NOTE TO USERS

The original manuscript received by UMl contains pages with slanted print and margins that exceed the guidlines.

Pages were microfilmed as received.

This reproduction is the best copy available

Gene Trap Studies of the Mouse Genome and Embryonic Development

Mehran Sam

A thesis subrnitted in conformity with the requirements for the degree of Doctor of Philosophy

Graduate Department of Molecular and Medical Genetics University of Toronto

O Copyright by Mehran Sam 1998

National Library Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliogcaphic Services services bibliographiques

395 Weilington Street 395, rue Wellington Ottawa ON K I A ON4 Ottawa ON K1A ON4 Canada Canada

The author has granted a non- L'auteur a accordé une Licence non exclusive licence allowing the exclusive permettant à la National Lïbrary of Canada to ~ibfiothe~ue nationale du Canada de reproduce, loan, distri'bute or seil reproduire, prêter, distri'buer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic fonnats. la forme de microfiche/film, de

reproduction sur papier ou sur fomat électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts £kom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

Gene Trap Studies of the Mouse Genome and Embryonic Developmeat Mehran Sam Doctor of Phiiosophy, 1998 Graduate Department of Molecular and Medical Genetics University of Toronto

In recent years, gene trap insertional mutagenesis strategy has been developed and

successfully used to disrupt and study developmeotally regulated genes in the mouse

embryos. Using a gene trap approach, two independent mouse embryonic stem (ES) cell

clones were idenWied and investigated for developmental expression pattern, dismpted

gene sequences and phenotypic effects. The strategy made use of a promoterless lac2

reporter gene to disrupt and mark genomic loci and to determine the expression pattern of

genes. Il93 and R140 gene trapped ES cell clones were isolated in a retinoic acid (RA)

response screen and were then characterized by myself as described in this dissertation.

The RA-induced Dr repetitive element and RA-induced Aquarius gene were

cloned from the Il93 clone. Dr elements are highly similar in sequence, show peculiar

structural features and are expressed in a spatially resericted manner. The expression of

Dr transcripts seems to be part of an ES ce11 differentiation prograrn induced by RA.

Aquarius is a novel gene with weak similarity to RNA-dependent RNA polymerases

(RRP) of the murine hepatitis viruses and it contains an RRP motif. During embryonic

development. Aquarius is expressed in mesoderm, neural crest and neural crest-derived

tissues. and in neuroepithelium. Mice homozygous for the Aquarius gene trapped allele

are viable, are normal in size and weight, and breed normally.

The gene trapped locus, designated rnyocyte maintenance (mym), was iden tified

and characterized in the R140 clone. myrn is expressed in heart, limbs, gut, meninges and

choroid plexus. The disrupted mym locus was traasmitted through the mouse germ Iine

on to different genetic backgrounds. Homozygous mutant embryos died during

midgestation due to heart failure. Myocardiai wall and trabeculae were found to be

diminished in thickness and size, and were hypoplastic. In mutant myocytes the

organization of actin-myosin filaments and cell junctions were disrupted and the

expression of myosin light chah isoforms was deregulated. mym is required for the

maintenance and organization of heart myocytes during midgestation development. The

results presented in this thesis indicate the facility of gene trapping to identify and study

developmentally regulated rnammaiian genes and genetic elements.

iii

1 would like to express my appreciation and hank my supervisor, Alan Bernstein, for

providing a rich and fertile environment to l e m and to do science. 1 also thank the

mernbers of my graduate committee, Henry Krause and Janet Rossant, for their help and

guidance throughout this work The members of the Bemstein lab, including postdocs,

graduate students and technicians also provided me with guidance and assistance

whenever 1 needed them. For this and for their friendship. 1 thank them ail. 1 take with me

many valuables from this period of my studies including many lasting mernories. I

dedicate this work to my dear wife, Azita, and to my parents.

Attribution of Data

1 am responsible for ail the expcriments reported in this thesis except the following. The

original retinoic acid gene trap screen and the isolation of gene trap clones were

perfonned by Wolfgang Wurst and Lesley Forrester. The blastocyst injections and

generation of chimeric anirnals were carried out by Sandra Tondat. FISH analyses were

perfomed by Henry Heng. In Chapter 3, RNA in situ hybridization was performed by

Michael Klüppel. and the whole mount in situ hybridization by Ou Jin. Transmission

electron rnicroscopy in Chapter 4, was done by Douglas Holrnyard and myself.

Table of Contents

. . .................................................................................................................... ABSTRACT il ........................................................................................................... Acknowledgments iv .......................................................................................................... Attribution of Data v

Table of Contents ............................................................................................................ vi ... ............................................................................................... List of Tables and Figures viii

Chapter 1 ........................................................................................ Historical Background 1

Genetics of mode1 organisrns ............................................................................. 1 The mouse as a mode1 for rnammalian development ........................................ 6

............................................................................ Chimeric mice ..................... .. 6 Mouse embryonic stem celis ......................................................................... 8 Transgenic mice .................................................................................................. 9 Gene targeting by homologous recoa bination ................................................. 11 Gene trap insertional mutagenesis .................................................................... 13 Retinoic acid response gene trap screen ............................................................. 20

Chapter 2 A Novel Family of Repeat Sequences . n the Mouse Genome Responsive to Retinoic Acid .................................................................................................. 23

........................................................................................................ ABSTRACT 24 INTRODUCTION .............................................................................................. 24

........................................................................ MATERIALS AND METHODS 26 Gene trapped embryonic stem (ES) ceU lines and tissue culture ............ 27 5' RACE, PCR DNA amplification and cloning ................................. .... 27 cDNA cloning and sequencing ............................................................... 28 Retinoic acid induction, RNA preparation and Northern blot

................................................................................................... analyses 28 RNA in situ hybridization ................................................................... 29 Genomic screening and Southem blot analysis .................................. 29 FISH ........................................................................................................ 30

