-461.

181 Ferguson, E.L. and Anderson, K.V. (1992b). Localized enhancement and repression of the activity of the TGF-(3 family member, decapentaplegic, is necessary for dorsal-ventral pattern formation in the Drosophila embryo. Development 114, 583-597.

Fey, J. and Hausen, P. (1990). Appearance and distribution of laminin during development of Xenopus laevis. Differentiation 42, 144-152.

Finelli, A.L., Bossie, C.A., Xie, T. and Padgett, R.W. (1994). Mutational analysis of the Drosophila tolloid gene, a human BMPl homolog. Development 120, 861-870.

Finelli, A.L., Xie, T., Bossie, C.A., Blackman, R.K. and Padgett, R.W. (1995). The tolkin gene is a tolloidfEMPl homologue that is essential for Drosophila development. Genetics 141, 271-281.

Francois, V. and Bier, B. (1995). Xenopus chordin and Drosophila short gastrulation genes encode homologous proteins functioning in dorsal-ventral axis formation. Cell 80, 19-20.

Francois, V., Solloway, M., O’Neill, J.W., Emery, J. and Bier, E. (1994). Dorsal-ventral patterning of the Drosophila embryo depends on a putative negative growth factor encoded by the short gasrulation gene. Genes Dev. 8, 2602-2616.

Frazier, W.A. (1991). Thrombospondins. Curr Op. Cell Biol. 3, 792-799.

Fukagawa, M., Suzuki. N., Hogan, B.L.M. and Jones, C M. (1994). Embryonic expression of mouse bone morphogenetic protein-1 (BMP-1), which is related to the Drosophila dorsoventral gene tolloid and encodes a putative astacin . Dev. Biol. 163, 175-183.

Gambino, R., Salamone, M., Aliamo, M.G., Saitta, B. and Ghersi, G. (1996). Presence of different collagens and collagen mRNAs during embryogenesis and in adult tissues of the sea urchin, Paracentrotus lividus. J. Submicros. Cytol. Pathol. 28, 41-47.

Gearing, A.H., Beckett, P., Christodoulou, M., Churchill, M., Clements, J.M., Crimmin, M., Davidson, A.H., Drummond, A.H., Galloway, W.A., Gilbert, R., Gordon, J.L., Leber, T.M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S., Thomas, W., Wells, G., Wood, L.M and Woolley, K. (1995). Matrix metalloproteinases and processing of pro-TNF-alpha. J. Leukoc. Biol. 57, 774-777.

182 Gearing, A.H., Beckett, P., Christodoulou, M., Churchill, M., Clements, J.M., Davidson, A.H., Drummond, A.H., Galloway, W.A., Gilbert, R., Gordon, J.L., Leber, T.M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S., Thomas, W., Wells, G., Wood, L.M and Woolley, K. (1994). Processing of Tumor-Necrosis-Factor-alpha precursor by metalloproteinases. Nature 370, 555-557.

Gelbert, W.M. (1989). The decapentaplegic gene: a TGF-p homologue controlling pattern formation in Drosophila. Development 122 (Suppl), 65-74.

Gentry, L.E. and Nash, B.W. (1990). The pro domain of pre-pro transforming growth factor p 1 when independently expressed is a functional binding protein for the mature growth factor. Biochemistry 29, 6851-6857.

George, E.L., Georges-Labouesse, E.N., Patel-King, R.S., Rayburn, H. and Hynes, R.O. (1993). Defects in mesoderm, neural tube and vascular development in mouse embryos lacking fibronectin. Development 119, 1079-1091.

Gerhart, J., Danilchik, M. Doniach, T., Roberts, S., Rowning, B. and Stewart, R. (1989). Cortical rotation of ih t Xenopus egg: consequences for the anterioposterior pattern of embryonic dorsal development. Development Suppl. 107, 37-51.

Gerhart, J. and Keller, R. (1986). Region-specific cell activities in amphibian gastrulation. Ann. Rev. Cell Biol. , 201-229.

Ghersi, G., Salamone, M., Dolo, V., Levi, G. and Vittorelli, M.L. (1993). Differential expression and function of cadherin-like proteins in the sea urchin embryo. Mech. Dev. 41, 47-55.

Giannelli, G., Falk-Marzillier, J., Schiraldi, O., Stetker-Stevenson, W.G. and Quaranta, V. (1997). Induction of cell migration by matrix metalloprotease-2 cleavage of laminin 5. Science 277, 225-228.

Godsave, S.F. and Slack, J.M.W. (1989). Clonal analysis of mesoderm induction in Xenopus laevis. Dev. Biol. 134, 486-490.

Goodman, S., Albano, R., Wardle, F., Matthews, G., Tannahill, D. and Dale, L. (1998). BMPl-related metalloproteases promote the development of ventral mesoderm in early Xenopus embryos. Dev. Biol, (in press).

Graff, J.M., Bansal, A. and Melton, D.A. (1996). Xenopus mad proteins transduce distinct subsets of signals for the TGF-p a superfamily. Cell 85, 479-487. 183 Graff, J.M., Thies, R.S., Song, J.J., Celeste, AJ. and Melton, D.A. (1994). Studies with a Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override dorsal signals in vivo. Cell 79, 169-179.

Gray, A.M. and Mason, A.J. (1990). requirement for activin A and transforming growth factor p 1 pro-regions in homodimer assembly. Science 247, 1328-1330.

Green, J.B.A. and Smith, J.C. (1990). Graded changes in dose of a. X enopus Activin A homologue elicits stepwise transitions in embryonic cell fate. Nature 347, 391-394.

Griffith, D.L., Keck, P.C., Sampath, S.K., Rueger, D C. and Carlson, W.D. (1996). Three-dimensional structure of recombinant human osteogenic protein 1: Structural paradigm for the transforming growth factor p superfamily. Proc. Natl Acad. Sci. USA 93, 878-883.

Gumbiner, B.M. (1996). Cell adhesion: the molecular basis of tissue architecture and morphogenesis. Cell 84, 345-357.

Hammerschmidt, M., Pelegri, P., Mullins, M.C., Kane, D.A., van Eeden, F.J.M., Granato, M., Brand, M., Furutani-Seiki, M., Haffter, P., Heisenberg, C.P., Jiang, Y.J., Kelsh, R.N., Odenthal, J., Warga, R.M., Nusslein-Volhard, C. (1996a). dino and mercedes, two genes regulating dorsal development in the zebrafish embryo. Development 123, 95-102.

Hammerschmidt, M., Serbedzija, G.N. and McMahon, A.P. (1996b). Genetic analysis of dorsoventral pattern formation in the zebrafish: requirement of a BMP-like ventralizing activity and its dorsal repressor. Genes Dev. 10, 2452-2461.

Hansen, C.S., Marion, C D., Steele, K., George, S. and Smith, WC. (1997). Direct neural induction and selective inhibition of mesoderm and epidermis inducers by Xnr3. Development 124, 483-492.

Harland, R.M. (1991). In situ hybridization: an improved whole-mount method for Xenopus embryos. Meth. Cell Biol. 36, 685-695.

Hawley, S.H.B., Wunnenberg-Stapleton, K., Hashimoto, C., Laurent, M.N., Watabe, T., Blumberg, B.W. and Cho, K.W.Y. (1995). Disruption of BMP signals in embryonic Xenopus ectoderm leads to direct neural induction. Genes Dev. 9, 2923-2935.

184 Hayashi, H., Abdollah, S., Qiu, Y.B., Cai, J.X., Xu, Y.Y., Grinnell, B.W., Richardson, M.A., Topper, J.N., Gimbrone, M.A., Wrana, J.L. and Falb, D. (1997). The MAD- related protein Smad7 associates with the TGF beta receptor and functions as an antagonist of TGF-beta signaling. Ce//89, 1165-1173.

Hazama, M., Aono, A., Ueno, N. and Fujisawa, Y. (1995). Efficient expression of a heterodimer of bone morphogenetic protein subunits using a baculovirus expression system. Biochem. Biophys. Res. Comm. 209, 859-866.

He, X., Saint-Jeannet, J-P., Woodgett, J R., Varmus, H E., and Dawid, I B. (1995). Glycogen synthase kinase-3 and dorsoventral patternig in X enopus embryos. Nature 3 7 4 , 617-622.

Heasman, J., Crawford, A., Goldstone, K., Garner-Hamrick, P., Gumbiner, B., McCrea, P., Kintner, C., Yoshida Noro, C. and Wylie, C. (1994a). Overexpression of cadherins and underexpression of p-catenin inhibit dorsal mesoderm induction in early X enopus embryos. Cell 79, 791-803.

Heasman, J., Ginsberg, D., Geiger, B., Goldstone, K., Pratt, T, Yoshida-Noro, C. and Wylie, C. (1994b). A functional test for maternally inherited cadherin in Xenopus shows its importance in cell adhesion at the blastula stage. Development 120, 49-57.

Helde, K.A. and Grunwald, D.J. (1993). The DVR-1 (Vgl) transcript of zebrafish is maternally supplied and distributed throughout the embryo. Dev. Biol. 159, 418-426.

Heldin, C H., Miyazono, K. and ten Dijke, P. (1997). TGF-p signalling from cell membrane to nucleus through Smad proteins. Nature 390, 465-471.

Hemmati-Brivanlou, A., Frank, D., Bolce, M, Brown, B., Sive, H. and Harland, R. (1990). Localization of specific mRNAs in Xenopus embryos by whole-mount in situ hybridization. Development 110, 325-330.

Hemmati-Brivanlou, A., Kelly, O.G. and Melton, D.A. (1994). Follistatin, an antagonist of activin, is expressed in the Spemann organizer and displays direct neuralizing activity. Cell 77, 283-295.

Hemmati-Brivanlou, A. and Melton, D.A. (1992). A truncated activin receptor inhibits mesoderm induction and formation of axial structures in Xenopus em bryos. Nature 3 5 9 , 609-614.

185 Hemmati-Brivanlou, A. and Melton, D. (1997). Vertebrate embryonic cells will become nerve cells unless told otherwise. Cell 88, 13-17.

Hemmati-Brivanlou, A. and Thomsen, G.H. (1995). Ventral mesodermal patterning in Xenopus embryos: expression patterns and activities of BMP2 and BMP4. Dev. Genet. 17, 78-89.

Hemmati-Brivanlou, A., de la Torre, J.R., Holt, C. and Harland, R.M. (1991). Cephalic expression and molecular characterization of Xenopus En-2. Development 111, 715-724.

Henry, G.L., Brivanlou, I.H., Daniel, S.K., Hemmati-Brivanlou, A. and Melton, D.A. (1996). TGF-p signals and a prepattem in Xenopus laevis endodermal development. Development 122, 1007-1015.

Hishida, R., Ishihara, T., Kondo, K. and Katsura, I. (1996). hch-1, a gene required for normal hatching and normal migration of a neuroblast in C.elegans, encodes a protein related to Tolloid and BMPl. EMBO J. 15, 4111-4122.

Hogan, B.L.M. (1995). Upside-down ideas vindicated. Nature 376, 210-211.

Hogan, B.L.M. (1996a). Bone morphogenetic proteins: multifunctional regulators of vertebrate development. Genes Dev. 10, 1580-1594.

Hogan, B.L.M. (1996b). Bone morphogenetic proteins in development. Curr. Op. Genet. Dev. 6, 432-438

Hojima, Y., Vanderrest, M. and Prockop, D.J. (1985). Type-I procollagen carboxyl- terminal proteinase from chick-embryo tendons - purification and characterization. J. Biol. Chem. 260, 5996-6003

Holley, S.A. and Ferguson, E.L. (1997). Fish are like flies are like frogs: conservation of dorsal-ventral patterning molecules. Bioessays 19, 281-283.

Holley, S.A., Jackson, P.D., Sasai, Y., Lu, B., De Robertis, E.M., Hoffman, F.M. and Ferguson, E.L. (1995). A conserved system for dorso-ventral patterning in insects and vertebrate involving sog and chordin. Nature 376, 249-253.

Holley, S.A., Neul, J.L., Attisano, L., Wrana, J.L., Sasai, Y, O’Connor, M B., De Robertis, E.M. and Ferguson, E.L. (1996). The Xenopus dorsalizing factor Noggin ventralizes Drosophila embryos by preventing DPP from activating its receptor. Cell 86, 607-617. 186 Hoodless, P.A., Haerry, T., Abdollah, S., Stapleton, M., O’Connor, M.B., Attisano, L. and Wrana, J.L. (1996). MADRl, a MAD-related protein that functions in BMP2 signaling pathways. Cell 85, 489-500.

Hooper, N.M., (1994). Families of zinc metalloproteases. FEBS Lett. 354, 1-6.

Hoppler, S., Brown, J.D. and Moon, R.T. (1996). Expression of a dominant-negative Wnt blocks induction of MyoD in Xenopus embryos. Genes Dev. 10, 2805-2817.

Horstadius, S. (1973). Experimental Embryology of Echinoderms. Oxford, Clarendon Press.

Rotten, G., Neidhardt, H., Schneider, C. and Pohl, J. (1995). Cloning of a new member of the TGF-p family - a putative new activin p (C) chain. Biochem. Biophys. Res. Comm. 206, 608-613.

Howard J.E., Hirst, E.M.A. and Smith, J.C. (1992). Are pi integrins involved in Xenopus gastrulation? Mech. Dev. 38, 109-120.

Howard J.E. and Smith, J.C. (1993). Analysis of gastrulation: different types of gastrulation movement are induced by different mesoderm-inducing factors in Xenopus laevis. Mech. Dev. 43, 37-48.

Hwang, S-P. L., Partin, J.S. and Lennarz, W.J. (1994). Characterisation of a homolog of human bone morphogenetic protein 1 in the embryo of the sea urchin, Strongylocentrotus purpuratus. Development 120, 559-568.

Hynes, R.O. (1992). Integrins: versatility, modulation and signalling in cell adhesion. Cell 62, 11-25.

Hynes, R.O. (1996). Targeted mutations in cell adhesion genes: what have we learned from them? Dev. Biol. 180, 402-412. lannaccone, P.M., Zhou, X, Khokha, M, Boucher, Drosophila and Kuehn, M R. (1992). Insertional mutation of a gene involved in growth regulation of the early mouse embryo. Dev. Dynam. 194, 198-208.

Imamura, T., Takase, M., Nishihara, A., Oeda, E., Hanai, J., Kawabata, M. and Miyazono, K. (1997). Smad 6 inhibits signalling by the TGF-p superfamily. Nature 3 8 9 , 622-626.

187 Irish, V F. and Gelbert, W.M. (1987). The decapentaplegic gene is required for dorsal- ventral patterning of the Drosophila embryo. Genes Dev. 1, 868-879.

Jeffery, W.R. (1992). Axis determination in sea urchin embryos: from confusion to evolution. Trends Genet 8, 223-225.

Jiang, W. and Bond, J.S. (1992). Families of metalloendopeptidases and thier relationships. FEBS Lett. 312, 110-114

Jones, C M, Armes, N. and Smith, J.C. (1996a). Signalling by TGF-P family mebers: short range effects of Xnr2 and BMP4 contrast with long-range effects of activin. Curr. Biol. 6, 1468-1475.

Jones, C M., Dale, L., Hogan, B.L.M., Wright, C.V.E. and Smith, J.C. (1996b). Bone morphogenetic protein-4 (BMP-4) acts during gastrula stages to cause ventralization of Xenopus embryos. Development 122, 1545-1554.

Jones, C M., Kuehn, M.R., Hogan, B.L., Smith, J.C .and Wright, C.V.E. (1995). Nodal-related signals induce axial mesoderm and dorsalize mesoderm during gastrulation. Development 121, 3651-3662.

Jones, C M., Lyons, K.M., Lapan, P.M., Wright, C.V.E. and Hogan, B.L.M. (1992a). DVR-4 (Bone Morphogenetic Protein-4) as a posterior-ventralizing factor in Xenopus mesoderm induction. Development 115, 639-647.

Jones, C M., Simon, C D., Guenet, J.L. and Hogan, B.L.M. (1992b). Isolation of Vgr-2, a novel member of the transforming growth factor p-related family. Mol. Endocrinol. 6, 1961-1968.

Jones, E.A. and Woodland, H.R. (1987). The development of animal cap cells in Xenopus: a measure of the start of animal cap competence to form mesoderm. Development 101, 557-563.

Jongeneel, C.V., Bouvier, J., Bairoch, A. (1989). A unique signature identifies a family of zinc-dependent metallopeptidases. FEBS Lett. 242, 211-214

Joos, T.O., Whittaker, C.A., Meng, F.Y., DeSimone, D.W., Gnau, V. and Hausen, P.