..................................................................................................... RESULTS 3 0 5' RACE-PCR cloning of fusion Dr-1 transcripts from three

........ independent gene-trapped ES ce11 lines induced with retinoic acid 31 Molecular cloning and nucleotide sequence analysis of Dr-2 and

............................................................................................ Dr-3 cDNAs 32 Dr-containing transcripts are induced by retinoic acid in ES celis ......... 37 Developrnental expression of Dr repeats ................... ... ..................... 37 Genomic representation and organization of Dr sequences .................... 43

................... DISCUSSION .. ............................................................................ 46

Chapter 3 Aquarius. a novel gene isolated by gene trapping with a RNA-dependent

....................................................................................... RNA polymerase motif 50 ............................... ............................. ABSTRACT ... 51

.............................................................................................. INTRODUCTION 51 MATERIALS AND METHODS ..................... .. ....................................... 54

Gene trapped embryonic stem (ES) ce11 lines. lac2 staining and germ line transmission ....................................................................... 5 4 Mouse genotyping and F2 breeding in different backgrounds ............... 55

RACE . PCR amplification. cloning of the 5' Aqr exon. and genomic PCR .......................................................................................... 55 cDNA cloning, sequencing, GenBank and intemet utiiities ................... 56 Retinoic acid induction . RNA preparation. Northem blot analyses, d R P R ............................................................................................ 58 RNA and whole mount in situ hybrïdization ......................................... 59 FISH .............................. ., ...................................................................... 59

RES ULTS ......................................................................................................... 6 0 ........................................ Cloning of the RA-responsive Aquarius gene 60

Aquarius has weak similarity to RNA-dependent RNA polymerases (RRPs) ................................................................................ 64

.................. Gem Line transmission of the gene trapped Aquarius locus 66 .............. Embryonic expression in neural crest and mesodemal tissues 76

........................................... Chromosome localization by FISH analysis 82 DISCUSSION ...................................... ,, . 88

Chapter 4 The Novel Gene Trapped mym Locus 1s Required for Midgestation Heart

...................................................................................................... Development 9 2 ABSTRACT' ........................................................................................................ 93 INTRODUCTION ............................ ,, ......... ... 93 MATERIALS AND METHODS ....................................................................... 99

Gene trapped embryonic stem (ES) ce11 lines, lac2 staining and gerrn line transmission .......................................................................... 99 Mouse and embryo genotyping and F2 breeding in different

............................................................................................ backgrounds 99 RACE, PCR ampiifkation, cloning of the 5' mym exon ........................ 99

........................................................................ Genomic PCR, and FISH 100 ....................................................................... RNA in situ hybridization 101

............................ ......................... Scanning electron microscopy .... 101 ................................................................................................................. Results 102 RI40 gene trap ES cells and race-pcr products ....................................... 102 Genomic disruption and chromosome localization by FISH analysis ................................................................................................... 103

.............................. Lac2 expression pattern of the mym trapped locus 103 ............................... Germ line transmission and analysis of F2 progeny 111

............................................................ 129 males show reduced feaility 111 Homozygous embryos die during midgestation ........... ... ................... 116

.................................................. mym is requhd for heart development 121 Expression of myocyte-specinc genes in mym mutant hearts ................ 127

.............................. Failure in myocyte maintenance ......................... ... 127 DISCUSSION ..................................................................................................... 133

Chapter 5 ........................................................... .......................... Concluding Remarks ... 138

...................................................................... Summary and General Discussion 139 .................................... ..................................................... Dr repeats .... 140

.................................................................................................. Aquarius 141 ......................... mym .. ....................................................................... 143 ......................................................................................................... Conclusions 145

.................................................................................. Literanire Cited .......................... ,.. 147

vii

List of Tables and Fipres

Figure 1 .1 ........................................................................................................................ 15 Figure 1.2 ................................................................................... ............................ 22

Figure 2.1 ....................................................................................................................... 33 Figure 2.2 ....................................................................................................................... 35 Figure 2.3 ........................................................................................................................ 39 Figure 2.4 ......................................................................................................................... 41 Figure 2.5 ....................... ... ....................................................................................... 44

Figure 3.1 ............................ ... ....................................................................................... 62 Figure 3.2 ......................................................................................................................... 68 Figure 3.3 ....................... ... ......................................................................................... -72

....... Figure 3.4 .......................... .. .. .......................................................................... -74 Figure 3.5 ..................... ....... ......................................................................................... 78 Figure 3.6 ....................... .. ........................................................................................... -80 Figure 3.7 ......................................................................................................................... 84

............................................................................................ Figure 3.8 ....................... .. 86

Figure 4.1 ..................................... ......................................................................-. 1 0 5 Figure 4.2 ....................................................................................................................... 107 Figure 4.3 .................................................................................................................... - 1 09 Figure 4.4 .......................... ... ................................................... ... ................................ 112 Figure 4.5 ............................. ...... .............................................................................. 1 14 Figure 4.6 ................................................................................................................... 1 19

........................................................................................ Figure 4.7 .......................... .. 1 23 Figure 4.8 ...................................................................................................................... 125 Figure 4.9 ...................................................................................................................... 1 29 Figure 4.10 ................................................................................................................... 131

Table 3.1 ......................................................................................................................... -70 Table 4.1 ............................ ,., ......................................................................................... 117

viii

Chapter 1: Historical Background

As biologists. we want to understand the living world in ali its splendor and wonder. It is

perhaps Our awe for the breathtaking beauty and variety of iife forms that drives us in our

desire to know more. Or it is perhaps our fortitude that we live just a short while, yet our

appetite is to fathom so much. Every time we leam something about a living organism,

we have learned something about ourselves. This thesis is about using genetics and gene

trapping in the mouse to understand something about mammalian development. Chapter

One provides a short overview of the developmental genetics of model organisms and

then briefly describes the historical development of the techniques and fields of studies

that have made gene trapping in the mouse useful and possible. This chapter is not

intended to be comprehensive. rather only to provide a general overview for an

appreciation of the scope of the field.

Genetics of model organisms

The most powerful tool in the arsenal of genetic analyses is the study of mutations. By

mutational analyses we gain knowledge about the function of genes even before we have

any information about the genes involved. The recent explosion in our understanding of

developmental processes at the molecular level is largely due to the extensive mutational

analyses, performed or in progress. in several model organisms, notably Saccharomyces,

Arabidopsis, Caenorhabditis, Drosophila, laboratory mice and most recently in the

vertebrate Danio r e h (zebrafish).

The most widely used eukaryotic microorganism for genetic analyses is the yeast

Saccharomyces cerevisiae (Guthrîe and Fink. 199 1). The rapid growth and ease of mutant

isolation makes Saccharomyces an ideal genetic system in which to unravel ce11 biology,

ce11 division, metabolism and chromosome structure and function. In addition, the stable

existence of yeast in both haploid and diploid States allows for facilitated analysis of

many mutations that may not be otherwise possible in other organisms . Mutagenesis

techniques such as gene targeting, disruption. and replacement, as well as allele rescue,

complementation and transformation are well established in yeast and make this organism

the most powemil system for biological studies of eukaryotes. In the late 1970s.

reproducible transformation of yeast was achieved by intmducing purified exogenous

DNA. Today, yeast remains the only eukaryotic organism that can be transformed

conveniently with synthetic oligonucleotides, allowing direct production of numerous

gene alleles which c m in tum be snidied in the whole organism. This is a powerful

molecular tool and a highly desirable one in any organisrn since it makes it feasible and

convenient to investigate the multiple domains and functions of any gene product. In

addition, yeast development of mating types, polarized morphogenesis and their response

to mating pheromones have been investigated as a model to understand eukaryotic

development at the molecular level (Chant 1996; Leberer et al., 1997). The complete

yeast genome sequence is now available as a fust complete eukaryotic genome which

will be used as a reference point for other eukaryotic genomes

(http:l/speedy.mips.biochem.rnpg.delmips/ye The complete set of expressed genes,

the transcriptome, from the yeast genome is also now under investigation and has made

feasible genome-wide expression studies in eukaryotes (Velculescu et al., 1997).

Arabidopsis thliana is a srnall flowering plant. Its diploid genetics, rapid growth

cycle and small genome size has made it the model for a plant genome project and for

developmental molecular genetic studies. Large scaie insertional mutagenesis and cDNA

and genomic mapping and sequencing are underway (Azpiroz-Leehan and Feldrnann,

1997; Delseny et aL. 1997). Approximately 4000 mutants have been so far identified and

more than 40 genes isolated. These efforts have ken instrumental in advancing Our

knowledge of the physiology. biochemistry and development of plants. An efficient

insertional mutagenesis enhancer and gene trap system has also been developed in

Arabidopsis that permits gene identification based on expression patterns dunng

development (Sundaresan et al.. 1995).

Since the 1960s. the nematode worm Caenorhubditis elegans has been utilized as

a mode1 organism to study animal development and behavior (Riddle et aL, 1997). Rapid

life cycle, small size, ease of cultivation, inbreeding by the self-fertilizing hermaphrodite

and crossing with males ali offers great potential for genetic analysis in this organism.

Anatomical simplicity ( l es than 1 0 ceils, the developmentai lineages of ail of which

are now known), large number of progeny and small genome (60% of the genome

sequence complete. http://www.sanger.ac.uk/-sj/C.elegansSHorneehtml) has contributed

to the astonishing progress we have made in understanding the genetics and development

of C. elegans. The ultimate goal is to determine the role of each gene in the development

and function of this organism, a goal that now seems within reach. Induced and target-

selected mutagenesis in C. elegans are now routine and have helped dissect genetic

pathways and functionai units in the worm genome. To date, 10% of the 15,000 total

genes estimated by the worm genome sequencing project have been genetically

identified. Mutations in about half of these genes display visible phenotypes. and about

half are essential genes (have lethal alleles). Lethality may be due to a developmental

blockage from egg to larval stages, steriiity or maternakffect. It is estimated that nul1

mutations in up to 50% of C. eleganî gens may yield no obvious phenotype either due to

redundant genes or pathways, or lack of specific environmental conditions necessary to

reveal a phenotype. On the other hand. while some essential genes have 'housekeeping'

functions, i.e. are necessary for generai ce11 processes such as intermediary metabolism.

mosaic screens indicate the majority of essential genes function in only specific ce11

lineages. A promoter trap screen has been performed in C. eleguns in which a worm

genomic DNA library was cloned upstream of a lac2 reporter gene and used to create

transgenic w o m (Hope 199 1). Tissue specific expression of lac2 wodd then infom of

the presence of functional regdatory elements upstrearn.

Earlier this century, work with Drosophiln rnelanogaster established many of the

major principles of genetics such as the chromosome theory of heredity, multiple

allelism. non-disjunction, chromosome aberrations, chernical and radiation mutagenicity,

and gene mapping (Ashburner 1989). Wilkins has divided the history of developmental

genetics into three fairly distinct phases based on the prevailing ideas and approaches in

each phase (Wilkins 1993). The 'classical' p e n d from the fmt decade of this century to

about 1960, was dominated by the study of mutant phenotypes to deduce the roles of

genes in developmat. In the second period, starting in 1961 and lasting approximately

two decades, molecular concepts and various techniques of clonai analysis (tracking ce11

lineages by the presence of distinctive genetic markers) gained influence. The new era of

developmental genetics, i.e. determining at the molecular level all the genes involved in a

particular developmental process, started with the classic saturation mutagenesis

approach in Drosophila (Nusslein-Volhard and Weischaus, 1980), and was transformed

by recombinant DNA technoiogy and other molecular biological techniques. One of the

main resources of the developmental biologists t d a y is the large mutant collections of

h i t flies collected during the period of classical developmental genetics.