(1995). Integrin as during early development of Xenopus laevis. Mech. Dev. 5 0 , 187-

199.

188 Joseph, E.M. and Melton, D.A. (1997). Xnr4: a Xenopus nodal-related gene expressed in the Spemann organizer. Dev. Biol. 184, 367-372.

Kadler, K.B., Holmes, D.F., Trotter, J.A. and Chapman, J.A. (1996). Collagen fibril formation. Biochem. J. 316, 1-11

Kao, K. R. and Elinson, R. P. (1988). The entire mesodermal mantle behaves as Spemann's organizer in dorso-anterior enhanced Xenopus laevis embryos. Dev. Biol. 127, 64-77.

Katow, H. (1990). A new technique for introducing anti-fibronectin antibodies and fibronectin-related synthetic peptides into the blastulae of the sea urchin, Clypeaster japonicus. Develop. Growth Differ. 32, 33-39.

Katow, H., Yazawa, Y. and Sofuku, S. (1990). A fibronectin-related synthetic peptide, Pro-Ala-Ser-Ser, inhibits fibronectin binding to the cell surface, fibronectin-promoted cell migration in vitro, and cell migration in vivo. Exp. Cell Res. 190, 17-24.

Kessler, D.S. and Melton, D.A. (1995). Induction of dorsal mesoderm by soluble, mature Vgl protein. Development 121, 2155-2164.

Kessler, E., Adar, R., Goldberg, B and Niece, R. (1986). Partial purification and characterization of a procollagen C-proteinase from culture medium of mouse fibroblasts. Collagen Rel. Res. 6, 249-266.

Kessler, E., Takahara, K., Biniaminov, L., Brusel, M and Greenspan, D.S. (1996). Bone morphogenetic protein-1: The type 1 procollagen C-proteinase. Science 271, 360-362.

Kim, J., Johnson, K., Chen, H.J., Carroll, S. and Laughon, A. (1997). Drosophila MAD binds to DNA and directly mediates activation of vestigial by decapentaplegic. Nature 3 8 8 , 304-308.

Kimelman, D., Christian, J.L. and Moon, R.T. (1992). Synergistic principles of development - overlapping patterning systems in Xenopus mesoderm induction. Development 116, 1-9.

Kingsley, D M, Bland, A.E., Grubber, J.M., Marker, P.C., Russell, L.B., Copeland, N.G. and Jenkins, N.A. (1992). The mouse short-ear skeletal morphogenesis locus is associated with defects in a bone morphogenetic member of the TGF-P superfamily. Cell 11, 399-410.

189 Kingsley, D.M, (1994). The TGF-p superfamily: new members, new receptors, and new genetic tests of function in different organisms. Genes Dev. 8, 133-146.

Kishimoto, Y., Lee, K., Zon, L., Hammerschmidt, M. and Schulte-Merker, S. (1997). The molecular nature of zebrafish swirl: BMP2 function is essential during early dorsoventral patterning, (in press).

Kreig, P.A. and Melton, D.A. (1984). Functional messenger RNAs are produced by SP6 in vitro transcription of cloned cDNAs. Nuc. Ac. Res. 12, 7057-7070.

Kreig, P.A. and Melton, D.A. (1987). In vitro RNA synthesis with SP6 RNA polymerase. Meth. Enzymol. 155, 397-415.

Kretzschmar, M., Liu, F., Hata, A., Doody, J. and Massagué, J. (1997).The TGF-beta family mediator Smadl is phosphorylated directly and activated functionally by the BMP receptor kinase. Genes Dev. 8, 984-995.

Kühl, M. and Wedlich, D. (1996).Xenopus Cadherins: Sorting out types and functions in embryogenesis. Dev. Dynam. 207, 121-134.

Kulkarni, A.B., Huh, C.G., Becker, D., Geiser, A., Lyght, M., Flanders. K.C., Roberts, A.B., Sporn, M.B., Ward, J.M. And Karlsson, S. (1993). Transforming growth factor- beta 1 null mutation in mice causes excessive inflammatory response and early death. Proc. Natl. Acad. Sci. USA 90, 770-774

Lallier, T.E., Whittaker, C.A. and DeSimone, D.W. (1996). Integrin expression is required for early nervous system development in Xenopus laevis. Development 1 2 2 , 2539-2554.

Lamb, T.M., Knecht, A.K., Smith, W.C., Stachel, S.E., Economides, A.N., Stahl, N. Yancopolous, G.D., and Harland, R.M (1993). Neural induction by the secreted polypeptide noggin. Science 262, 713-718.

Larabell, C.A., Torres, M., Rowning, B.A., Yost, C, Miller, J.R., Wu, M., Kimelman, D., Moon, R.T. (1997). Establishment of the dorso-ventral axis in Xenopus embryos presaged by early asymmetries in P-catenin that are modulated by the . J. Cell Biol. 136, 1123-1136.

190 Larue, L., Ohsugi, M., Hirchenhain, J. and Kemler, R. (1994). E-cadherin null mutant embryos fail to form a trophectoderm epithelium. Proc Natl Acad. Sci USA 91, 8263- 8267.

Lau, A.L., Nishimori, K. and Matzuk, M.M. (1996). Stmctural-analysis of the mouse Activin beta-C gene. Biochim. Biophys. Acta 1307, 145-148.

Lecuit, T., Brook, W.J., Ng, M., Calleja, M., Sun, H. and Cohen, S.M. (1996). Two distinct mechanisms for long-range patterning by Decapentaplegic in the Drosophila wing. N ature 381, 387-393.

Lee, C-H. and Gumbiner, B.M. (1995). Disruption of gastrulation movements in Xenopus by a dominant negative mutant for C-cadherin. Dev. B iol 171, 363-373.

Lee, G., Hynes, R.O. and Kirshner, M. (1984). Temporal and spatial regulation of fibronectin in early Xenopus development. Cell 36, 729-740.

Lee, J.J. and Costlow, N.A. (1987). A molecular titration assay to measure transcript prevalence levels. Meth. Enzymol 152, 633-648.

Lee, S.J. (1990). Identification of a novel member (GDF-1) of the transforming growth factor-beta superfamily. M ol Endocrinol. 4, 1034-1040.

Lemaire, P., Garrett, N. and Gurdon, J.B. (1995). Expression cloning of Siamois, a Xenopus homeobox gene expressed in dorsal-vegetal cells of blastulae and able to induce a complete secondary axis. Cell 81, 85-94.

LePage, T., Ghiglione, C. and Gache, C. (1992). Spatial and temporal expression pattern during sea urchin embryogenesis of a gene encoding for a protease homologous to the human protein BMPl and to the product of the Drosophila dorsal-ventral patterning gene tolloid. Development 114, 147-164.

Letsou, A., Arora, K, Wrana, J.L., Simin, K., Twombly, V., Jamai, J., Staehling- Hampton, K, Hoffman, P.M., Gelbert, W.M., Massagué, J and O’Connor, M.B. (1995). Drosophila dpp signaling is mediated by the punt gene product: a dual ligand-binding type II receptor of the TFG(J receptor family. Cell 80, 899-908.

Letterio, J.J., Geiser, A.G., Kulkarni, A.B., Roche, N.S., Sporn, M.B. and Roberts, A.B. (1994). Matemal rescue of transforming growth factor-p 1 null mice. Science 2 6 4 , 1936-1938.

191 Leyns, L., Bouwmeesyer, T., Kim, S-H, Piccolo, S. and De Robertis E.M. (1997). Frzb- 1 is a secreted anatgonist of wnt signalling expressed inn the Spemann organizer. Cell 88, 747-756.

Leytus, S.P., Kurachi, K., Sakariassen, K.S. and Davie, E.W. (1986). Nucleotide sequence of the cDNA coding for human complement Clr. Biochemistry 25, 4855-4863.

Li, S-W., Sieron, A.L., Fertala, A., Hojima, Y., Arnold, W.V. and Prockop, D.J. (1996). The C-proteinase that processes procollagens to fibrillar collagens is identical to the protein previously identified as bone morphogenetic protein-1. Proc. Natl Acad. Sci. USA 93, 5127-5130.

Lin, C.Q. and Bissell, M.J. (1993). Multi-faceted regulation of cell differentiation by extracellular matrix. FASEB J. 7, 737-743.

Lin, H.Y. and Lodish, H.F. (1993). Receptors for the TGF-p superfamily: multiple polypeptides and serine/threonine kinases. Trends Cell B iol 3, 14-19.

Lin, J.J., Maeda, R., Ong, R.C., Kim, J., Lee, L.M., Kung, H.F. and Maeno, M. (1997). XBMP-IB (Xtld), a Xenopus homolog of the dorso-ventral polarity gene in Drosophila, modifies tissue phenotypes of ventral explants. Develop. Growth & Dijf. 3 9 , 43-51.

Ling, N.; Ying, S.Y., Ueno, N., Shimasaki, S., Esch, F., Hotta, M.and Guillemin, R. (1986). Pituitary FSH is released by a heterodimer of the P-subunits from the two forms of inhibin. Nature 321, 779-82.

Liu, .F, Hata, A., Baker, J.C., Doody, J., Carcamo, J., Harland, R.M. and Massagué, J. (1996). A human mad protein acting as a BMP-regulated transcriptional activator. Nature 381, 620-623.

Liu, Q.R., Hattar, S., Endo, S., MacPhee, K., Zhang, H., Cleary, L.J., Byrne, J.H. and Eskin, A. (1997). A developmental gene (Tolloid/BMP-1) is regulated in Aplysia neurons by treatments that induce long-term sensitization. J. N euroscl 17, 755-764.

Livingston, B.T. and Wilt, E.H. (1990). Determination of cell fate in sea urchin embryos. Bioessays 12, 115-119.

Logan, C.Y. and McClay, D R. (1997), The allocation of early blastomeres to the ectoderm and endoderm is variable in the sea urchin embryo. Development 124, 2213-2223.

192 Lyons, K.M., Jones, C M. and Hogan, B.L.M. (1991). The DVR gene family in embryonic development. Trends Genet. 7, 408-412.

Lyons, K.M., Pelton, R.W., Hogan, B.L.M. (1989). Patterns of expression of murine Vgr-1 and BMP-2a RNA suggest that transforming growth factor-beta-like genes coordinately regulate aspects of embryonic development. Genes Dev. 3, 1657-68.

Macias-Silva, M., Abdollah, S., Hoodless, P.A., Pirone, R., Attisano, L. and Wrana, J.L. (1996). MADR2 is a substrate of the TGFp receptor and its phosphorylation is required for nuclear accumulation and signaling. Cell 87, 1215-1224.

MacNicol, A.M., Muslin, A.J. and Williams, L.T. (1993). Raf-1 kinase is essential for early Xenopus development and mediates the induction of mesoderm by FGF.Ce// 73, 571-83.

Maeno, M., Xue, Y., Wood, T.I., Ong, R.C. and King, H.F. (1993). Cloning and expression of cDNA encoding Xenopus laevis bone morphogenetic protein-1 during embryonic development. Gene 134, 257-261.

Marqués, G., Musacchio, M., Shimell, M.J., Wunnenburg-Stapleton, K., Cho, K.W.Y. and O'Connor, M.B. (1997). The Dpp activity gradient in the early Drosophila embryo is established through the opposing actions of the Sog and Tld proteins. Cell 91, 417-425.

Marrs, J.A. and Nelson, W.J. (1996). Cadherin cell adhesion molecules in differentiation and embryogenesis. Int. Rev. Cytol. 165, 159-205.

Marsden, M. and Burke, R.D. (1997). Cloning and characterization of novel p integrin subunits from a sea urchin. Dev. Biol. 181, 234-245.

Mason, A.J., Hayflick, J.S., Ling, N., Esch, F., Ueno, N., Ying, S.Y., Guillemin, R. Niall, H. and Seeburg, P.H. (1985). Complementary DNA sequences of ovarian follicular fluid inhibin show precursor structure and homology with transforming growth factor-p. Nature 318, 659-63.

Massagué, J. (1990). The transforming growth factor-p family. Annu. Rev. Cell Biol. 6, 597-641.

Massagué, J. (1992). Receptors for the TGF-p family. Cell 69, 1067-1070.

Massagué, J. (1996). TGF-p signalling: receptors, transducers and Mad proteins. Cell 85, 947-950. 193 Massagué, J., Hata, A. and Liu, F. (1997). TGF(3 signalling through the Smad pathway. Trends Cell. Biol. 7, 187-192.

Matthews, G. and Colman, A. (1991). A highly efficient, cell-free translation/translocation system prepared from Xenopus eggs. Nuc. Ac. Res. 19, 6405-6412.

Matzuk, M.M., Kumar, T.R., Vassalli, A., Bickenbach, J.R., Roop, D R., Jaenisch, R. and Bradley, A. (1995). Functional analysis of activins during mammalian development. Nature 374, 354-356

Maurer, P. and Engel, J. (1996). In 'The Laminins' (eds P. Ekblom and R. Timpl). pp 27- 49. Harwood, Amsterdam.

McDonald, N.Q. and Hendrickson, W.A. (1993). A structural superfamily of growth factors containing a cysteine knot motif. Cell 73,421-424.

McDonald, N.Q. and Kwong, P.D. (1993). Does Noggin head a new class of Kunitz domain? Trends Biochem. Sci. 18, 208-209.

McDowell, N., Zorn, A.M., Crease, D.J. and Gurdon, J.B. (1997). Activin has long range signalling activity and can form a concentration gradient by diffusion. Curr. Biol. 1, 671-681.

McGeehan, G.M., Becherer, J.D., Bast, R.C., Boyer, C.M., Champion, B., Connolly, K.M., Conway, J.G., Furdon, P., Karp, S., Kidao, S., McElroy, A.B., Nichols, J., Pryzwansky, K.M., Schoenen, F., Sekut, L., Truesdale, A., Verghese, M., Warner, J.and Ways, J.P. (1994). Regulation of Tumor-Necrosis-Factor-a processing by a metalloproteinase Nature 370, 558-561

McPherron, A C. and Lee, S.J. (1993). GDF-3 and GDF-9: Two new members of the transforming growth factor (3 superfamily containing a novel pattern of cysteines. J. Biol. Chem. 268, 3444-3449.

Meng, F.Y., Whittaker, C.A., Ransom, D.G. and DeSimone, D.W. (1997). Cloning and characterization of cDNAs encoding the integrin 0 2 and subunit from Xenopus laevis.

Mech. Dev. 67, 141-155.

Miller, J.R. and Moon, R.T. (1996). Signal transduction through p-catenin and specification of cell fate during embryogenesis. Genes Dev. 10, 2527-2539.

194 Miya, T., Morita, K., Ueno, N. and Satoh, N. (1996). An ascidian homolog of vertebrate BMPs-5-8 is expressed in the midline of the anterior neuroectoderm and in the midline of the ventral epidermis of the embryo. Mech. Dev 57, 181-190.

Mohun, T.J., Brennan, S., Dathan, N., Fairman, S. and Gurdon, J.B. (1984). Cell type specific activation of actin genes in the early amphibian embryo. Nature 311, 716-721

Molenaar, M., van der wetering, M, Oosterwegel, M., Peterson-Maduro, J., Godsave, S., Korinek, V, Roose, J., Destree, O. and Clevers, H. (1996). XTcfB transcription factor mediates (J-catenin-induced axis formation in Xenopus embryos. Cell 86, 391-399.

Moon, R.T., Campbell, R.M., Christian, J.L., McGrew, L.L., Shih. J. and Fraser, S. (1993). Xwnt-5A: a matemal Wnt that affects morphogenetic movements after overexpression in embryos of Xenopus laevis. Development 119, 97-111.

Moos, M., Wang, S. and Krinks, M. (1995). Anti-dorsalizing morphogenetic protein is a novel TGF-p homolog expressed in the Spemann organizer. Development 121, 4293- 4301.

Nakamura, T., Sugino, K., Titani, K. and Sugino, H. (1991). Follistatin, an activin- binding protein, associates with heparin sulphate chains of proteoglycans on follicular granulosa cells. J. Biol. Chem. 266, 19432-19437.

Nakamura, T., Takio, K., Eto, Y., Titani, K and Sugino, H. (1990). Activin-binding protein from rat ovary is follistatin. Science 247, 836-838.

Nakao, A., Afrakhte, M., Moren, A., Nakayama, T., Christian, J.L., Heuchel, R., Itoh, S, Kawabata, M., Heldin, N-E, Heldin, C-H and ten Dijke, P. (1997). Identification of Smad 7, a TGFp-inducible antagonist of TGF-p signalling. Nature 389, 631-635.

Nellen, D., Affolter, M., Basler, K. (1994). Receptor serine/threonine kinases implicated in the control of Drosophila body pattern by decapentaplegic. Cell 78, 225-37.