Drosophila was one of the f i a t eukaryotic organisms in which vectors with

dominant drug-resistant marken were used to gewticdy select for transformants (Steller

and Pirrotta, 1985). The selectable marker used in this experiment was the neomycinr

(neo) gene of bacterial Tt15 which confen resistance to the powerful inhibitor of protein

synthesis, the arninoglycoside G418 (Jorgensen et al., 1979). Flies transformed with the

neo gene. under the control of a suitable promoter were resistant to G4 18. The lac2 gene

of E. coli, coding for the enzyme ~galactosidase (kgal) and assayed by histochemistry,

was also one of the fmt markers used in Drosophila as a reporter of gene activity (Lis et

al., 1983). Finally, a chimeric ZacZ-P element gene under the control of the weak P

element promoter was used in Drosophila in the fmt enhancer trap experiments (O'Kane

and Gehnng, 1987). The integration of the enhancer trap element near a host sequence

upregulated its transcription and led to an increase in lac2 expression in a tissue-specific

manner.

The zebrafish, Danio rerio, has recently become a vertebrate mode1 system

because it is susceptible to both classical embryological manipulations and extensive

genetic analysis. Because eggs are fertiiized extemally and the embryo is transparent,

vimially every ce11 c m be examined in the Live embryo by light microscopy. Zebrafish

offer several advantageous features for genetic anaiysis. compared to other vertebrates,

including a short gestation t h e , and the ability to make genetic mosaics, haploids and

ttiploids (Felsenfeld 1996). These features made it possible and encouraged two

independent groups to undertake large scde chemical mutagenesis screens in zebrafish

with remarkable results (Driever et al, 1996; Haffter et al, 1996). A few thousand

mutations have been identified and several hundred of them characterized. Mutant

phenotypes have k e n identified with matemal effects. in gastmlation and ceil movement,

body axes, notochord, CNS, neural crest, blood. heart and intemal organs, eye and

behavior (http://zebrafsh. mgh.harvard.edu/database.hmil). Although positionai cloning

of these mutations will be labor-intensive, it is achievable as tools for genetic analysis,

mapping and cloning become available (Koapik et al, 1 996). Insertional mutagenesis and

rapid cloning of essential genes using a retroviral vector have d s o been successfully

attempted in zebrafish (Gaiano et al., 1996), which could circumvent positional cloning . Furthemore, gene interactions can be revealed in zebrafish by enhancer/suppressor

screens and conditional mutations (Johnson and Weston, 1995). Thus, zebrafsh has great

potential in enhancing our understanding of rnolecular and genetic mechanisms

underlying vertebrate development.

Avian and amphibian systems

vertebrate development through classical

will also continue to provide insights into

embryological manipulations. while the mouse

will continue to be the mode1 of choice in understanding human genetic disorders and

behavior. Creating nul1 as well as subtle mutations in known genes using the rnouse

embryonic stem ceii techology will also continue to provide insights about gene

function, regulation and interactions in mammaüan system.

The mouse as a mode1 for mammaüan development

The mouse has been the subject of embryological and developmental studies for almost a

century. Its smaii size, relative ease of maintenance and generation of inbred strains of

genetically homogeneous mice has made this organism the mode1 of choice in many

laboratories. Early work by Snell, Little, Castle and othea established the basis for future

studies in the fields of embryology, genetics, cancer, and transplantation (Russell 1985).

Today, cornplicated genetic and developmental problerns, i.e. genetic interactions,

modifier genes and complex genetic traits, can be addressed in the mouse due to advances

in genetic mapping and analysis such as the use of microsate1Iite repeats as genetic

markers (Dietrich et al., 1992). In addition, the burst in Our knowledge of mamrnalian

development is attributable to three inter-dependent technologicai advances in the mouse:

i) generation of chimeric animals, ii) establishment of mouse embryonic stem cells, and

üi) production of transgenic mice.

A rnarnmalian animal composed of genetically dissimilar cells is called a 'mosaic' if it is

derived from a single zygote, and a 'chirnera' if it is denved from different zygotes (Slack

199 1). In the early 1960s, chimenc mice were generated by aggregation of two eight-cell

stage embryos (Mintz 1962; Tarkowski 1961). The chimeras were generated to

investigate ce11 movement. assortment. and to a limited extent lineage development by

using genetic (lethal mutants). radioisotopic and cytologicai markers (Rossant 1987).

Sexshromosome mosaicisrn and hermaphroditism as a mode1 of sexual disorden were

also investigated in XXKY chimeras (Tarkowski 1965).

In 1968, Gardner succeeded in creating mouse chimeras by injecting cells into

blastocyst stage embryos, thus making it possible to controi the extent of chimerism by

the number of cells injected (Gardner 1968). Gardner detected the extent of chimerism by

coat color. Pluripotent embryonic stem cell lines were established in culture and chimeras

were generated by injection of these cells into blastocysts, followed by blastocyst transfer

into pseudopregnant recipient mice.

Today we can create mouse chimeras by simple aggregation of pluripotent

embryonic stem cells with morulae-stage embryos (Wood et al.. 1993). This new

technique is less cumbersome, more economical and just as efficient as blastocyst

injection. It is also possible now to generate mice and embryos that are derived

completely frorn embryonic stem cells by aggregating stem cells with developmentally

compromised tetraploid embryos (Nagy et al., 1993a).

ES-diploid and ES-tetraploid aggregate chimeras are of great value in direct

anaiysis of dominant negative, gain of function and homozygous mutations that may

interfere with normal development or genn h e transmission. For exaii-ple, the effect of a

mutation may not be as revealing when al1 cells are mutant as when mutant cells are

intermixed with wild type cells in a chimera. Mutant cells in a wild type environment can

reveal whether the mutation is lethai for al1 cells or has only a temporal or tissue-specific

effect. Furthemore, chimeric analysis can determine whether the gene product acts ce11

autonomously or whether neighboring wild type cells can rescue the mutant phenotype

(Beddington 1992). Chimeras are also a rich source of differentiated tissues and tissue

progenitors carrying genomic alterations (Chen et al., 1993).

Mouse embryonic stem ceïis

Plunpotent cells were known to be present in a mouse embryo until early post-

implantation stage; however, early attempts to culture these cells were unsuccessful. In

198 1. Evans and Kaufman established embryonic pluripotent cells in culture by direct in

vitro culture of blastocysts (Evans and Kaufman, 1981). These cells gave nse to

embryoid bodies and a complex of differentiated tissues in vitro, and to teratocarcinoma

in vivo, indicating their pluripotentiality. They carried a normal karyotype and generated

chimeric mice. In the same year, Martin established a s i d a r cell line from the imer celi

mass of early embryos (Martin, 1981). This cell culture was maintained in medium

conditioned by teratocarcinoma stem ceils.

Today several pluripotent embryonic stem (ES) cell lines capable of germ line

transmission have been established. These ceils proliferate on feeder layers of irradiated

cells, and in the absence of the feeder cells, differentiate into embryoid bodies. Leukemia

inhibitory factor can substitute for the feeder layer (Smith et al., 1988). ES cells can be

injected into blastocysts or aggregated with morulae stage embryos to create chimeras.

When the chimerism extends to the g e m cells, the genome of the ES cells, carrying any

genetic alteration, can be transmitted to the progeny of the chimeras, thus creating

transgenic mice with a desired genetic change (Robertson 1986). ES cells are now

routinely used to introduce mutations in the mouse germ line by homologous

recombination (Capecc hi 1 989).

As mentioned above, it is now possible by tetraploid aggregation to produce

totally ES cell-derived rnice, indicating the remarkable potential of cultured ES cells.

Howéver, in these experiments many ES cell-tetraploid aggregates die before reaching

term making the technique inefficient and infeasible for germ line transmission of a

genetic manipulation (Nagy et al., 1993a).

In vitro differentiation of ES cells depending on culture conditions allows access

to different developmental processes such as mesoderm commitment and neuronal and

hematopoietic development. In this system, experimental conditions can be controlled

and readily altered. and the amount of material available is neither as limited nor as

expensive as the in vivo system. The progressive differentiation of ES cells can be easily

followed by mo1ecu1ar marken and techniques such as cytohistochemistry or RT-PCR

(Wiles 1993).

Transgenic mice

Foreign DNA sequences can be stably inserted into the mouse genome to create a

transgenic animal. Transgenic mice can be created by microinjection, viral infection and

manipulation of ES cells in culture. In 1976, Jaenisch infected mouse embryos with

Molony Ieukemia virus and showed for the first time the incorporation of exogenous

DNA into the mouse germ line (Jaenisch 1976). In the next five years, several groups

achieved the integration of foreign DNA into the mouse genome through the

microinjection of DNA into the zygote pronuclei (Brinster et al., 1981; Costanthi and

Lacy, 198 1 ; Gordon and Ruddle. 198 1; Harbers et al., 198 1; Wagner et al., 198 1; Wagner

et al., 198 1). Such random integrations led to various genomic rearrangements such as

translocation, deletion, duplication and inversion. Before the advent of transgenesis,

investigators had been primady restricted to observation and description. Thus, the

ability to introduce at will genetic changes or new genes into animals represented a

dramatic advance.

To achieve transgene expression in differentiated adult cells, Brinster et ai. hised

the herpes simplex virus thymidine kinase gene with the rnetallothionein-I promoter

region (Brinster et al., 1981). The fusion gene was injected into fertilized eggs which

were then reimplanted into pseudopregnant rnice. The fusion gene was present in the

genome of the progeny mice and it was highly expressed in the liver and kidney. In one

of the fmt gene hinctional studies in transgenic mice, Lohler et al. examined the result of

insertional mutagenesis by the Moloney murine leukemia virus in the a l collagen gene in

the Movl3 mice (Lohler et al., 1984). The integration at the Movl3 locus resulted in a

recessive lethal mutation leading to midgestation embryonic lethality. The cause of death

was determined to be rupture of major blood vessels. There were also pathological events

in hemopoietic cells of the liver and necrosis of mesenchymd cells, indicating a

functioaal mie for al collagen in the normal development of these tissues.

in another transgenic experiment, the mouse mammary tumor v h s long terminal

repeat fused to the c-myc gene created an insertional mutation in the mouse genome that

led to a severe defect in limb patteming (Woychik et al., 1985). The inserted element

provided for a molecular link with the genetic control of limb formation and was used to

examine the integration site. The integration of the transgene deleted about 1 Kb of the

genomic sequences with no other gross rearrangement. The integration site was mapped

to the limb deformity (Id) region and in complementation tests was shown to be indeed an

d e l e of Id. These early transgenic experiments in which the mouse germ line was stably

transfonned by exogenous DNA are extensively reviewed by Palmiter and Brinster

(Pairniter and Brinster, 1986).

Subsequently, retroviral vectors were used to introduce transgenes into cultured

pluripotential stem cells for generation of chimeric mice and germ line transmission

(Gossler et al., 1986; Robertson et al., 1986). The proviral vector sequences containing

neo gene were stably transmitted through several generations and provided drug

resistance as weli as new chromosomal molecular markers for linkage studies. They also

caused insertional mutations. These and other similar experiments indicated the

feasibility of using the easily manipulatable embryonic stem ceils to produce transgenic

animals, opening up a whole new set of experimental approaches for in vitro selection

and screening for a desired genetic change in cultured stem cells before generating the

transgenic animal. Soon after, transgenes canying the E. coli p-galactosidase (la&)

reporter gene were introduced into the mouse genome and were detected with a

histochemical stain to foilow endogenous gene expression pattems and to study ce11

lineages in vivo (Allen et al., 1988; Kothary et al., 1988; Sanes et aL, 1986). Tissue-

specific and inducible promoter sequences were also successfuily used in transgenic mice

to direct spatio-temporal expression patterns of transgenes during mouse embryogenesis

(Kothary et aL, 1989; Rossant et al.. 1991). Today, it is possible to transfer intact yeast

artificial chromosome (YAC) DNA into transgenic mice enabling the analysis of large

genes and multigenic loci in vivo (Peterson et al., 1997).