Nellen, D., Burke, R., Struhl, G. and Basler, K. (1996). Direct and long-range action of a DPP morphogen gradient. Cell 85, 357-368.

Nguyen, T., Joumana, J., Shimell, M.J., Arora, K. and O'Connor, M.B. (1994). Characterization of tolloid-related-1: a BMPl-like product that is required during larval and pupal stages of Drosophila development. Dev. Biol. 166, 569-586.

195 Nieuwkoop, P.D. and Faber, J. (1967). Normal Table of Xenopus laevis (Daudin). Amsterdam, North Holland Publishing Co.

Nikaido, M., Tada, M., Saji, T. and Ueno, N. (1997). Conservation of BMP signaling in zebrafish mesoderm patterning. Mech. Dev. 61, 75-88

Oda, S., Nishimatsu, S., Murakami, K. and Ueno, N. (1995). Molecular cloning and functional analysis of a new activin p subunit: a dorsal mesoderm-inducing activity in Xenopus. Biochem. Biophys. Res. Commun. 210, 581-8.

Okada, A., Tomasetto, C., Lutz, Y., Bellocq, J.P., Rio, M.C. and Basset, P. (1997). Expression of matrix metalloproteinases during rat skin wound healing: evidence that membrane type-1 matrix metalloprotease is a stromal activator of pro-gelatinase A. J. Cell Biol. 137, 67-77

Otte, A.P., Roy, D., Siemerink, M., Koster, C.H., Hochstenbach, P., Timmermans, A. and Durston, A.J. (1990). Characterization of a matemal type VI collagen in Xenopus embryos suggests a role for collagen in gastrulation. J. Cell. Biol. Ill, 271-278.

Overall, C.M. (1991). Recent advances in matrix metalloproteinas research. TIGG 3, 384- 399.

Ozkaynak, B., Rueger, D C., Drier, B.A., Corbett, C., Ridge, R.J., Sampath, T.K., Oppermann, H. (1990). OP-1 cDNA encodes an osteogenic protein in the TGF-p family. EMBO J. 9, 2085-93.

Ozkaynak, B., Schnegelsberg, P.N.J., Jin, D.F., Clifford, G.M., Warren, F.D., Drier, B.A. and Oppermann, H. (1992). Osteogenic Protein-2 - A new member of the transforming growth factor p superfamily expressed early in embryogenesis. J. Biol. Chem. 267, 25220-25227.

Padgett, R.W., St Johnson, R.D. and Gelbert, W.M. (1987). The decapentaplegic gene complex of Drosophila encodes a protein homologous to the transforming growth factor-p gene family. Nature 325, 81-84.

Padgett, R.W., Wozney, J.M., and Gelbart, W.M. (1993). Human BMP sequences can confer normal dorsal-ventral patterning in the Drosophila embryo. Proc. Natl. Acad. Sci. USA 90, 2905-2909.

196 Panchenko, M.V., Stetler-Stevenson, W.G., Trubetskoy, O.V., Gacheru, S.N. and Kagan, H.M. (1996). Metalloproteinase activity secreted by fibrogenic cells in the processing of prolysyl oxidase. J. B iol Chem. 271, 7113-7119.

Panganiban, F.E., Reuter, R., Scott, M.P. and Hoffman, F.M. (1990). A Drosophila growth factor homolog, decapentaplegic, regulates homeotic gene expression within and across germ layers during midgut morphogenesis. Development 110, 1041-1050.

Paralkar, V.M., Nandedkar, A.K.N., Pointer, R.H., Kleinman, H.K. and Reddi, A.H. (1990). Interaction of osteogenin, a heparin binding bone morphogenetic protein, with type IV collagen. J. B iol Chem. 265, 17281-17284.

Pelham, H.R.B. and Jackson, R.J. (1976). An efficient mRNA-dependent translation system from reticulocyte lysates. Eur. J. Biocem. 67, 247-256.

Piccolo, S. Agius, E., Lu, B., Goodman, S., Dale, L. and De Robertis, E.M. (1997). Cleavage of Chordin by the Xolloid metalloprotease suggests a role for proteolytic processing in the regulation of Spemann organizer activity. Cell 91, 407-416.

Piccolo, S., Sasai, Y., Lu, B. and De Robertis, E.M., (1996). Dorsoventral patterning in Xenopus: inhibition of ventral signals by direct binding of chordin to BMP-4. Cell 86, 589-598.

Raftery, L.A., Twombly, V., Wharton, K., Gelbart, W.M. (1995). Genetic screens to identify elements of the decapentaplegic signaling pathway in Drosophila. Genetics 1 3 9 , 241-54.

Ramos, J.W. and DeSimone, D.W. (1996). Xenopus embryonic cell adhesion to fibronectin: position-specific activation of RGD/Synergy site-dependent migratory behaviour at gastrulation. J. Cell. B iol 134, 227-240.

Ramos, J.W., Whittaker, C.A. and DeSimone, D.W. (1996). Integrin-dependent adhesive activity is spatially controlled by inductive signals at gastrulation. Development 122, 2875- 2883.

Ransom, D.G., Hens, M.D. and DeSimone, D.W. (1993). Integrin expression in early amphibian embryos: cDNA cloning and characterization of Xenopus pi, P2, p3, and p6 subunits. Dev. Biol. 160, 265-275.

Ransick, A. and Davidson, E.H. (1993). A complete second gut induced by transplantaed micromeres in the sea urchin embryo. Science 259, 1134-1138. 197 Rawlings, N.D. and Barrett, A.J. (1993). Evolutionary families of peptidases. Biochem J. 290, 205-218.

Rebagliati, M R. and Dawid, I.E. (1993). Expression of activin transcripts in follicle cells and oocytes of Xenopus laevis. Dev. Biol. 159, 574-580.

Rees, D.J.G., Jones, M., Handford, P.A, Walter, S.J., Esnouf, M.P. Smith, K.J. and Brownlee G.G. (1988). The role of p hydroxyaspartate and adjacent carboxylate residues in the first EGF domain of human factir Ix. EMBO J. 7, 2053-2061.

Reynolds, S.D., Angerer, L.M., Palis, J., Nasir, A. and Angerer, R.C. (1992). Early mRNAs, spatially restricted along the animal-vegetal axis of sea urchin embryos, include one encoding a protein related to tolloid and BMP-1. Development 114, 769-786.

Richter, K., Grunz, H. and Dawid, I B. (1988). Gene expression in the embryonic nervous system of Xenopus laevis. Proc. Natl. Acad. Sci. USA 85, 8086-8090.

Riou, J-F, Darribere, T., Shi, D.L., Richoux, V. and Boucaut, J-C. (1987). Synthesis of laminin-related polypeptides in oocytes, eggs and early embryos of the amphibian Pleurodeles waltii. Roux Arch. Dev. Biol. 196, 328-332.

Rissi, M., Wittbrodt, J., Delot, E., Naegeli, M., and Rosa, F.M. (1995). Zebrafish Radar: a new member of the TGF-p superfamily defines dorsal regions of the neural plate and the embryonic retina. Mech. Dev. 49, 223-234.

Roberts, A.B. and Sporn, M.B. (1992). In "Peptide growth factors and their inhibitors.' Eds Sporn, M.B. and Roberts, A.B. pp419-472. Springer-Verlag. Berlin.

Rosenzweig, B.L., Imamura, T., Okadome, T., Cox, G.N., Yamashita, H., ten-Dijke, P., Heldin, C.H. and Miyazono, K. (1995). Cloning and characterization of a human type II receptor for bone morphogenetic proteins. Proc. Natl. Acad. Sci. USA 92, 7632-7636.

Ruberte, E., Marty, T., Nellen, D., Affolter, M and Basler, K. (1995). An absolute requirement for both the type II and type I receptors, punt and thick veins, for dpp signaling in vivo. Cell 80, 889-897.

Rubin, K., Gullberg, D., Tomasini-Johansson, B., Reed, R.K., Ryden, C. and Borg, T.K. (1996). In "Extracellular Matrix' Vol. 2. (ed.W.D. Comper). pp 262-309. Harwood, Amsterdam.

198 Ruiz i Altaba, A. and Melton, D.A. (1989). Bimodal and graded expression of the Xenopus homeobox gene Xhox3 during embryonic development. Development 106, 173- 183.

Rupp, R.A.W., Snider, L. and Weintraub, H. (1994). Xenopus embryos regulate the nuclear localization of XMyoD. Genes Dev. 8, 1311-1323.

Ruppert, R., Hoffman, E. and Sebald, W. (1996). Human bone morphogenetic protein 2 contains a heparin-binding site which modifies its biological activity. Eur. J. Biochem 237, 296-302.

Ruoslahti, E. and Yamaguchi, Y. (1991). Proteoglycans as modulators of growth factor activities. Cell 64, 867-869.

Ruoslahti, E., Yamaguchi, Y., Hildebrand, A and Border, W.A. (1992). Extracellular matrix/growth factor interactions. J. Histochem. Cytochem. 44, 875-889.

Sambrook, J., Fritsch, E.H. and Maniatis, T. (1989). Molecular cloning: a laboratory manual. Second edition. Cold Spring Harbor Laboratory, Cold Spring Harbor.

Sampath, T.K., Rashka, K.E., Doctor, J.S., Tucker, R.F. and Hoffman, F.M. (1993). Drosophila transforming growth factor p superfamily proteins induce endochondral bone formation in mammals. Proc. Natl. Acad. Sci. USA 90, 6004-6008.

Sanger, F., Niklen, S. and Coulson, A.R. (1977). DNA sequencing with chain terminating inhibitors. Proc. Nat. Acad. Sci. USA 74, 5463-5467.

Sarras, M.P. Jr. (1996). BMP-1 and the astacin family of metalloproteinases: a potential link between the extracellular matrix, growth factors and pattern formation. Bioessays 18, 439-442.

Sasai, Y., Lu, B., Piccolo, S. and De Robertis, E.M. (1996). Endoderm induction by the organizer-secreted factors Chordin and Noggin in Xenopus animal caps. EMBO J. 15, 4547-4555.

Sasai, Y., Lu, B., Steinbeisser, H. and De Robertis, E.M. (1995). Regulation of neural induction by the Chd and BMP4 antagonistic patterning signals in Xenopus. Nature 3 7 6 , 333-336.

199 Sasai, Y., Lu, B., Steinbeisser, H., Geissert, D., Gont, L.K. And De Robertis, E.M. (1994). Xenopus Chordin - a novel dorsalizing factor activated by organizer-specific homeobox genes. Cell 79, 779-790.

Sastry, S.K. and Horwitz, A.F. (1993). Integrin cytoplasmic domains: mediators of cytoskeletal linkages and extra- and intracellular initiated transmembrane signalling. Curr. Op. Cell Biol 5, 819-831.

Sastry, S.K. and Horwitz, A.F. (1996). Adhesion-growth factor interactions during differentiation: an integrated biological response. Dev. Biol. 180, 455-467.

Sato, H., Okada, Y. and Seiki, M. (1997). Membrane-type matrix metalloproteinases (MT- MMPs) in cell invasion. Thromb. Haemost. 78, 497-500.

Sato, H., Takino, T., Okada, Y., Gao, J., Shinagawa, A., Yamamoto, E and Seiki, M. (1994). A expressed on the surface of invasive tumour cells. Nature 370, 61-65.

Sato, S.M. and Sargent, T.D. (1991). Localized and inducable expression of Xenopus posterior (Xpo), a novel gene active in early frog embryos, encoding a protein with a "CCHC" finger domain. Development 112, 747-753.

Savage, C., Das, P., Finelli, A.L., Townsend, S.R., Sun, C.Y., Baird, S.E., Padgett, R.W. (1996). Caenorhabditis elegans genes sma-2, sma-3, and sma-4 define a conserved family of transforming growth factor beta pathway components. Proc. Natl Acad. Sci. USA 93, 790-4.

Schlunegger, M.P. and Grutter, M.G. (1992). An unusual feature revealed by the crystal structure at 2.2A resolution of human transforming growth factor-(32. Nature 358, 430- 434.

Schulte-Merker, S., Lee, K.J., McMahon, A.P. and Hammerschmidt, M. (1997). The zebrafish organizer requires chordino. Nature 387, 862-863.

Schulte-Merker, S., Smith, J.C. and Dale, L. (1994). Effects of truncated activin and FGF receptors and of follistatin on the inducing activities of BVgl and activin: does activin play a role in mesoderm induction. EMBO J. 13, 3533-3541.

Schultz-Cherry, S. and Murphy-Ullrich, I.E. (1993). Thrombospodin causes activation of latent transforming growth factor-p secreted by endothelial cells by a novel mechanism. J. C ell B io l 122, 923-932. 200 Sekelsky, J.J., Newfeld, S.J., Raftery, L.A., Chartoff, E.H., Gelbart, W.M. (1995). Genetic characterization and cloning of mothers against dpp, a gene required for decapentaplegic function in Drosophila melanogaster. Genetics 139, 1347-58.

Seleiro, E.A.P., Connolly, D.J. and Cooke, J. (1996). Early developmental expression and experimental axis determination by the chicken Vgl gene. Curr. B iol 6, 1476-1486.

Shelton-Inloes, B.B., Titani, K. and Sadler, I.E. (1986). cDNA sequences for human von Willebrand factor reveal five types of repeated domains and five possible protein sequence polymorphisms. Biochemistry IS, 3164-71

Shimell, .J., Ferguson, E.L., Childs, S.R. and O'Connor, M B. (1991). 'Tht Drosophila dorso-ventral patterning gene tolloid is related to human bone morphogenetic protein 1. Cell 67, 469-481.

Shull, M.M., Ormsby, I., Kier, A.B., Pawlowski, S., Diebold, R.J., Yin, M., Allen, R., Sidman, C., Proetzel, G., Calvin, D., et al. (1992). Targeted disruption of the mouse transforming growth factor-(3 gene results in multifocal inflammatory disease. Nature 3 5 9 , 693-699.

Sive, H.L. (1993). The frog prince-ss - a molecular formula for dorsoventral patterning in Xenopus. Genes Dev. 7, 1-12.

Sive, H.L., Hattori, K. and Weintraub, H. (1989). Progressive determination during formation of the anteroposterior axis in Xenopus laevis. Cell 58, 171-180.

Slack, J.M.W. (1991). ‘From Egg to Embryo. Regional specification in early development' 2nd Ed. Cambridge University Press.

Slack, J.M.W., Darlington, B.G., Heath, J.K. and Godsave, S.F. (1987). Mesoderm induction in early Xenopus embryos by heparin-binding growth-factors. Nature 3 2 6 , 197-

200.

Smith, J.C., Price, B.M.J., Green, J.B.A., Weigle, D. and Herrmann, B.G. (1991). Expression of a Xenopus homologue of brachyury (T) is an immediate-early response to mesoderm induction. Cell 67, 79-87.

Smith, W.C. and Harland, R.M. (1992). Expression cloning of noggin, a new dorsalizing factor localized to the Spemann organizer in Xenopus embryos. Cell 70, 829-840.

201 Smith, W.C., McKendry, R., Ribisi, S. Jr., and Harland, R.M. (1995). A nodal-related gene defines a physical and functional domain within the Spemann organizer. Cell 82, 37- 46.

Sokol, S.Y. (1996). Analysis of Dishevelled signalling pathways during Xenopus development. Curr. Biol. 6, 1456-1467.

Spencer, F.A., Hoffmann, F.M. and Gelbert, W.M. (1982). Decapentaplegic: a gene complex affecting morphogenesis in Drosophila embryos. Cell 28,451-461.

St. Johnston, R.D. and Gelbert, W.M. (1987). decapentaplegic transcripts are localized along the dorsal-ventral axisof the Drosophila embryo. EMBO J. 6, 2785-2791.

Steinbeisser, H., Fainsod, A., Niehrs, Y., Sasai, Y. and De Robertis, E.M. (1995). The role of gsc and Bmp-4 in dorsal-ventral patterning of the marginal zone in Xenopus: A loss of function study using antisense RNA. EMBO J. 14: 5230-5243.

Stenzel, P., Angerer, L.M., Smith, B.J., Angerer, R.C. and Vale, W.W. (1994). The univin gene encodes a member of the transforming growth factor-p superfamily with restricted expression in the sea urchin embryo. Dev. Biol. 166, 149-158.

Stephens, L.E., Sutherland, A.E., Klimanskaya, I V., Andrieux, A., Meneses, J., Pedersen, R.A. and Damsky, C.H. (1995). Deletion of pi integrins in mice results in inner cell mass failure and peri-implantation lethality. Genes Dev. 9, 1883-1895.

Stocker, W., Gomis-Riith, F-X., Bode, W. and Zwilling, R. (1993). Implications of the three-dimensional structure of astacin for the structure and function of the astacin family of zinc-endopeptidases. Eur. J. Biochem. 214, 215-231.

Stocker, W., Grams, F., Baumann, U., Reinemer, P., Gomis-Riith, F.X., McKay, D.B. and Bode, W. (1995). The metzincins - topological and sequential relations between the astacins, adamalysins, serralysins, and matrixins (collagenases) define a superfamily of zinc-endopeptidases. Protein Sci. 4, 823-840

Stocker, W. and Zwilling, R. (1995). Astacin. Meth. Enzymol. 248, 305-325.

Storm, E.E., Huynh, T.V., Copeland, N.G., Jenkins, N.A., Kingsley, D M. and Lee, S.J. (1994). Limb alterations in brachypodism mice due to mutations in a new member of the TGFp-superfamily. Nature 368, 639-643.

202 Su, M.W., Suzuki, H.R., Bieker, J.J., Solursh, M. and Ramirez, F. (1991). Expression of two nonallelic type II procollagen genes during Xenopus laevis embryogenesis is characterized by stage-specific production of alternatively spliced transcripts. J. Cell Biol. 115, 565-575.

Suzuki, A., Kaneko, B., Maeda, J. and Ueno, N. (1997). Mesoderm induction by BMP-4 and -7 heterodimers. Biochem. Biophys. Res. Comm. 232, 153-156.

Suzuki, A., Thies, R.S., Yamaji, N., Song, J.J., Wozney, J.M., Murakami, K. and Ueno, N. (1994). A tmncated bone morphogenetic protein-receptor affects dorsal-ventral patterning in the early Xenopus embryo. Proc. Natl. Acad. Sci. USA 91, 10255-10259.

Suzuki, N., Labosky, P.A., Furata, Y., Hargett, L., Dunn, R., Fogo, A.B., Takahara, K., Peters, D M.P., Greenspan, D.S. and Hogan, B.L.M. (1996). Failure of body wall closure in mouse embryos lacking procollagen C-proteinase encoded by B m p l, a mammalian gene related to Drosophila tolloid. Development 122, 3587-3595.

Taipale, J. and Keski-Oja, J. (1997). Growth factors in the extracellular matrix. FASEB J 11, 51-59.

Taipale, J., Miyazono, K., Heldin, C.H.and Keski-Oja, J. (1994). Latent transforming growth factor-p 1 associates to fibroblast extracellular matrix via latent TGF-(3 binding protein. J. Cell Biol. 124, 171-81.

Takahara, K., Brevard, R., Hoffman, G.G., Suzuki, N., Greenspan, D.S. (1996). Characterization of a novel gene product (mammalian tolloid-like) with high sequence similarity to mammalian tolloid/bone morphogenetic protein 1. Genomics 34, 157-165.

Takahara, K., Lyons, G.E. and Greenspan, D.S. (1994). Bone morphogenetic protein and a mammalian tolloid homologue (mTld) are encoded by alternatively spliced transcripts which are differentially expressed in some tissues. J. Biol. Chem. 269, 32572-32578.

Tannahill, D. and Melton, D.A. (1989). Localized synthesis of the Vgl protein during early Xenopus development. Development 106, 775-785.

Tarentino, A.L., Quinones, G., Grimwood, B.G., Hauer, C.R. and Plummer, T.H. Jr., (1995). Molecular cloning and sequence analysis of flavastacin; An O-glycosylated prokaryotic zinc metalloendopeptidase. Arch. Biochem. Biophys. 319, 281-285

ten Dijke, P., Miyazono, K. and Heldin, C.H. (1996). Serine/Threonine kinase receptors. Curr. Op. Cell Biol. 8, 139-145. 203 Thielens, N.M., van Dorsselaer, A., Gagnon, J. and Arlaud, G.J. (1990). Chemical and functional characterization of a fragment of Cls containing the epidermal growth factor homology region. Biochemistry 29, 3570-3578

Thomsen, G.H. and Melton, D.A. (1993). Processed Vgl protein is an axial mesoderm 'mducQT in Xenopus. Cell 74, 433-441.

Thomsen, G.H., Woolf, T., Whitman, M., Sokol, S., Vaughan, J., Vale, W. and Melton, D.A. (1990). Activins are expressed early in Xenopus embryogenesis and can induce axial mesoderm and anterior structures. Cell 63, 485-493.

Titani, K., Torff, H-J., Hormel, S., Kumer, S., Walsh, K.A., Rodl, J., Neurath, H. and Zwilling, R. (1987). Amino acid sequence of a unique protease from the crayfish Astacus fluviatlis. Biochemistry 26, 222-226.

Tomita, M., Kinoshita, T., Izumi, S., Tomino, S. and Yoshizato, K. (1994). Characterizations of sea urchin fibrillar collagen and its cDNA clone. Biochim. Biophys. Acta. 1217, 131-140.

Tosi, M., Duponchel, C., Meo, T. and Julier, C. (1987). Complete cDNA sequence of human complement Cls and close physical linkage of the homologous genes Cls and Clr. Biochemistry 26, 8516-8524.

Tsuneizumi, K., Nakayama, T., Kamoshida, Y., Kornberg, T.B., Christian, J.L. and Tabata, T. (1997). Daughters against dpp modulates dpp organizing activity in Drosophila wing development. Nature 389, 627-631.

Tucker, A.M. and Slack, J.M.W. (1995). Tail bud determination in the vertebrate embryo. Curr. Biol. 5, 807-813.

Turner, A.P., Brown, D., Heasman, J., Cook, G.M., Evans, J., Vickers, L. and Wylie, C.C. (1992). Involvement of a neutral glycolipid in differential cell adhesion in the Xenopus blastula. EM BO J. 11, 3845-3855.

Turner, D.L. and Weintraub, H. (1994). Expression of achaete-scute homologue 3 in Xenopus embryos converts ectodermal cells to a neural fate. Genes Dev. 8, 1443-1447.

Vale, W., Rivier, J., Vaughan, J., McClintock, R., Corrigan, A., Woo, W., Karr, D. and Spiess, J. (1986). Purification and characterization of an FSH releasing protein from porcine ovarian follicular fluid. Nature 321, 776-779.

204 van Wart, H E. and Birkedal-Hansen, H. (1990). The cysteine switch: a principle of regulation of metalloproteinase activity with potential applicability to the entire matrix metalloproteinase gene family. Proc. Natl Acad. Sci USA 87, 5578-5582.

Vassalli, A., Matzuk, M.M., Gardner, H.A., Lee, K.F. and Jaenisch, R. (1994). Activin/inhibin (3 B subunit gene disruption leads to defects in eyelid development and female reproduction. Genes Dev. 8, 414-427

Vincent, J-P., Oster, G.F. and Gerhart, J.C. (1986). Kinematics of grey crescent formation in Xenopus eggs : The displacement of subcortical cytoplasm relative to the egg surface. Dev. B io l 113, 484-500.

Vize, P.D. (1996). DNA sequences mediating the transcriptional response of the Mix.2 homeobox gene to mesoderm induction. Dev. B iol 177, 226-231. von Dassow, G., Schmidt, I.E. and Kimelman, D. (1993). Induction of the Xenopus organizer: expression and regulation of Xnot, a novel FGF and activin-regulated homeobox gene. Genes Dev. 7, 355-366.

Vukicevic, S., Latin, V., Chen, P., Batorsky, R., Reddi, A H. and Sampath, T.K. (1994). Localization of osteogenic protein-1 (bone morphogenetic protein 7) during human embryonic development: high affinity binding to basement membrane. Biochem. Biophys. Res. Comm. 198, 693-700.

Wall, N.A. and Hogan, B.L.M. (1994). TGF-(3 related genes in development. Curr. Op. Genet. Dev. 4, 517-522.

Wang, S., Krinks, M., Lin, K., Luyten, F.P. and Moos, M. Jr. (1997). Frzb, a secreted protein expressed in the Spemann organizer, binds and inhibits Wnt-8. Cell 88, 757-766.

Watabe, T., Kim, S., Candia, A., Rothbacher, U., Hashimoto, C., Inoue, K and Cho, K.W.Y. (1995). Molecular mechanisms of Spemman’s organsizer formation: conserved growth factor synergy between Xenopus and mouse. Genes Dev. 9, 3038-3050.

Watanabe, T.K., Suzuki, M., Omori, Y., Hishigaki, H., Horie, M., Kanemoto, N., Fujiwara, T., Nakamura, Y. and Takahashi, E. (1997). Cloning and characterization of a novel member of the human Mad gene family (MADH6). Genomics 42, 446-451.

Weeks, D.L. and Melton, D.A. (1987) A maternal mRNA localized to the vegetal hemisphere in Xenopus eggs codes for a growth factor related to TGF-p. Cell 51, 861- 867. 205 Wells, J.M. and Strickland, S. (1994). Aprotinin, a Kunitz-type protease inhibitor, stimulates skeletal muscle differentiation. Development 120, 3639-3647.

Wharton, K.A., Thomsen, G.H., Gelbart, W.M. (1991). Drosophila 60A gene, another transforming growth factor beta family member, is closely related to human bone morphogenetic proteins. Proc. Natl. Acad. Sci. USA 88, 9214-9218.

Whitman, M. (1997). Feedback from inhibitory SMADs. Nature 389, 549-551

Whittaker, C.A. and DeSimone, D.W. (1993). Integrin a subunit mRNAs are differentially expressed in early Xenopus embryos. Development 108, 229-238.

Wiersdorff, V., Lecuit, T., Cohen, S.M. and Mlodzik, M. (1996). Mad acts downstream of Dpp receptors, revealing a differential requirement for dpp signaling in initiation and propagation of morphogenesis in the Drosophila eye. Development 122, 2153-2162.

Wieser, R., Wrana, J.L. and Massagué, J. (1995). GS domain mutations that constitutively activate TpR-1, the downstream signaling component in the TGF-(3 receptor complex. EMBO J. 14, 2199-2208.

Wikramanayake, A.H. Brandhorst, B.P. and Klein, W.H. (1995). Autonomous and non- autonomous differentiation of ectoderm in different sea urchin species. Development 1 2 1 , 1497-1505.

Wikramanayake, A.H. and Klein, W.H. (1997). Multiple signalling events specify ectoderm and pattern the oral-aboral axis in the sea urchin embryo. Development 1 2 4 , 13- 20.

Wilson, P. A. and Hemmati-Brivanlou, A. (1995). Induction of epidermis and inhibition of neural fate by BMP4. Nature 376, 331-333.

Wilson, P. A., Lagna, G., Suzuki, A. and Hemmati-Brivanlou, A. (1997). Concentration-

dependent patterning of Ûiq Xenopus tctod&rm by BMP4 and its signal transducer Smadl. Development 124, 3177-3184.

Wilt, F.H. (1997). Looking into the sea urchin embryo you can see local cell interactions regulate morphogenesis. BioEssays 19, 665-668.

Wozney, J.M., Rosen, V., Celeste, A.J., Mitsock, L.M., Whitters, M.J., Kriz, R.W., Hewick, R.M. and Wang, E.A. (1988). Novel regulators of bone formation; Molecular clones and activities. Science 242, 1528-1534. 206 Wozney, J.M. (1989). Bone morphogenetic proteins. Prog. Growth Factor Res. 1, 267- 280.

Wrana, J.L. and Attisano, L. (1996). MAD-related proteins in TGF-P signalling. Trends Genet. 12, 493-497.

Wrana, J.L., Attisano, L., Carcamo, J., Zentella, A., Doody, J., Laiho, M., Wang, X.F. and Massagué, J. (1992). TGFp signals through a heteromeric protein kinase receptor complex. Cell 71, 1003-1014.

Wrana, J.L., Attisano, L., Wieser, R., Francesc, V. and Massagué, J. (1994). Mechanism of activation of the TGF-p receptor. Nature 370, 341-347.

Xu, R.H., Dong, Z., Maeno, M., Kim, J., Suzuki, A., Ueno, N., Sredni, D., Colburn, N.H., Kung, H.F. (1996). Involvement of Ras/Raf/AP-1 in BMP-4 signaling during Xenopus embryonic development. Proc. Natl. Acad. Sci. USA 93, 834-838.

Yamaguchi, K., Shirakabe, K., Shibuya, H., Irie, K., Oishi, I., Ueno, N., Taniguchi, T., Nishida, E. and Matsumoto, K. (1995). Identification of a member of the MAPKKK family as a potential mediator of TGF-P signal transduction. Science 270, 2008-2011.

Yamashita, H., ten Dijke, P., Huylebroeck, D., Sampath, T.K., Andries, M., Smith, J.C., Heldin, C.H. and Miyazono, K. (1995). Osteogenic protein-1 binds to activin type II receptors and induces certain activin-like effects. J. Cell. Biol. 130, 217-226.

Yang, J.T., Rayburn, H. and Hynes, R.O. (1993). Embryonic mesodermal defects in a^- integrin-deficient mice. Development 119, 1093-1105.

Yang, M., Murray, M.T. and Kurkinen, M. (1997). A novel matrix metalloprotease gene (XMMP) encoding vitronectin-like motifs is transiently expressed in Xenopus laevis early embryo development. J. Biol. Chem. 272, 13527-13533.

Yaswen, L., Kulkarni, A.B., Fredrickson, T., Mittleman, B., Schiffman, R., Payne, S. Longenecker, G, Mozes, E. and Karlsson, S. (1996). Autoimmune manifestations in the transforming growth factor-p 1 knockout mouse. Blood 87, 1439-1345.

Yayon, A., Klagsbrun, M., Esko, J.D., Leder, P. and Omitz, D M. (1991). Cell surface, heparin-like molecules are required for binding of basic fibroblast growth factor to its high affinity receptor. Cell 64, 841-848.

207 Zhang, Y., Feng, X., We, R. and Derynck, R. (1996). Receptor-associated Mad homologues synergize as effectors of the TGF-p response. Nature 383, 168-72

Zhou, X, Sasaki, H, Lowe, L, Hogan, B.L.M. and Kuehn, M R. (1993). Nodal is a novel TGF- -like gene expressed in the mouse node during gastrulation. Nature 361, 543-547.

Zimmerman, L.B., De Jesus-Escobar, J.M. and Harland, R.M. (1996). The Spemann organizer signal noggin binds and inactivates Bone Morphogenetic Protein 4. Cell 86, 599- 606.

208 APPENDIX BMPl-Related Metalloproteinases Promote the Development of Ventral Mesoderm in Early Xenopus Embryos

Shelley A. Goodman^, Rodolphe Albano^, Fiona C. Wardle^, Glenn Matthews^,

David TannahilP and Leslie Dale^’^

1. Department of Anatomy and Developmental Biology, University College London,

Gower Street, London, WCIE 6BT, U.K.

2. Department of Surgery, University of Birmingham, Queen Elizabeth Hospital,

Edgbaston, Birmingham, B15 2TH, U.K.

3. Department of Anatomy, University of Cambridge, Downing Street,

Cambridge, CB2 3DY, U.K.

4. Author of correspondence

Tel: 0171 419 3061

Fax: 0171 380 7349

E.Mail: [email protected]

Running title: BMPs in Xenopus development

Key words: XgMopwy/ventralization/BMPl/Tolloid/metalloendoprotease

11 SUMMARY

Bone morphogenetic protein 1 (BMPl) is a metalloproteinase closely related to Drosophila Tolloid

(Tld). Tld regulates dorsoventral patterning in early Drosophila embryos by enhancing the activity of

Dpp, a member of the TGF-B family most closely related to BMP2 and BMP4. In Xenopus BMP4 appears to play an essential role in dorsoventral patterning, promoting the development of ventral fates during gastrula stages. To see if BMP 1 has a role in regulating the activity of BMP4, we have isolated cDNAs for Xenopus BMPl and a novel closely related gene that we have called xolloid (xld).

Whereas xb m p l is uniformly expressed at all stages tested, the initial uniform expression of xld becomes localised to two posterior ectodermal patches flanking the neural plate and later to the inner ectoderm of the developing tailbud. xld is also expressed in dorsal regions of the brain during tailbud stages and is especially abundant in the ventricular layer of the dorsal hindbrain caudal to the otic vesicle. Overexpression of either gene inhibits the development of dorsoanterior structures in whole embryos, and ventralises activin-induced dorsal mesoderm in animal caps. Since ventralisation of activin induced animal caps can be blocked by coinjecting a dominant-inhibitory receptor for BMP2 and BMP4, we suggest a role for BMPl and Xld in regulating the ventralising activity of these molecules.