Gene targethg by homoIogous recombination

In transgenic mice, gene integration into the genome is usually random, based on non-

homologous recombination, and expression is often unpredictable. The ability to target

genes to specific locations in the yeast genome by homologous recombination (Hinnen et

al., 1978) allowed for specifc gene inactivation, mutation, and correction, and prompted

investigaton to seek a simiiar technique in the mouse. Reconstruction of a mutant herpes

simplex virus thymidine kinase (tk) gene in mammalian cells was one of the earliest

indications that the genome of rnammalian cells can undergo homologous recombination

with exogenous DNA (Small and Scangos, 1983). Mutants of dominant selectable

markers like neo gene introduced into mouse or human cells also generated a functional

gene by homologous recombination (Kucherlapati et al., 1984). Homologous

recombination was also shown to correct a mutation in the selectable hypoxanthine

phosphoribosyl transferase (HPRT) gene and to inactivate a wild type HPRT gene in

mouse ES cells (Doetschman et al., 1987; Thomas and Capecchi, 1987).

In 1988. Mansour and colleagues developed a positive-negative selection method

that allowed for positive selection for the correct homotogous recombination events,

while at the same time selected against non-homologous events, thereby substantially

enriching for the recovery of targeted events (Mansour et al., 1988). Gene targeting is

now widely used in the mouse ES cells to understand gene function in the whole animal.

When the targeted gene is transmitted through the germ line of a chirneric animal, it can

be made homozygous by breeding and its phenotypic effects can be examined. Based on

the number of knockout mice published by 1995, it can be estimated that to date more

than 800 genes have been mutated by gene targeting in the mouse ES cells (Brandon et

al., 1995).

The combination of homologous recombination with site-specific recombination

systems such as the bacteriophage Cre-LoxP has allowed the more powefil tissue- and

tirne-specific modification of genes in the mouse genome during development (Gu et al.,

1993; Orban et al., 1992). The Cre-loxP recombination system has also been used to

induce site-directed chromosomal translocations and chromosome loss in the mouse

(Lewandoski and Martin, 1997; Smith et al., 1995).

Genetic manipulation of mice by random transgenic integration and by

hornologous recombination has increasingly allowed genetic dissection of development,

studies of gene function, expression and regdation, and the production of mouse models

of human disease. More than 100 mouse models of human disease exist where the

homologous mouse and human genes have k e n mutated, and in a majority of cases the

mouse phenotype closely resembles the human disease, thus providing valuable models

to understand disease and to devise ways to prevent or treat such diseases (Bedeil et al.,

1997). A transgenic database on the internet provide an update of aü mouse transgenics

and knockouts to date (http://www.gdb.org/Dan/tbase/tbase.html).

Gene trap insertional mutagenesis

Two main strategies for obtaining and studying mutations are: i) seeking genes

and mutations that display specific phenotypes of interest (forward genetics). and ii)

starting with a gene of choice and then looking for the phenotype of a mutation in that

gene (reverse genetics). In forward genetics, one selects and is sure about the phenotype

but may miss redundant genes or modifien that do not play an important role under the

set of conditions used. in reverse genetics, the gene of interest is moleculady defined but

may lead to a variety of phenotypes ranging from lethality to no obvious phenotype. Gene

trapping provides for a third alternative in obtaining and studying mutations where

random gene mutations are created by gene trap vector insertion and then mutations are

selected for further study based on. i) regulatory or signaling pathways involved. ii) in

vitro or Ni vivo expression pattern of the trapped gene, or iii) partial sequence information

of the gene.

ui a trapping expriment in the mouse, a construct consisting of a promoterless

lac2 reporter gene and a selection marker, commonly the neo gene, is randody

integrated into the genome of ES cells grown in culture. ES clones that have successfi.dly

integrated the trapping vector are selected in the presence of the drug G418. Then.

histochemical staining for -galactosidase ac tivity reveals the expression of lac2 gene

under the regulation of the trapped endogenous genomic sequences. Expression of lacZ

indicates a successful trapping event.

in order to achieve lac2 expression, the exogenous lac2 constmct must insert near

a gene enhancer (enhancer trap), in a gene intron (gene trap), or in a gene exon (promoter

trap) (Figure 1.1) (Zachgo and Gossler, 1996). In the case of the enhancer trap. lac2

camies a minimai prornoter at its 5' end and its activation by genomic enhancer sequences

indicates a trapping event (Figure 1.1A). In the case of gene trap, the l a d construct lacks

any promoter sequences, instead carries a 5' spiice acceptor site. Thus, its integration into

an intron leads to its expression and splicing to 5' exons of the endogenous active gene

(Figure 1 . lB). Such a genomic insertion leads to the generation of a fusion transcript

between lac2 and the 5' exons of the endogenous gene. This fusion transcript may be

used in 5' rapid amplification of cDNA ends-polymerase chain reaction (5' RACE-PCR)

to clone cDNA sequences of the trapped endogenous gene. The experiments described in

this thesis are based on a sirnilar gene trap strategy. Finally. in the case of the promoter

trap, lac2 carries no promoter or splice acceptor sites in its 5' end. The LocZ sequence

starts in this case with its open reading frame and it therefore must integrate i n - h e into

an exon of an active gene to be translated (Figure 1.1C).