INTRODUCTION

Bone morphogenetic proteins (BMPs) were initially identified in bone extracts that induced ectopic bone formation in rodents (Wozney et al., 1988). Most of these molecules define a subgroup of closely related members of the TGF-6 family, that act as extracellular signals in a diverse range of developmental processes (reviewed by Hogan, 1996). In early Xenopus embryos BMP4 plays an important role in patterning the dorsoventral axis, promoting the development of both ventral mesoderm and epidermis. Overexpression of BMP4 converts dorsal into ventral mesoderm during gastrula stages (Dale et al., 1992; Jones et al., 1992; 1996), while inhibition of BMP signalling converts ventral into dorsal mesoderm (Graff et al., 1994; Maeno et al., 1994; Suzuki et al., 1994). In early gastrulae bmp4 is expressed throughout the animal cap and in ventral and lateral sectors of the marginal zone, but is excluded from a dorsal region of the marginal zone known as the organiser

(Fainsod et al., 1994; Schmidt et al., 1995; Hemmati-Brivanlou and Thomsen, 1995). The organiser releases signals that are responsible for both dorsalising lateral mesoderm and neuralising the animal

iii cap, signals that can be mimicked by three secreted factors - chordin, noggin and follistatin (reviewed by Hemmati-Brivanlou and Melton, 1997; Thomsen, 1997). All three proteins are synthesised by the organiser during gastrula stages and directly bind BMP4, thereby neutralising its activity (Piccolo et al., 1996; Zimmerman et al., 1996; Fainsod et al., 1997). The antagonistic interaction between these organiser signals and BMP4 may establish a morphogenetic gradient of BMP4 activity across the dorsoventral axis, cells adopting different fates as a function of its concentration (Dosch et al., 1997).

A similar system of extracellular signals is responsible for dorsoventral patterning in the early

Drosophila embryo, although the axis has been inverted (reviewed by De Robertis and Sasai, 1996).

Genetic studies have identified a small number of zygotic genes that regulate dorsoventral patterning in the early embryo, including decapentaplegic (dpp), tolloid (tld) and short gastrulation (sog)

(Ferguson and Anderson, 1992). The critical element in this system is Dpp, a member of the TGF-B family that is structurally and functionally homologous to BMP2 and BMP4 (Sampath et al., 1993;

Padgett et al., 1993; Holley et al., 1995). Dpp appears to function as a morphogen, specifying dorsal structures when activity is high and more ventral structures when activity is low (Ferguson and

Anderson, 1992). Moreover, the activity of Dpp is antagonised by ventrally expressed Sog, a structural and functional homologue of the Xenopus organiser signal chordin (Holley et al., 1995). In contrast, Tld is a metalloproteinase that enhances the activity of Dpp, possibly by releasing Dpp from inactive complexes with inhibitory binding proteins such as Sog (Shimell et al., 1991; De Robertis and Sasai, 1996).

In addition to the N-terminal metalloproteinase domain, Tld also contains five C-terminal CUB repeats and two EGF repeats that are thought to mediate protein-protein interactions (Shimell et al., 1991). A similar structure is also found in the mTld splicing variant of vertebrate BM Pl, the only BMP that is not a member of the TGF-B superfamily, while a shorter splicing variant has only three CUB repeats and single EGF repeat (Wozney et al., 1988; Fukagawa et al., 1994; Takahara et al., 1994). The fact that BMPl was copurified with BMP2 and BMP3 (Wozney et al., 1988) suggests that it may physically associate with one or more of the TGF-B-related BMPs, perhaps regulating the release of active BMP by proteolytic cleavage. More recently, BMPl was shown to be identical to procollagen

IV C-proteinase (PCP), an enzyme responsible for cleaving the C-terminal peptide from types I, II and

III procollagen (Kessler et al., 1996; Li et al., 1996). PCP is also capable of correctly processing prolysyl oxidase (Panchenko et al., 1996), an enzyme involved in intermolecular cross-linking in collagen fibres. This suggests that BMPl plays an important role in the assembly of mature collagen fibres, and a loss-of-function mutation in the mouse bmpl gene disrupts the normal processing and deposition of collagen (Suzuki et al., 1996). However, this does not preclude a wider role for BM Pl, or related proteinases, which may have many target proteins.

If BMP 1-related metalloproteinases are responsible for regulating the activity of BMP4, we might expect them to be involved in patterning the mesoderm during early Xenopus development.

Preliminary experiments with human BMPl suggested that this may indeed be the case, injected embryos typically lack a notochord (Dale, 1997). To further explore the role of BMP 1-related metalloproteinases in Xenopus development, we have isolated cDNAs for Xenopus BM Pl {xbm pl) and a closely related gene that we have called xolloid (xld). We find that both genes are uniformly expressed at blastula and gastrula stages, but that x/J becomes localised to the tip of the developing tail during neurula stages. Injection of synthetic mRNA for either gene into early X enopus embryos generates a similar, albeit weaker, phenotype to that previously described for injection of bmp4 RNA.

The resulting embryos lack dorsoanterior structures, including the notochord. XBMPl and Xld will also ventralise dorsal mesoderm induced in animal caps by activin, an effect that can be blocked by coinjection of a dominant-inhibitory receptor for BMP2 and 4. These results implicate BMP 1-related metalloproteinases as regulators of TGF-B-related BMPs during dorsoventral patterning in early amphibian embryogenesis.

MATERIALS AND METHODS

Embryo culture and manipulation

Xenopus laevis embryos were obtained by artificial fertilisation, allowed to rotate and then dejellied with 2% cysteine hydrochloride. Injected embryos were scored using the Dorso-Anterior Index of

Kao and Elinson (1988), in which a score of 5 = normal embryos, 4 = reduced eyes and forehead, 3

= cyclopic, 2 = microcephalic, 1 = acephalic and 0 = fully ventralised. Animal caps were isolated from mid-blastulae and incubated in full-strength MBS (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCOg, 15 mM HEPES, 0.3 mM CaNOg, 0.4 mM CaCl 2 , 0.8 mM MgSO^, 25|ig/l gentamycin, adjusted to pH

7.6) containing 10 units of human activin A until controls had reached stage 26.

Embryos required for histology were fixed in 4% paraformaldehyde and stained overnight in a saturated solution of borax carmine (in 35% ethanol). Paraffin sections (10 pm) were cut and counter stained with 0.1% naphthalene black in saturated aqueous picric acid. Whole mount in situ hybridisations were performed on albino embryos essentially as described by Harland (1991), with the exception that the RNase digestion step was omitted, CHAPS buffer was replaced by Maleic Acid buffer (100 mM Maleic Acid, pH 7.8; 150 mM NaCl; 0.1% Tween 20), and BM Purple (Boehringer) was used as a substrate. DIG labelled probes were made from the same DNA fragments as RNase protection probes (see below).

cDNA library screening

A Xenopus oocyte cDNA library prepared in lambda ZAP (Shuttleworth et al., 1988) was plated on

E. coli BB4 cells. Approximately 5x10^ plaques were screened with a random primed full-length probe for human BMPl (Wozney et al., 1988). Hybridisation was performed in 5 x SSPE, 20% formamide, 1 x Denhardt's solution and 100 pg/ml yeast tRNA at 42°C for 18 hours. Filters were washed in 2 x SSC and 0.1% SDS at 60°C. Three positive clones were isolated and sequenced using standard dideoxynucleotide chain-termination techniques. Two of the positive clones contained 3.2 and 4.8 kb cDNAs encoding xld, the larger one encoding the entire protein. The third clone contained a 2.4 kb cDNA encoding xbm pl, but lacked the initiating ATG and signal sequence. An additional 80 nucleotides, encoding the missing 25 N-terminal amino acids, was obtained by PCR from a Xenopus gastrula cDNA library prepared in pcDNAl (Invitrogen). PCR was performed with a 5' anchored primer - 5-GGTACCTAATACGACTCACTATAGGG-3' - and a 3' gene specific primer - 5'-

ACATAAGGAATGACCCC-3'. The 480 bp PCR product was ligated to the Xho I site of the 2.4 kb xbm pl cDNA, such that the entire open reading frame was generated.

Microinjections

VI xb m p l was subcloned into a modified version of pSP64T containing multiple cloning sites. A stop codon was introduced by PCR at the C-terminus of the metalloproteinase domain, generating the construct Axbm pl which was subcloned into pRN3. The Eco RI-Pvm II fragment of xld, encoding the full protein sequence, was also subcloned into pRN3. Human interstitial collagenase (m m pl) in pSP64 was provided by Dr. Dylan Edwards (University of Calgary). Six myc-tags were added to the

C-terminus of XBMPl and Xld by using PCR to subclone these cDNAs into the Bam HI site of pCS2^MT. Capped synthetic mRNAs were transcribed with either SP6 or T3 polymerase and taken up in sterile RNase free water. Embryos were injected with 10-15 nl of mRNA (0.2 }ig/|xl) at the 1-4 cell stage in 10% MBS plus 3% Ficoll (Sigma, Type 400). Fully grown Xenopus oocytes were injected with 20-30 nl of mRNA (1 |ig/p.l) and incubated for 48 hours in full-strength MBS plus 1%

BSA.

RNA Analysis

Poly (A)+ mRNA was isolated on oligo dT columns (GIBCO-BRL) according to the manufacturer's instructions. For Northern blots, RNAs were separated on formaldehyde-agarose (1.5%) gels, transferred to nylon (Amersham), and probed with random primed DNA probes made from the full- length cDNAs. RNase protections for xbm pl and xW were performed in 50% formamide using up to

10 |ig of total RNA per hybridisation. Probes were: 230 bp Eco Rl-Xho I fragment o f xb m p l, 250 bp

H ind III fragment of xld, and a 90 bp PCR fragment for ornithine decarboxylase (ode) (Isaacs et al.,

1992). Following hybridisation single stranded RNA was digested with 700 units/ml of RNase T1

(GIBCO-BRL) and 30 |ig/ml RNase A (Sigma). Control hybridisations with 10 |ig of tRNA were always negative and are not included in the Figures. RNase protections for a-actin (Mohun et al.,

1988) and a-globin (Walmsley et al., 1994) were performed using the Ambion RPA II kit according to the manufacturer's instructions, except that RNase digestion was carried out with 700 units/ml of

RNase T1 alone.

Protein Analysis

Capped synthetic mRNAs were translated in a message dependent Xenopus cell-free egg extract

(Matthews and Colman, 1991) in the presence of 1 mCi/ml of ^^S-methionine (Amersham). To test

v ii for glycosylation we included the tripeptide (acetyl)-Asn-Tyr-Thr-(amide) at 5 mM which inhibits N- glycosylation in this cell-free system. mRNAs were also translated in a message dependent rabbit reticulocyte lysate (Promega) according to the manufacturer's instructions. Radiolabelled proteins were separated by SDS-PAGE on a minigel apparatus (BioRad) according to the manufacturer's

instructions. After electrophoresis, gels were fixed and prepared for fluorography with Enhance

(DuPont).

Oocytes were homogenised in 100 mM Tris-HCl (pH 7.5), 1% Triton XlOO, 1 mM PMSF and

microfuged for 15 minutes to remove yolk platelets. Oocyte extracts and culture media were separated

by SDS-PAGE and transferred to Nylon membranes (Amersham). XBMPl and Xld were detected by

ECL (Amersham), according to the manufacturer’s instructions, using an antibody (9E10) against the

myc-tag epitope.

RESULTS

Isolation and characterisation of Xenopus BMPl-related cDNAs

cDNAs for Xenopus bmpl (%Wp7)and a novel gene we have called xolloid {xld), were isolated from

a screen of a Xenopus oocyte cDNA library (see Materials and Methods). A 2.5 kb cDNA for xb m p l

encodes a protein of 735 amino acids (Fig. lA) that is 93% identical to human bm p l (Wozney et al.,

1988), and 95.5% identical to a previously described partial cDNA for xb m p l (Maeno et al., 1993).

In addition to the N-terminal metalloproteinase domain, this cDNA encodes three C-terminal CUB

repeats and a single EGF repeat, a domain structure most similar to the short-splicing variant of

human BMPl (Wozney et al., 1988; Takahara et al., 1994). A longer splicing variant of xb m p l has

recently been described by Lin et al., (1997).

A 4.8 kb cDNA for xld encodes a protein of 1019 amino acids (Fig. lA) that is closely related to

BMPl. The N-terminal signal sequence is followed by a 121 amino acid pro-region with little

homology to that of BMP 1, except for a 21 amino acid sequence that is also found in the pro-regions

of human BMPl, mouse Tolloid-like, sea urchin BMPl, Aplysia TBL-1 and Drosophila Tld-related 1

(Fig. IB). Although the published sequence for murine BMPl lacks the first 12 amino acids of this

Vlll sequence (Fukagawa et al., 1994), they can be found in a different reading frame of the cDNA. These

12 amino acids are included in a 35 amino acid stretch that is 89% identical to the equivalent region in human BM Pl, suggesting that an error during cloning may have placed it in a different frame to the rest of the protein. There is currently no information to indicate the role of this conserved sequence, but it appears to contain an unpaired cysteine which may form an interchain disulphide bond with an as yet unidentified protein. Alternatively, this cysteine may act as a "cysteine switch", as proposed for the unpaired cysteine in the proregion of the matrix metalloproteinases (Van Wart and Birkedal-

Hansen, 1990). In these latter proteins, the unpaired cysteine inactivates the protease by interacting with the active site zinc atom, activation is achieved by breaking this interaction.

The 201 amino acid metalloproteinase domain of Xld is preceded by a short basic sequence (RVRR) believed to be a cleavage site for intracellular processing enzymes (Barr, 1991). A similar site is found at the same location in most BMPl-related metalloproteinases, and cleavage at this site is thought to be a requirement for enzymatic activity (Bode et al., 1992). The metalloproteinase domain of Xld is most closely related to mouse Tolloid-like (87% identity) and both Xenopus and mammalian BMPl (85% identity) (Fig. 1C). Like BMPl, the metalloproteinase domain of Xld contains two highly conserved sequences, HExxHxxGFxHExxRxDRD and YDxxS(IW)MHY (Fig. 1 A), that are responsible for co­ ordinating a zinc atom in the active site of astacin (Bode et al., 1992). The protease domain of Xld is followed by two CUB repeats, an EGF repeat and a third CUB repeat which together share 77% identity with Xenopus BMP\, xld also encodes additional EGF (xl) and CUB (x2) repeats, a similar structure to Drosophila Tld (Shimell et al., 1991) and the Tld splicing variant of BMPl (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997).

XBMPl and Xld are secreted glycoproteins.

Synthetic mRNA for xb m p l gave a translation product of Mr 80 kD in a rabbit reticulocyte cell-free translation system (Fig. 2A, track 6), quite close to the 83.6 kD predicted from the amino acid

sequence, while a product of Mr 90 kD was obtained when the same mRNA was translated in a

Xenopus cell-free system that possesses a functional endoplasmic reticulum (track 4) (Matthews and

Colman, 1991). That this increase in Mr was due to glycosylation was demonstrated by translating

IX xbmpl in the presence of an inhibitor of N-linked glycosylation, the tripeptide (acetyl)-Asn-Tyr-Thr-

[amide), which resulted in a smaller product of Mr 77.5 kD (track 5). Translation of synthetic mRNA for xld gave a product of Mr 117 kD in a rabbit reticulocyte cell-free translation system (track 3), which is quite close to the 115 kD predicted from the amino acid sequence. An almost identical product was obtained following translation of xld in the Xenopus cell-free system in the presence of the tripeptide inhibitor of glycosylation (track 2). A product of 135 kD was obtained in the absence of the inhibitor (track 1) demonstrating that Xld is also glycosylated. To confirm that XBMPl and Xld are secreted, we made constructs which added six myc-tags to the C-terminus of each protein and injected the mRNAs into Xenopus oocytes. Oocytes and culture media were collected after 48 hours, and proteins analysed by Western blots using the 9E10 myc antibody (Fig. 2B). Although most of the detectable BMPl and Xolloid was retained by the oocyte (tracks 3 and 5), both proteins could also be detected in the media (tracks 4 and 6). Expression of BMPl and Xld

Three transcripts of 2.5, 4.5 and 5.6 kb were detected when xbm pl was used to probe a Northern blot of poly (A)+ mRNA isolated from cleavage stages (2-8 cell) and an early tailbud stage (stage 28)

(Fig. 3A, tracks 1 and 2). This is similar to the situation found in mammals where the two largest transcripts are detected by probes specific for the mtld splicing variant (Takahara et al., 1994). When the blot was reprobed with xld, a major transcript of approximately 5.0 kb was detected at both stages

(tracks 3 and 4), while two minor transcripts of approximately 3.5 kb were present only at stage 28

(track 4). Since a full-length probe for xld does not detect any of the transcripts detected by x b m p l, and a nearly full-length probe for xb m p l does not detect transcripts for xld, it is clear that xld is a novel gene rather than a splicing variant oixbm pl.

The temporal and spatial expression patterns of xbm pl and xld were analysed in detail by RNase protections (Fig. 3B, C). Since the probe for xbm pl encompasses the proregion, it does not discriminate between the long and short-splicing variants, xbm pl and xld are expressed at all developmental stages tested, although transcript levels drop during gastrula and neurula stages before rising again in early (xlcl) and late {xbm pl) tailbud stages (Fig. 3B). A probe for the uniformly expressed ornithine decarboxylase {ode) gene confirms that roughly equivalent levels of RNA were

added to each assay. The result for xb m p l contrasts with those previously reported by Maeno et al.

(1993), who failed to detect any maternal expression of xb m p l. In our hands, maternal xb m p l expression has been confirmed using three independent assays: Northern blotting (Fig. 3A), RNase protection (Fig. 3B) and RT-PCR (data not shown). Moreover, oux xbm pl cDNA was isolated from a oocyte library. RNase protection of total RNA isolated from animal and vegetal hemispheres dissected from blastulae (stage 8) suggested that both xbm pl and xld are expressed in both hemispheres (Fig.

3C). A similar result was obtained at stage 10 (early gastrula), when xbm pl was also expressed at

similar levels in dorsal, lateral and ventral marginal zones. In contrast, there appeared to be slightly

more xld RNA in the dorsal marginal zone (track 4) than in lateral or ventral marginal zones (tracks 3

and 5). At stage 28, expression of xld, but not xbm pl, was largely restricted to the tail (track 9).

XI Finally, the spatial expression patterns of xbm pl and xld were analysed by whole mount in situ hybridisation, xb m p l was detected in most cells of the embryo at all stages tested (stage 10-28), although it appeared to be particularly abundant in the head and dorsal axis of tailbud stages (Fig. 4A,

B). The lack of signal in the endoderm is quite common using this technique. However, this probe also detects the more abundantly expressed long-splicing variants, which may mask differential expression of the short-splicing variant used in this study. In contrast, xld could not be detected until neurula stages, when expression was observed in two ectodermal patches of positive cells interspersed by negative cells that flank the posterior neural plate (Fig. 4C). This region of the embryo ultimately gives rise to the epidermis of the tail (Tucker and Slack, 1995), and at early tailbud stages xld is abundantly expressed in the inner layer of epidermis at the distal tip of the tail (Fig. 4D, F). A weaker signal was seen throughout the forebrain and in dorsal regions of the midbrain and hindbrain at tailbud stages (Fig. 4D), and was particularly prominent in the dorsal ventricular layer of the caudal hindbrain just posterior to the otic vesicle (Fig. 4D, E). The results are largely consistent with the

RNase protection data presented in Fig. 3C.

Overexpression of BMPl and Xld disrupts anterodorsal development

Preliminary experiments had shown that injection of human bm pl mRNA resulted in embryos that lacked dorsoanterior structures, including the notochord (Dale, 1997). To further investigate this phenotype we injected 7 or xld mRNA into each blastomere at the 2-cell stage. As controls we injected synthetic mRNA for A xbm pl, from which the C-terminal CUB and EGF repeats have been deleted (see Materials and Methods), or human collagenase (m m pl). Injected embryos developed normally until late gastrula stages, when a slight delay in blastopore closure was observed in embryos injected with xb m p l or xld. By tailbud stages it was clear that most of these embryos had anterior defects (AD embryos in Table 1, see Fig. 5B), but the proctodeum was also frequently enlarged (EP embryos in Table 1, see Fig. 5D) and in some cases an additional tail-like structure was observed (TD embryos in Table 1, see Fig. 5C). Since these tail-like structures rarely contained neural tissue or muscle, never contained notochord, and did not express Xpo (data not shown), a tail marker, it is unlikely that they represent true tails.

Xll Embryos were scored using the dorsoanterior index (DAI in Table 1) of Kao and Elinson (1988) in

which a score of 5 represents a normal embryo and a score of 0 a fully ventralised embryo (see

Materials and Methods), xb m p l and xld injections resulted in embryos with a weakly ventralised

phenotype and mean DAI scores of 3.84 and 3.68 respectively (Table 1). Embryos were rarely scored

as fully ventralised and only 10-20% were scored as 2 or lower; anterior defects rarely extended into

the hindbrain as indicated by normal expression of kroxlO in whole mount in situ hybridisations (data

not shown). In contrast, anterior defects were rarely observed in embryos injected with control RNAs

and the mean DAI scores were 4.90 for Axbm pl and 4.83 for m m p l. When analysed histologically

the most severely truncated embryos were found to lack a notochord and as a consequence the somitic

muscles had fused across the midline (Fig. 6B). The phenotypes we have described following the

injection of xb m p l and xld RNAs are very similar to those seen following the injection of low

concentrations of xbmp4 RNA (Dale et al., 1992), and are suggestive of a partial ventralisation of the

mesoderm.

: If xbmpl and xld ventralise dorsal mesoderm then localised injection of RNA into dorsal blastomeres

should result in a stronger phenotype than injection into ventral blastomeres. This was confirmed by

injecting either xb m p l or xld mRNAs into two dorsal or two ventral blastomeres at the 4-cell stage.

[ Although injection of xb m p l RNA was less effective than in previous experiments, anterior defects

; were seen in 30% of embryos (DAI = 4.53) following dorsal injections (Fig. 5B) but only in 6% of

I embryos (DAI = 4.94) following ventral injections. Similarly, whereas dorsal injections of xld caused

I anterior defects in 68% of cases (DAI = 3.68), ventral injections caused anterior defects in only 13%

I of cases (DAI = 4.59) (Fig. 5D, E). In contrast, an enlarged proctodeum and tail-like stmctures were

I more frequent following ventral injections (Fig. 5C). Injection of mRNA for A xbm pl had little effect

I on anterior development irrespective of the site of injection (Table 2 and Fig. 5). The results suggest

that both xbm pl and xld can ventralise dorsal mesoderm, resulting in anterior defects.

BMPl and Xld inhibit dorsal mesoderm in animal caps treated with activin

We have previously shown that bmp4 blocks the formation of dorsal mesoderm in animal caps treated

with activin (Dale et al., 1992; see also Jones et al., 1992; 1996), a potent inducer of mesoderm in this

Xlll system. Since both xbm pl and xld have a similar, albeit weaker, effect to bmp4 on whole embryos,

we wished to know if they could also inhibit dorsal mesoderm formation in animal caps. To test this

j animal caps were isolated at midblastula stages from embryos injected with either x b m p l, xld or

I control mRNAs and then incubated in the presence of 10 units of activin. These caps did not elongate

in the absence of activin (Fig. 7A), and no mesoderm was detected upon histological examination j (data not shown). In contrast, activin caused control caps to elongate (Fig. 7B), and histological

analysis showed that 80-100% of them differentiated muscle (data not shown). Elongation did not

occur in animal caps injected with either xbm pl or xld mRNAs (Fig. 1C, D), and muscle was

replaced by ventral mesoderm in all xld injected caps and in 79% of xbm pl injected caps (data not

shown). In a separate experiment, injected caps treated with activin were incubated until control

efnbryos had reached stage 25 and assayed by RNase protection for the expression of muscle-specific

a-actin, a dorsal marker, and a-globin, a ventral marker (Fig. BA). Animal caps from embryos

injected with either xb m p l or xld (tracks 4 and 5) had greatly reduced levels of a-actin when

compared to controls (tracks 2, 3 and 6), while animal caps injected with xld exhibited a

corresponding increase in a-globin (track 5). The results are consistent with xbm pl and xld

converting dorsal mesoderm into ventral mesoderm, as previously observed for bmp4 (Dale et al.,

1992; Jones et al., 1992; 1996).

One explanation for these results is that XBMPl and Xld are activating TGF-B-related BMPs such as

BMP2 or BMP4, which in turn ventralise the mesoderm. If this were the case we would expect

, inhibitors of BMP signalling pathways to block the ventralising activity of XBMPl and Xld. To test

, this we have coinjected xbm pl and xld mRNAs with mRNA encoding a dominant-inhibitory receptor

I (tBR) for BMP2 and BMP4. This receptor lacks the cytoplasmic serine-threonine kinase domain and

, blocks the response of animal caps to both BMP2 and BMP4, but not activin (Suzuki et al., 1994).

Animal caps were isolated from injected embryos at stage 8 and incubated in the presence of 10 units

of activin. In contrast to animal caps injected with either o ix ld alone (Fig. 8B, tracks 2 and 4),

animal caps coinjected with tbr both elongated (data not shown) and expressed a-actin (tracks 3 and

5). This result indicates that the ventralising activity of both XBMPl and Xld requires a functional

XIV BMP signalling pathway, consistent with a role for these metalloproteinases in regulating the activity of BM P2 and/or BMP4.

DISCUSSION

In this paper, we describe the isolation and characterisation of two Xenopus cDNAs encoding astacin- like metalloproteinases closely related to mammalian BMPL Whereas the first of these encodes the short-splicing variant of Xenopus BMPl, the second encodes a novel metalloproteinase we have called Xolloid (Xld). The domain structure of Xld is identical to that of Drosophila Tld (Shimell et al.,

1991), both proteinases containing additional CUB and EGF repeats when compared to the short-

splicing variant of BMPL This domain structure is also shared with the Tld splicing variant of both

mammalian and Xenopus BMPl (Fukagawa et al., 1994; Takahara et al., 1994; Lin et al., 1997), a

novel mouse gene called mTld-like (Takahara et al., 1996), Aplysia Tld/BMPl-like (Liu et al., 1997)

and Drosophila Tld-related 1 (Nguyen et al., 1994; Finelli et al., 1995). A different arrangement of

CUB and EGF repeats are found in astacin-like metalloproteinases from the sea urchin (BP 10 and

SpAN) (Lepage et al., 1992; Reynolds et al., 1992) and C. elegans (HCH-1) (Hishida et al., 1996).

In addition, the Xenopus hatching enzyme (XHE) is an astacin-like metalloproteinase with two C-

terminal CUB, but no EGF, repeats (Katagiri et al., 1997). Since CUB domains are thought to

mediate protein-protein interactions (Bork and Beckmann, 1993), differences in the amino acid

sequence and/or arrangement of these domains may target these metalloproteinases to different I ' substrates. Evidence in favour of different substrate specificities has been provided by studies in

I Drosophila, where Tld and Tld-related 1 are not interchangeable (Nguyen et al., 1994; Finelli et al.,

I 1995).

I I Genetic studies in Drosophila suggest that the BMPl-related metalloproteinase Tld enhances the

activity of Dpp, a structural and functional homologue of both BMP2 and BMP4 (Shimell et al.,

1991; Ferguson and Anderson, 1992). In this paper, we provide evidence that BMPl and Xld may

play a similar role in early Xenopus embryos. First, overexpression of BMPl and Xld results in

embryos that lack the most anterior structures of the head, and in the trunk fail to differentiate a

notochord. A similar phenotype is also seen following injection of low concentrations of BMP4

XV mRNA (Dale et al., 1992; Dosch et al., 1997). Second, BMPl and Xld ventralise animal caps exposed activin, a member of the TGF-B family that induces dorsal mesoderm. Once again, similar results have been obtained for BMP4 (Dale et al., 1992; Jones et al., 1996). Finally, we have shown that the ventralising activity of BMPl and Xld, can be blocked by coexpressing a dominant-negative

receptor for BMPs. Since this receptor blocks signalling by BMP2 and BMP4, but not activin (Graff

^ et al., 1994; Suzuki et al., 1994), our results demonstrate that a functional BMP signalling pathway is

required for the ventralising activity of these metalloproteinases. The simplest explanation is that

I overexpression of BMPl and Xld increases the activity of Xenopus TGF-6-related BMPs during

I blastula and gastrula stages. These may include BMP2, BMP4, BMP? and a BMP-related gene called

I antidorsalizing morphogenetic protein (ADMP) (Dale et al., 1992; Nishimatsu et al., 1992; Hawley et i al., 1995; Hemmati-Brivanlou and Thomsen, 1995; Moos et al., 1995).

: In contrast to the results presented here, Lin et al. (1997) have provided evidence that Xenopus BM Pl

' dorsalises ventral mesoderm. These authors injected mRNAs for either XBMPl or the more

I abundantly expressed XTld splicing variant into both ventral blastomeres at the 4-cell stage, and

I subsequently isolated the ventral marginal zone (VMZ) from early gastrulae. Whereas XBMPl had no

I effect on the differentiation of injected VMZs, XTld converted ventral mesoderm (e.g. blood) into I dorsal mesoderm (muscle) and upregulated expression of the dorsal marginal zone markers goosecoid

and chordin. These are results expected of a molecule that inhibits BMPs, rather than one that

activates them. We have repeated these experiments using XBMPl and Xld, and obtained no evidence

that these metalloproteinases are capable of dorsalising VMZs (data not shown). One explanation for

these different results is that the XTld splicing variant used by Lin et al. (1997), which contains

additional CUB and BGF repeats, has a different substrate specificity to the short-splicing variant of

BMPl used in our studies.

Our results suggest that XBMPl and Xld ventralise dorsal mesoderm by enhancing the activity of

TGF-B-related BMPs, but how is this achieved? One explanation is that these metalloproteinases

interact directly with the TGF-B-related BMPs themselves, as suggested by co-purification of BMPl

with BMP2 and BMP3 (Wozney et al., 1988). Like most members of the TGF-B superfamily BMPs

XVI are first synthesised as large inactive precursors from which the active C-terminal dimers are proteolytically cleaved. Perhaps the function of BMPl and Xld is to serve as the processing enzyme responsible for this cleavage event, although the cleavage sites more closely resemble the consensus sequence (RX[K/R]R) for subtilisin-like serine proteinases (Barr, 1991). While cleavage sites for Xld have yet to be determined, the known cleavage sites for BMPl do not match this consensus sequence

(Kessler et al., 1986; Panchenko et al., 1996). Alternatively, these metalloproteinases could release

TGF-B-related BMPs from latent complexes with their own proregions. Although such complexes have not been described for BMPs, they have been well characterised for TGF-B itself and proteinase treatment has been shown to release the active molecule (Lyons et al., 1988). However, XBMPl and

Xld do not enhance the ventralising activity of coexpressed XBMP4 (data not shown), suggesting that these metalloproteinases do not activate BMP4 directly.

A second explanation is that BMPl and Xld degrade components of the extracellular matrix (ECM), thereby releasing bound TGF-B-related BMPs. ECM proteins are common substrates for metalloproteinases and astacin, the prototypical enzyme for BMPl, completely degrades the native

I triple helix of type I collagen (Stocker and Zwilling, 1995). Moreover, we have evidence that SpAN,

a sea urchin BMPl-related metalloproteinase (Reynolds et al., 1992), degrades fibronectin when

■ expressed in early Xenopus embryos, although this is a property not shared by either XBMPl or Xld

I (Wardle and Dale, unpublished). At present we do not know if Xld acts on the ECM, but BMPl is

I responsible for cleaving the C-terminal peptides from procollagens I, II and III (Kessler et al., 1996; 1 I Li et al., 1996). BMPl also correctly activates prolysyl oxidase (Panchenko et al., 1996), an enzyme

I involved in intermolecular cross-linking in collagen fibres, and a loss-of-function mutation in the

I mouse bm pl gene disrupts the assembly of collagen fibres during embryogenesis (Suzuki et al.,

I 1996). These results suggest that BMPl is responsible for the correct assembly of collagen fibres,

rather than degradation. However, the presence of multiple BMPl-related genes (Takahara et al.,

1996), which might compensate for the loss of BMPl, means that other roles cannot be excluded. In

Xenopus fibrillar collagens are not expressed until neurula stages (Su et al., 1991), so it seems likely

that maternal BMPl, and Xld, have other substrates. Some of these may be components of the ECM

that sequester BMPs.

xvii A third explanation is that BMPl and Xld release TGF-B-related BMPs sequestered by inhibitory binding proteins such as chordin, noggin and follistatin (Piccolo et al., 1996; Zimmerman et al., 1996;

Fainsod et al., 1997). We have recently shown that Xld can cleave chordin, but not noggin, in vitro, and provided evidence that this releases active BMPs (Piccolo et al., 1997). Xld also blocks the dorsalising activity of chordin in vivo, but not that of noggin and follistatin, suggesting that specific cleavage of chordin may also be its in vivo role. Consistent with this, xld is expressed in the inner epidermal layer of the developing tailbud, in close proximity to the chordoneural hinge which expresses chordin (Sasai et al., 1995). At gastrula stages chordin is localised to the organiser (Sasai et al., 1995), while xld is expressed by all cells and may even be more abundant in the organiser. This is different from the situation in Drosophila where Tld is coexpressed with Dpp at the cellular blastoderm stage, and not in cells expressing the chordin homologue Sog (Shimell et al., 1991; Francois et al.,

1994). These differences may reflect the fact that chordin directly binds both BMP2 and BMP4

(Piccolo et al., 1996), and the fact that bmp2 is expressed in the organiser (Hemmati-Brivanlou and

Thomsen, 1995). In addition, the organiser also expresses both bmp7 and admp (Hawley et al., 1995;

Moos et al., 1995), which may also interact with chordin. Although the function of these

“ventralising” genes in the organiser has yet to be elucidated, a requirement for Xld in regulating inhibitory interactions with chordin may explain expression of xld in the organiser. At present we do not know if XBMPl also cleaves chordin, or whether it cleaves other inhibitory binding proteins such as noggin or follistatin. However, since XBMPl and Xld do not act synergistically in coinjection experiments, the DAI score of coinjected embryos is similar to that of singularly injected embryos

(data not shown), it is possible that they act on the same substrates.