Figure 1.1 Structure and activation of different trapping constnicts. In the enhancer trap

construct, lac2 has an initiation ATG codon. In the gene trap and promoter trap

constnicts. lac2 may or may not have ui initiation ATG codon, resulfing respectively in a

distinct Lac2 protein or a fusion Luc2 protein. [ATG]: optional ATG translation initiation

codon. eRE: endogenous regulatory element, mP: minimal promoter, lacZ bacterid P-

galactosidase gene, neo: neomycine resistance gene, P: promoter, PA: polyadenytation

signal, SA: splice acceptor site, line: intron or intergenic genomic sequences, solid boxes:

endogenous genomic exons, gray boxes: promoter regions on the vector, open boxes:

reporter and selector genes on the construct.

A. Enhancer Trap

l a ~ f neo ATG

Genomic locus aAer insertion

B. Gene Trap neo

1 WGI -- - - - veaor

Genomic locus

- Genomic locus after i d o n

C. Promoter Trap l o c ~ neo d

Genomic locus

Genomic locus after insertion

Following the enhancer trap experiments in Drosophih mentioned above, Gossler

et al. successfully attempted gene and enhancer trap experiments in mouse ES cells. In

these experiments, a lac2 reporter gene with a minimal promoter (Le. an enhancer trap

vector) or with a 5' splice acceptor site (Le. a gene trap vector) was inserted into the

genome and activated by flanking mouse sequences (Gossler et ni., 1989).

Developrnental regulation of lac2 expression was demonstrated in chimenc mice and

indicated the potential to introduce insertional mutations into the mouse germ line. Using

the vector sequences as genomic markers also indicated the possibility for molecular

cloning of the flanking endogenous genes. Thus. these trapping experiments

demonstrated that it was now feasible to: i) insertionally mutagenise random genes; ii)

reveal their developmental expression pattern; and iii) clone the gene sequences using the

trapping vector as an aid.

In a gene trap screen of ES cells, it was found that a trapping constnict bearing the

fusion lac2 and neo gene (Bgeo) is much more efficient in picking up lac2 expressing

trapped ES clones as compared to when the neo gene is independently driven by its own

promoter (95% versus 5%) (Friedrich and Soriano, 1991). In this experiment simi1a.r

efficiencies were observed when either plasmid-based or retroviral constructs were used.

The trapped loci were transrnitted through the germ line and the lac2 expression patterns

were examined in 27 strains of mice. Nine strains showed restncted expression patterns.

ten had widespread and eight ubiquitous expression pattems. Heterozygotes were crossed

in 24 strains, nine of which led to embryonic lethality in the homozygotes. The rest did

not manifest any overt phenotype, suggesting that a substantial portion of genes identified

in this approach are not essential for development.

In another gene trap approach, three insertions were examined for sequence.

expression and phenotype by transmission through the germ line (Skames et al., 1992).

The results demonstrated correct use of splice acceptor sites in the fusion transcripts and

revealed distinct lac2 expression patterns. Abnomalities were detected in two of the

three mutant insertions in the homozygous state. Ribonuclease protection assays detected

negligible amounts of normal endogenous transcripts in the homozygotes as a result of

splicing around the gene trap insertion.

Using retroviral promoter trap vectors, Melchner et ai. characterized nine trapped

clones, one of which was the previously characterized gene, REX- 1, and two of the four

clones passed to the germ line resulted in homozygous embryonic lethality (Melchner et

al., 1992). The murine homolog of the yeast RNA 1 gene, fugl , was also identified and

mutated in a gene trap screen and was shown to be required for gastrulation in

postimplantation mouse development (DeGregori et al., 1994). A novel gene, jumonji,

expressed at the midbrain-hindbrain boundary was captured in another gene trap

experiment, and shown to be required for neural tube formation (Takeuchi et al., 1995).

In the gene trap construct used, an intemal ribosome entry site (IRES) was inserted

between the neo and lac2 genes to allow dicistronic expression of the two genes. The

nuclear transporting signal peptide from SV40 large T gene was also placed at the 5' end

of lac2 to concentrate P-galactosidase in the nucleus to facilitate detection of l a d

expressing cells. Several other genes, including a-E-colenin, the Eck receptor tyrosine

kinase and cordon-bleu were also identified and/or characterized in gene trap approaches

(Chen et al., 1996; Gasca et al., 1995; Torres et al., 1997). In addition, in vitro

prescreening of gene trapped ES cells has allowed isolation of trapped ES clones

expressing lac2 specifically in the embryonic nervous system (Shirai et al., 1996). By

including an N-terminal signal sequence in the Lac2 protein, Skmes et al. captured

genes targeted to ce11 membrane and involved in mouse developrnent (Skames et al.,

1995).

Large scale gene trap screens have also been performed to identify and

characterize developmentally regulated genes. Wurst et al. isolated 393 ES ce11 clonai

integrations with lac2 reporter gene expression (Wurst et al., 1995). Of 279 clones that

generated chimeras, 13% showed restncted expression at E8.5 embryos, 32% showed

widespread expression, and 55% failed to show any expression at E8.5. One-third of

these latter clones displayed reporter gene activity at E12.5, indicating spatio-temporal

gene regulation during embryogenesis for a large proportion of the genes expressed in ES

cells. Baker et al. isolated 86 gene trapped ES ce11 clones and differentiated them to

embryoid bodies in vitro to assay for lac2 expression in different ceU types (a strategy

they refer to as 'in vitro preselection') (Baker et al., 1997). Twenty three percent of the

clones exhibited uniform expression, 22% exhibited no expression and 55% expressed

lac2 in limited ce11 types. Combined with partial sequence analysis from the clones, the

in vitro preselection is efficient and cost effective in prescreening for trapped clones of

interest before generating mice.