Recently it has become clear that dorsoventral patterning in the early embryo of both Drosophila and

X enopus, is mediated by a conserved system of extracellular signals (reviewed by De Robertis and

Sasai, 1996). Dorsoventral fates are specified by the graded activity of two closely related members of the TGF-B family, Dpp {Drosophila) and BMP4 (Xenopus), gradients that may be established by diffusion of the inhibitory binding proteins Sog (Drosophila) and chordin (Xenopus). Our results suggest that an additional layer of regulation may also be conserved between Drosophila and

XVlll Xenopus, in that BMPl-related metalloproteinases may play an essential role in establishing the

activity gradient of Dpp/BMP4.

Acknowledgements

We would like to thank Elizabeth Wang for providing the human BMPl cDNA, Dylan Edwards for

i the human Collagenase cDNA, Naoto Ueno for the tBR cDNA, John Shuttleworth for the oocyte

cDNA library, Jim Smith for human activin A and Kate Nobes for the 9E10 antibody. This work was

funded by the British Medical Research Council and the Wellcome Trust. FCW is a Wellcome Prize

student.

REFERENCES

Barr, P.J. (1991). Mammalian subtilisins: The long-sought dibasic processing endoproteases. Cell 66, 1-3

Bode, W., Gomis-Rtith, F.X., Huber, R., Zwilling, R. and Stocker, W. (1992). Structure of astacin and implications for activation of astacins and zinc-ligation of collagenase. Nature 358, 164- 167.

Bork, P. and Beckman, G. (1993). The CUB domain a widespread molecule in developmentally regulated proteins. J. M o l B iol 231, 539-545.

; Dale, L. (1997). Bone morphogenetic proteins in amphibian development. In ''The Astacins: Structure and Function of a New ”. (R. Zwilling and W. Stocker, eds). pp 235-246.

: Dale, L., Howes, G., Price, B.M.J. and Smith, J. C. (1992). Bone morphogenetic protein 4: A ventralising factor in early Xenopus development. Development 115, 573-585. j De Robertis, E.M. and Sasai, Y. (1996). A common plan for dorsoventral patterning in Bilateria. I N ature 380, 37-40.

I Dosch, R., Gawantka, V., Delius, H., Blumenstock, C. and Niehrs, C. (1997). Bmp-4 acts as a 1 morphogen in dorsoventral mesoderm patterning in Xenopus. Development 124, 2325-2334.

[ Fainsod, A., Steinbeisser, H. and De Robertis, E.M. (1994). On the function of BMP-4 in patterning I the marginal zone of the Xenopus embryo. EMBO J. 13, 5015-5025.

I Fainsod, A., Deissler, K., Yelin, R., Marom, K., Epstein, M., Pillemer, G., Steinbeisser, H. and 1 Blum, M. (1997). The dorsalizing and neural inducing gent follistatin is an antagonist of BMP- I 4. Meek. Dev. 63, 39-50.

Ferguson, E.L. and Anderson, K.V. (1992). Localized enhancement and repression of the activity of the TGF-B family member, decapentaplegic, is necessary for dorsal-ventral pattern formation in the Drosophila embryo. Development 114, 583-597.

Finelli, A.L., Xie, T., Bossie, C.A., Blackman, R.K. and Padgett, R.W. (1995). The tolkin gene is a tolloidfOMP-l homologue that is essential for Drosophila development. Genetics 141, 271- 281.

XIX Francois, V., Solloway, M., O'Neill, J.W., Emery, J. and Bier, E. (1994). Dorso-ventral patterning of the Drosophila embryo depends on a putative negative growth factor encoded by the short gastrulation gene. Genes Dev. 8, 2602-2616.

Fukagawa, M., Suzuki, N., Hogan, B.L.M. and Jones, C.M. (1994). Embryonic expression of mouse bone morphogenetic protein-1 (BMP-1), which is related to the Drosophila dorsoventral gene tolloid and encodes a putative astacin metalloendopeptidase. Dev. Biol. 163, 175-183.

I Graff, J.M., Thies, R.S., Song, J.J., Celeste, A.J. and Melton, D.A. (1994). Studies with a I Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override dorsal signals I in vivo. Cd/79, 169-179.

Harland, R.M. (1991). In situ hybridization: an improved whole mount method for X enopus embryos. Methods Cell Biol. 36, 685-695.

Hawley, S.H.B., Wunnenberg-Stapleton, K., Hashimoto, C., Laurent, M.N., Watabe, T., Blumberg, B.W. and Cho, K.W.Y. (1995). Disruption of BMP signals in embryonic X enopus ^ ectoderm leads to direct neural induction. Genes & Dev. 9, 2923-2935.

Hçmmati-Brivanlou, A. and Melton, D.A. (1997). Vertebrate embryonic cells will become nerve cells unless told otherwise. Cell 88, 13-17.

Hemmati-Brivanlou, A. and Thomsen, G. (1995). Ventral mesodermal patterning in X enopus embryos: expression patterns and activities of BMP-2 and BMP-4. Dev. Genet. 17, 78-89.

Hishida, R., Ishihara, T., Kondo, K. and Katsura, I. (1996). hch-1, a gene required for normal hatching and normal migration of a neuroblast in C. elegans, encodes a protein related to TOLLOID and BMP-1. EMBO 7. 15,4111-4122.

■ Holley, S.A., Jackson, P.D., Sasai, Y., Lu, B., De Robertis, E., Hoffmann, E.M. and Ferguson, ; E.L. (1996). A conserved system for dorsal-ventral patterning in insects and vertebrates involving sog and chordin. Nature 376, 249-253.

I Isaacs, H.V., Tannahill, D. and Slack, J.M.W. (1992). Expression of a novel FGF in the Xenopus embryo. A candidate inducing factor for mesoderm formation and anteroposterior specification. Development 114, 711-720.

: Jones, C M., Dale, L., Hogan, B.L.M., Wright, C.V.E. and Smith, J.C. (1996). Bone morphogenetic protein-4 (BMP-4) acts during gastrula stages to cause ventralization of Xenopus I Qmhryo^. Development 122, 1545-1554.

I Jones, C M., Lyons, K.M., Lapan, P.M., Wright, C.V.E. and Hogan, B.L.M. (1992). DVR-4 I (Bone Morphogenetic Protein-4) as a posterior-ventralizing factor in Xenopus mesoderm I induction. Development 115, 639-647.

Kao, K. R. and Elinson, R. P. (1988). The entire mesodermal mantle behaves as Spemann's organizer in dorso-anterior enhanced Xenopus laevis embryos. Dev. Biol. 127, 64-77.

Katagiri, C., Maeda, R., Yamashika, C., Mita, K., Sargent, T.D. and Yasumasu, S. (1997). Molecular cloning of Xenopus hatching enzyme and its specific expression in hatching gland cells. Int. J. Dev. Biol. 41, 19-25.

Kessler, E., Adar, R., Goldberg, B. and Niece, R. (1986). Partial purification and characterization of a procollagen C-proteinase from culture medium of mouse fibroblasts. Collagen Rel. Res. 6, 249-266.

Kessler, E., Takahara, K., Biniaminov, L., Brusel, M. and Greenspan, D.S. (1996). Bone morphogenetic protein-1: The type 1 procollagen C-proteinase. Science 271, 360-362.

XX Lepage, T., Ghiglione, C. and Gâche, C. (1992). Spatial and temporal expression pattern during sea urchin embryogenesis of a gene coding for a protease homologous to the human protein BMP-1 and to the product of the Drosophila dorsal-ventral patterning gene tolloid. Development 1 1 4 , 147-164.

Li, S-W., Sieron, A.L., Fertala, A., Hojima, Y., Arnold, W.V. and Prockop, D.J. (1996). The C- proteinase that processes procollagens to fibrillar collagens is identical to the protein previously identified as bone morphogenetic protein-1. Proa. Natl. Acad. Sci. USA 93, 5127-5130.

Lin, J.J., Maeda, R., Ong, R.C., Kim, J., Lee, L.M., Kung, H.F. and Maeno, M. (1997). XBMP- IB (Xtld), a Xenopus homolog of the dorso-ventral polarity gene in Drosophila, modifies tissue phenotypes of ventral explants. Develop. Growth & Diff. 39, 43-51.

Liu, Q-R., Hattar, S., Endo, S., MacPhee, K., Zhang, H., Cleary, L.J., Byrne, J.H. and Eskin, A. (1997). A developmental gene {Tolloid/BMP-1) is regulated in Aplysia neurons by treatments that induce long-term sensitization. J. Neurosci. 17, 755-764.

Lyons, R., Keski-Oja, J. and Moses, H. (1988). Proteolytic activations of latent transforming growth factor-8 from fibroblast conditioned medium. J. Cell Biol. 106, 1659-1665.

Maeno, M., Xue, Y., Wood, T.I., Ong, R.C. and Kung, H.F. (1993). Cloning and expression of cDNA encoding /flevw bone morphogenetic protein-1 during embryonic development. Gene 134, 257-261.

Maeno, M., Ong, R.C., Suzuki, A., Ueno, N., Kung, H.F. (1994). A truncated bone morphogenetic protein 4 receptor alters the fate of ventral mesoderm to dorsal mesoderm: roles of animal pole tissue in the development of ventral mesoderm. Proc. Natl. Acad. Sci. USA. 91, 10260-10264.

Matthews, G. and Colman, A. (1991). A highly efficient, cell-free translation/translocation system prepared from Xenopus eggs. Nucl. Acids Res. 19, 6405-6412.

Mohun, T.J., Garrett, N., Stutz, F. and Spohr, G. (1988). A third striated muscle actin gene is expressed during early development in the amphibian Xenopus laevis. J. Mol. Biol. 2 0 2 , 67- 76.

Moos, M., Wang, S. and Krinks, M. (1995). Anti-dorsalizing morphogenetic protein is a novel TGF- 8 homolog expressed in the Spemann organizer. Development 121, 4293-4301.

Nguyen, T., Jamal, J., Shimell, M.J., Arora, K. and O'Connor, M B. (1994). Characterization of tolloid-related-1: A BMP-1-like product that is required during larval and pupal stages of Drosophila development. Dev. Biol. 166, 569-586.

Nishimatsu, S., Suzuki, A., Shoda, A., Murakami, K. and Ueno, N. (1992). Genes for bone morphogenetic proteins are differentially transcribed in early amphibian embryos. Biochem. Biophys. Res. Comm. 186, 1487-1495.

Padgett, R.W., Wozney, J. and Gelbert, W.M. (1993). Human BMP sequences can confer normal dorsal-ventral patterning in the Drosophila embryo. Proc. Natl. Acad. Sci. 90, 2905-2909.

Panchenko, M.V., Stetler-Stevenson, W.G., Trubetskoy, Q.V., Gacheru, S.N. and Kagan, H.M. (1996). Metalloproteinase activity secreted by fibrogenic cells in the processing of prolysyl oxidase. J. Biol. Chem. 271, 7113-7119.

Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L. and De Robertis, E.M. (1997). Cleavage of Chordin by the Xolloid metalloprotease suggests a role for proteolytic processing in the regulation of Spemann organizer activity. Cell. 91, 407-416.

XXI Piccolo, S., Sasai, Y., Lu, B. and De Robertis, E.M., (1996). Dorsoventral patterning in Xenopus: inhibition of ventral signais by direct binding of chordin to BMP-4. Cell 86, 589-598.

Reynolds, S.D., Angerer, L.M., Palis, J., Nasir, A. and Angerer, R.C. (1992). Early mRNAs, spatially restricted along the animal-vegetal axis of sea urchin embryos, include one encoding a protein related to tolloid and BMP-1. Development 114, 769-786.

Sampath, T.K., Rashka, K.E., Doctor, J.S., Tucker, R.F. and Hoffman, P.M. (1993). Drosophila transforming growth factor B superfamily proteins induce endochondral bone formation in mammals. Proc. Natl. Acad. Sci. 90, 6004-6008.

Schmidt, I.E., Suzuki, A., Ueno, N. and Kimelman, D. (1995). Localized BMP-4 mediates dorso­ ventral patterning in the early Xenopus embryo. Dev. Biol. 169, 37-50.

Shimell, M.J., Ferguson, E.L., Childs, S.R. and O'Connor, M B. (1991). Tho, Drosophila dorsal- ventral patterning gene tolloid is related to human Bone Morphogenetic Protein 1. Cell 6 7 , 469- 481.

Shuttleworth, J., Godfrey, R. and Colman, A. (1988). p40^O15 ^ c<7c2-related protein kinase involved in negative regulation of meiotic maturation of Xenopus oocytes. EM BO J. 9, 3233- 3240.

Stocker, W. and Zwilling, R. (1995). Astacin. Methods. Enzymol. 248, 305-325.

Su, M.W., Suzuki, H.R., Bieker, J.J., Solursh, M. and Ramirez, F. (1991). Expression of two nonallelic type II procollagen genes during Xenopus laevis embryogenesis is characterized by stage-specific production of alternatively spliced transcripts. J. Cell Biol. 115, 565-575.

Suzuki, A., Thies, R.S., Yamaji, N., Song, J.J., Wozney, J.M., Murakami, K., & Ueno, N. (1994). A truncated BMP receptor affects dorso-ventral patterning in the early Xenopus embryo. Proc. Natl. Acad. Sci. USA 91, 10255-10259.

Suzuki, N., Labosky, P.A., Furata, Y., Hargett, L., Dunn, R., Fogo, A.B., Takahara, K., Peters, D M.P., Greenspan, D.S. and Hogan, B.L.M. (1996). Failure of ventral body wall closure in ! mouse embryos lacking procollagen C-proteinase encoded by B m p l, a mammalian gene related to Drosophila tolloid. Development 122, 3587-3595.

I Takahara, K., Lyons, G.E. and Greenspan, D.S. (1994). Bone morphogenetic protein-1 and a I mammalian Tolloid homologue (mTld) are encoded by alternatively spliced transcripts which are differentially expressed in some tissues. J. Biol. Chem. 269, 32572-32578.

[ Takahara, K., Brevard, R., Hoffman, G.G., Suzuki, N. and Greenspan, D.S. (1996). Characterisation of a novel gene product of (mammalian tolloid-like) with high sequence I similarity to mammalian tolloid/bone morphogenetic protein-1. Genomics 34, 157-165.

I Tucker, A.M. and Slack, J.M.W. (1995). Tail bud determination in the vertebrate embryo. Curr. I Biol. 5, 807-813.

Thomsen, G.H. (1997). Antagonism within and around the organizer: BMP inhibitors in vertebrate body patterning. Trends Genet. 13, 209-211.

Van Wart, H E. and Birkedal-Hansen, H. (1990). The cysteine switch: A principle of regulation of metalloproteinase activity with potential applicability to the entire matrix metalloproteinase gene family. Proc. Natl. Acad. Sci. USA 87, 5578-5582.

Walmsley, M.E., Guille, M.J., Bertwistle, D., Smith, J.C., Pizzey, J.A. and Patient, R.K. (1994). Negative control of Xenopus GATA-2 by activin and noggin with eventual expression in precursors of the ventral blood islands. Development 120, 2519-2529.

xxii Wozney, J.M., Rosen, V., Celeste, A.J., Mitsock, L.M., Whitters, M J., Kriz, R.W., Hewick, R.M. and Wang, E.A. (1988). Novel regulators of bone formation: molecular clones and activities. Science 242, 1528-34.

Zimmerman, L.B., De Jesus-Escobar, J.M. and Harland, R.M. (1996). The Spemann organizer signal noggin binds and inactivates Bone Morphogenetic Protein 4. Cell 86, 599-606.

^ Figure Legends

' Figure 1. (A) Alignment of the deduced amino acid sequences of Xld and XBMPl (only non-

‘ conserved amino acids are indicated for XBMPl) are presented beneath a schematic of the domain

structure of Xld, the identity of each domain is indicated above the schematic. Note that relative to

I XBMPl, Xld contains one additional EGF repeat and two additional CUB repeats. Conserved

; sequences in the proregion are underlined as are the highly conserved metalloproteinase domain

sequences responsible for co-ordinating a zinc atom in the active site of astacin. These sequences will

appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under the accession

I numbers Y09660 and Y09661. (B) Alignment of the highly conserved 21 amino acid sequence in the

1 proregion of the BMPl-related metalloproteinases Xolloid (Xld), Xenopus BMPl (XBMPl), human j BMPl (HBMPl), murine BMPl (MBMPl), mammalian Tolloid-like (MTLL), sea urchin BMPl i: I (SBMPl), Drosophila Tolloid-related 1 (DTLRl) and Drosophila Tolloid (DTLD). The murine

sequence has been modified to take account of a possible error in the published sequence, the first 12

amino acids (underlined) are in a different reading frame to the last 9 amino acids. Although DTLD

also has a single cysteine in the proregion, the flanking amino acids share no homology to the other

seven proteins. (C) Percentage identity between Xolloid and other BMPl related proteins is presented

beneath schematics illustrating the domain stmcture of XBMPl and Xolloid. The percentage identity

between each domain is presented separately, figures for the highly diverged proregion are not

I included. We also include the Xenopus and human Tld splice variants of BMPl (XTld and HTld),

murine Tld-like (MTll) and Drosophila Tld (DTld) and Tld-related 1 (DTlrl).

I

I Figure 2. XBMPl and Xld are secreted glycoproteins. (A) In vitro translation oi xbmpl and xld.

Translation products of 80 (track 6) and 117 kD (track 3) were obtained for xbmpl and xld in a rabbit

reticulocyte lysate (RL), while translation products of 90 (track 4) and 135 kD (track 1) were obtained

XXlll for the same mRNAs in 2i Xenopus cell-free extract (E-). Since translation products of only 78 (track

5) and 117 kD (track 2) were obtained in the Xenopus extract in the presence of a tripeptide inhibitor of glycosylation (E+), both proteins must enter the endoplasmic reticulum and be glycosylated. The f I lower Mr for XBMP1 in track E-k relative to that in track RL is probably a consequence of signal peptide removal, this is not always evident by SDS-PAGE as can be seen for Xld under the same i conditions. Translation products were separated on a 7% SDS-polyacrylamide gel. (B) Western blot analysis of myc-tagged XBMPl (tracks 3 and 4) and Xld (tracks 5 and 6) in Xenopus oocytes (O) and culture media (M). The equivalent of 1/2 an oocyte of cell extract and the culture media for 4 oocytes were separated on a 10% polyacrylamide gel. Although most of the protein is retained by the oocyte

(tracks 3 and 5), both proteins are also secreted into the culture media (tracks 4 and 6). No signal could be detected in uninjected oocytes and culture media (tracks 1 and 2).

Figure 3. Temporal and spatial expression of xb m p l and xolloid in Xenopus embryos. (A)

Northern blot analysis of poly (A)4- mRNA (2pg/track) from 2 to 8-cell (tracks 1 and 3) and stage 28 embryos (tracks 2-4). The blot was initially probed with x b m p l, stripped and then reprobed with xolloid. Whereas three transcripts of approximately 5.6, 4.5 and 2.4 kb were detected with the xb m p l probe at all stages tested (tracks 1 and 2), only a single transcript of approximately 5.0 kb was detected with the xolloid probe in early cleavage stages through to neurula stages (tracks 3 and 4). i J Two weaker bands of approximately 3.5 kb were also detected at stage 28 with the xolloid probes

I (track 4). (B) RNase protection analysis of total RNA (10|ig/protection) isolated from indicated

I stages of Xenopus development using probes for either xbm pl or xld. Each protection also contained

I probe for the ubiquitously expressed ornithine decarboxylase {ode) gene. Stages shown are 16-cell

I (stage 5), early blastula (stage 7), late blastula (stage 9), early gastrula (stage 10), late gastrula (stage

I 12), neurula (stage 16), late neurula (stage 20), early tailbud (stage 25), tailbud (stage 30) and tadpole (stage 40). (C) RNase protection analysis of total RNA (10p.g/protection) isolated from embryo

fragments at the indicated stages using probes for either xb m p l or xolloid. Each protection also

contained probe for the ubiquitously expressed ornithine decarboxylase (ode) gene. The fragments

analysed were: AP = animal pole; VP = vegetal pole; DMZ = dorsal marginal zone; VMZ = ventral

XXIV ^marginal zone; LMZ = lateral marginal zone. Head and Tail are self explanatory, Dorsal = dorsal abdomen; Ventral = ventral abdomen.

Figure 4. Whole mount in situ hybridisation analysis of xbmpl and xolloid in Xenopus embryos.

(A) Early gastrulae (stage 10) hybridised with a probe for xbmpl, the left embryo is viewed from the animal pole (AP) and the right embryo from the vegetal pole (VP). Transcripts are uniformly expressed in the animal hemisphere but cannot be detected in the vegetal hemisphere, failure to detect vegetal transcripts can be quite common using this whole mount in situ hybridisation technique. (B)

Tailbud stage embryo (stage 28) showing widespread expression of xbmpl, transcripts are most abundant in the head and dorsal axis. (C) Neurula stage embryo (stage 17) showing punctate expression of xld in two dorsal patches flanking the posterior neural plate (NP). Anterior (Ant) is to the left and posterior (Post) to the right. (D) Tailbud stage embryo (stage 28) showing expression of xld at distal tip of the tail, flanking the posterior endoderm and in dorsal parts of the brain. The arrow indicates expression of xld in the caudal hindbrain. (E) Transverse section through the hindbrain showing expression of xld in the dorsal ventricular layer posterior to the otic vesicle (see arrow in D).

NT = neural tube; No = notochord; So = somite. (F) Transverse section through the tip of the tail

(stage 28) showing xld expression in the inner layer of the epidermis. D = dorsal; V = ventral.

! Figure 5. Overexpression of xbmpl or xolloid results in anterior defects. (A) Control embryo i injected with 2ng of Axbmpl RNA into each dorsal blastomere at the 4-cell stage. (B) Embryo

, injected with 2ng of xbmpl RNA into each dorsal blastomere at the 4-cell stage exhibiting reduced head structures and a shorter anteroposterior axis. Note the continued presence of blood in the ventral blood island (Bl). (C) Embryo injected with 2ng of xbmpl RNA into each ventral blastomere at the 4- I ; cell stage exhibiting a tail-like structure (T2) emerging from the proctodeum. Note the presence of melanocytes in the tail-like structure, although histological analysis showed that it contained no dorsal axial tissues. (D) Embryo injected with 2ng of xld RNA into each dorsal blastomere at the 4-cell stage exhibiting reduced head structures and a shorter anteroposterior axis. Note the continued presence of blood in the ventral blood island (Bl) and the enlarged proctodeum (P). (E) Embryo injected with 2ng

XXV of xld RNA into each ventral blastomere at the 4-cell stage, note the normal morphology apart from an

abnormal accumulation of blood (BL) in the tail.

Figure 6. Dorsal axial defects in xbm pl injected embryos. (A) Transverse section through the tmnk

of a control embryo injected with 2ng of A xbm pl RNA into dorsal blastomeres at the 4-cell stage,

note the normal arrangement of notochord (No), somites (So) and neural tube (NT). (B) Transverse

section through the trunk of an embryo injected with 2ng of xbm pl RNA into dorsal blastomeres at

the 4-cell stage, note the absence of a notochord and fusion of the somites (So) across the midline.

Figure 7. Inhibition of dorsal morphogenetic movements in activin treated animal caps. (A)

Uhinjected animal caps do not elongate in the absence of human activin A. An identical result was

obtained for animal caps injected with 3ng of either Axbm pl or m m pl RNA. (B ) Uninjected animal

caps elongate when incubated in the presence of human activin A. An identical result was obtained for

animal caps injected with 3ng of either Axbm pl or m m pl RNA and incubated in the presence human

activin A. (C) Animal caps injected with 3ng of xbm pl RNA fail to elongate when incubated in the

presence of human activin A. (D) Animal caps injected with 3ng of xld RNA fail to elongate when

incubated in the presence of human activin A. Embryos were injected at the 2-cell stage with the

indicated RNA, 1.5ng per blastomere, and animal caps isolated from mid-blastulae (stage 8). Animal

caps were incubated in the absence (A) or presence (B-D) of human activin A until sibling embryos [ had reached stage 17. I Figure 8. XBMPl and Xld ventralise dorsal mesoderm in a BMP dependent manner. (A) RNase

protection analysis showing that XBMPl and Xolloid ventralise dorsal mesoderm induced in animal

: caps by activin. Total RNA isolated from the same batch of animal caps was analysed with all of the

I indicated probes; MS-Actin = a-actin, C-Actin = cytoskeletal actin, a-globin = a-globin, ODC =

ornithine decarboxylase. Embryos were injected at the 2-cell stage with 3 ng of the indicated RNAs,

1.5 ng/blastomere. Animal caps were isolated from mid-blastulae (stage 8) and incubated until sibling

embryos had reached stage 25. (B) RNase protection analysis of activin treated animal caps showing

that a dominant-negative receptor for BMP2 and BMP4 (tBR) blocks the ventralising activity of both

XXVI XBMPl and Xolloid. Total RNA was isolated from injected animal caps treated with activin and probed for muscle-specific actin {a-actin). Embryos were injected at the 2-cell stage with 1.5 ng of the indicated RNAs, 750 pg/blastomere. Animal caps were isolated from mid-blastulae (stage 8) and incubated until sibling embryos had reached stage 25.

xxvii Table 1. Overexpression of BMPl and Xolloid reduces the development of anterior structures.

RNA AD DAI TD EP GD N

None 0% 5.00 0% 0% 1% 267

Axbm pl 10% 4.90 0% 0% 25% 101

xbm pl 70% 3.84 10% 40% 20% 103

xld 69% 3.68 0% 70% 8% 118

m m pl 3% 4.97 0% 0% 5% 41

Embryos were injected at the 1-2 cell stage with 3ng of the indicated RNAs and scored at stage 30. AD = Anterior Defects. DAI = Dorso Anterior Index. TD = Tail Duplications. EP = Enlarged Proctodeum. Embryos scored as TD and EP were also scored for anterior defects and are included in the DAI score. GD = Gastrulation Defects; in these embryos the blastopore failed to close resulting in a spina bifida-like phenotype, they were not scored for anterior defects and are not included in the DAI score. N = total number of embryos scored, including GD embryos.

XXVI n Table 2. Dorsal injections give more extreme anterior defects than ventral injections

RNA Blastomeres ADDAITDEP GDN

None None 3% 4.92 0% 0% 1% 284

Axbm pl Dorsal 5% 4.95 0% 0% 18% 26

Axbm pl Ventral 0% 5.00 0% 0% 6% 33

xb m p l Dorsal 30% 4.53 0% 62% 8% 185

xb m p l Ventral 6% 4.94 13% 37% 7% 153

xld Dorsal 68% 3.68 0% 90% 22% 96

xld Ventral 13% 4.59 0% 82% 33% 51

The indicated RNAs were injected into either dorsal or ventral blastomeres at the 4-cell stage. Both blastomeres were injected with 2ng of RNA per blastomere, the resulting embryos were scored at stage 38. Abbreviations are the same as in Table 1. Embryos scored as TD and EP were also scored for anterior defects and are included in the DAI score.

XXIX Figure 1

A)

signal Peptide Proregion

X l d MSCGSPOVMMTLWTLTCVGLILLGATRLSLGLDYDLESFDYLMEDNPEEFDYKDPCKAAAYWGDIALDEDDLKW X B M P l .PW.GSPPLLLWA. . AGMLHLGQGQEF. DYSYD.E . W E .SI ...... F ...... E. .AN

IFKNKSNDLRNTRHNQTHPTTDNFSEKLGTGSQNETSSNLNSKKVKKGSRLKLLIAEKAATETNSTFQVQTSND FKVARIV..TKHTISTVSGAAS.S .RPQRGRRTRKQRRRS------

Metalloprotease

RVRRAATSRTERIWPGGIIPYAIAGNFTGTQRAIFKQAMRHWKKHTCVTFVERTDEESFIVFTYRPCGCCSYVG .E...... P..V..D.V...V.S...S.S..... R ...... E ...... L .....D.Y.I......

RRGGGPOAISIGKNC DK F G I W HELGHWGFWHE HTRPDRDEHVSIIRENIOPGOE YNFLKMEPGEVSSLGET Y ...... I ...... D ...... E. .E.I----

CUBl

DFDSIMHYARNTFSRGVFLDTILPRRIDTSVRPTIGORIRLSQGDIAQAKKLYKCPACGETLODSSGNFSAPGY ...... I ...... KYDVNG. ..P----T...S...... R ...... Q ----S..F

PSGYPSYTHCIWRISVTPGEKIILNFTTMDLFKSRLCWYDYIEIRDGYWRKAALLGRLCGDKLPDPIISSDSKL * N *.S A ...... SL..YR...... V...F.K..P.R.«F....I.E.... TE . R .

CUB2

WIEFRSSSNILGKGFFAAYEAICGGDIKKDSGQIQSPNYPDDYRPAKECIWKITVSEGFLVGLSFQAFEIERHD ...... WV ____Q.V...L____V .....H ...... N.A.V..LS..... H..I...S......

EGFl

NCAYDYLEVRDGFSEDHALIGRFCGYEKPEDIKSTSNKLWIKFASDGSINKAGFSANFFKEMDECSRPDNGGCS S ...... I...S. .SSP...... D..D....ST.Q..V..V...... L.Y...V______N ____E

QRCVNTLGSYKCVCEPGFELTADKKSCEAACGGFITQLNGTITSPGWPKEYPTNKNCVWQWAPAQYRISLQFE ...... A.D..Y..GP...... K...S...... P ...... L...T...... K.D

EGF2

VFELEGNDVCK Y D YLEIRSGLSSESKLHGKFCGPEKPEVITSQGNTVRIEFKSDNTVSKKGFKANFFSDKDEC S Q T ...... FV.V T . D ...... T.L.A.....Y.NM...... Q .....E.KNNV

CUB4

KDNGGCQHDCVNTFGSYICQCKNGFILHENGHDCKEAGCEQKLLNAEGTISSPNWPEKYPSRKECTWDISVTAG QKLQQQNEVNRGQQQNQAPKRGRPRMRLRTMKKTRPP

HRVKLVFTDFEIEQHQECAYDHLELYDGPNGKAAILGRFCGSKEPSPWASTNNMFLRFYSDASVQRKGFQAKY

CUB5

SPECGGRLKAEIQTNDIYSHAQFGDNNYPVQSNCEWVIVAEDGYGVELIFQTFEIEEESDCGYDYMEVYDGYDS

TAPRLGRYCGSGPPEEMYSAGDSIMIRFHTDDTINKKGFHGQYTSTKFQDALHMRRK

XXX Figure 1 B)

Xld YKDPCKAAAYWGDIALDEDDL (52-72) XBMPl E.. (44-64) HBMPl ...... FL ... (51-71) MBMPl MTLL ...VF...... DE . . (50-70) SBMPl vas....SGFL... (33-53) DTLRl DM. . ...GGFM... ..P.G.S (146-166) DTLD PESE DFDFKEQPEDFFGILD (33-53)

C) XBMPl

Xolloid

XBMPl 85 72 76 83 79 MTll 87 77 78 70 80 83 66 71 XTld 85 71 75 83 80 80 54 69 HTld 85 73 78 73 82 83 64 76 DTlrl 67 41 56 53 46 65 42 41 DTld 54 30 50 55 44 65 44 36

PrePro Region 0 CUB Domains Metalloprotease ^ EGF-like

XXXI (Q CD to 00 00 m

NO NO

Ci Ci I Uninjected + Un injected m -j- Uninjected 1 ro + Xld X en # # -j. ABMP1 en DJ Ci + Xld+tBR m + BMPl m 11 + BM Pl -I- Xolloid 1 t t t

+ BM P1+tBR + MMP1

T| o in

AP ■n (O NO • VP t VMZ T T # f DMZ 2-8 cell en m m f LMZ I Stage 28 AP œ. en i # 2-8 cell 0 VP ! it Stage 28 00 1 Head I i:7 t t 1 %% %7îv CD * ♦ 0 Tail 0 Dorsal

1 Ventral B Fig. 5

&'

Ant HP P ost

Bl

$ k 'M

Bl

B *\

xxxni