To assess the feasibility of large scale rnolecular cloning, 55 cDNA clones were

isolated and characterized from gene trapped ES ce11 lines using the SApgeo vector

(Chowdhury et ai., 1997). SAPgeo vector consists of a 5' splice acceptor (SA) site and the

fusion P-galactosidase-neomycin genes. isolated DNA sequences varied between 52 and

882 bp. Thirty percent encoded known gene products including nuclear, cytoplasmic,

cytoskeletal. membrane and extraceliular proteins; 21 % were present in the databanks as

ESTs; and 50% were new. These results suggested stochastic insertion of gene trap

vectors in the ES cells. Additionally. the site of the SAPgeo vector insertion within

known genes was random. In a sirnilar large scale approach, Hicks et al. devised a gene

trap shuttle vector to disrupt genes in ES cells and quickly obtain genomic sequence

information from the 5' flanking region (Hicks et al., 1997). Out of 4-00 inserts cloned and

sequenced, 42 (10%) were in known genes and 2 1 (5%) were present in EST databanks

indicating 85% unknown sequences, many of which could be intronic sequences. Gene

targets identified were various and included al1 major classes of cellular proteins. Sixteen

mutations were introduced into the germline, six of which resulted in obvious

phenotypes; however. in at least one case, insertion in an intron did not ablate

endogenous gene expression. Improvements in direct sequencing of RACE-PCR products

make sequence prescreening of gene trap integrations possible before further

characterization is undertaken (Towniey et al.. 1997). Sixty percent of trapped cell lines

in the study by Townley et al. were either spliced inefficiently or contained deletions of

the vector arguing for caution in such experiments.

It is of great interest to know if al1 genes can be eventually trapped in the mouse.

All genornic loci rnay not be equally accessible to insertion of exogenous DNA.

Nevertheless, at present. it seems possible to trap al1 genes expressed in ES cells,

estimated to be around 10,000 (Evans et al., 1997). The goal would be to create a gene

trap library of ES clones and perhaps thousands of mouse strains in an international effort

to extract biological information from the genomic sequences. Because gene trap

strategies cannot introduce subtle or gain-of-function mutations, gene trapping will not

replace gene targeting by homologous recombination; rather it provides a rapid stratew

for gene ciiscovery , mapping, mutagenesis, and expression studies.

In a gene trap screen, ES-tetraploid chimeras can be used to visuaiize gene

expression patterns. Generation of such chimeras is less demanding than the blastocyst

injection techniques and is more likely to represent accurately the endogenous gene

activity as ES cells, but not the tetraploid ceiis, contribute to embryonic tissues.

One shortcorning of the gene trap strategy is that a single clone of each gene trap

integration is available; thus independent confirmation of the resulting phenotype is not

possible. In contrast, gene targeting generates several independently derived clones of

targeted ES cells allowing the confmation of the phenotype.

Retinoic acid response gene trap screen

By preselecting gene trapped ES cells in vitro, Forrester et al. developed a novel

retinoic acid induction approach that selected for integrations into genes that lie

downstream of the RA signalhg pathways (Forrester et al., 1996). The screening strategy

is described in Figure 1.2. Essentially. it is based on the fact that ES colonies can be

replica plated and while the master plates of colonies may be saved for further

characterization, replicas of the plates are used in in vitro screens for lac2 response to

RA. Once colonies of interest are detected in the screen. master colonies can be used for

g e m line transmission. expression studies and cloning of the trapped genes. This

approach can be adapted to other receptorAigand-mediated signaling pathways.

In the above-mentioned RA screen. 20 gene trap integrations were identified, 9 of

which were induced and 11 were repressed afier exposure to exogenous RA. Al1 but one

of these integrations showed unique spatially restricted or tissue-specific expression

patterns during embryogenesis. Five out of six aoalyzed cDNA sequences were novel

genes and one was the protooncogene c-fyn. Germ-line transmission and breeding of 4

integrations uncovered one homozygous embryonic lethal (RI40 described in Chapter 4

of this thesis), and three homozygous viable insertions.

The work described in this thesis is based on characterization of two gene trapped

clones, 1193 and R140, isolated in this RA screen. Chapter 2 describes the novel RA-

induced Dr repetitive elements cloned from the 1193 gene trapped ES ceIl line, the Dr

nucleotide sequence and its expression pattern- Chapter 3 describes the RA-induced

Aquariur gene, its sequence <anaiysis and expression pattern during embryonic

development. Chapter 4 describes the identification and characterization of the mym gene,

from the RI40 gene trapped ES ce11 line. Finally. 1 summarize in Chapter 5 the work

done for this dissertation and describe sorne future experiments that might be undertaken

to better understand the Aquarius and mym genes.

Electroporate gene trap into ES ce&

vecto

Select G418 resistant colonies 1 for 6-48 hrs

Masterplaie (Freeze and save)

Histochemicd stiiining for lac2 expression

Pick lac2 positive colonies & test for RA response

I 1 1 -"a I

of lac2 staining

\ / Expand and fieeze

RA responsive clones (for RACE-PCR cloning, chimera

& genn line transmissiou) -

Figure 1.2 Retinoic acid gene trap screen in ES cells. Replica plating allows for in viho manipulation and screening of gene trapped ES colonies before M e r characterization. Gene trapped clones that are induced or repressed in response to RA are selected for cloning cDNA sequences and generating chimeric animals.

Chapter 2: A Novel Family of Repeat Sequences in the Mouse Genome

Responsive to Retinoic Acid

Chapter 2 is a version of the following publication: Sam M, Wurst W, Forrester L, Vauti F, Heng & Bernstein A. (1996) A Novel Family of Repeat Sequences in the Mouse Genome Responsive to Retinoic Acid. Mammalian Genome 7,7411748.

ABSTRACT

Repetitive DNA sequences form a substantial portion of eukaryotic genomes and

exist as members of families which differ in copy number, length and sequence. Various

functions, including chromosomal integnty, gene regulation, and gene remangement

have been ascribed to repetitive DNA. Although there is evidence that some repetitive

sequences may participate in gene regulation, little is known about how their own

expression may be regulated. During the course of gene trapping experirnents with

embryonic stem (ES) cells, we identified a novel class of expressed repetitive sequences

in the mouse using 5' rapid amplification of cDNA ends-polymerase chah reaction (5'

RACE-KR) to clone fusion transcnpts from these lines. The expression of these repeats

was induced by retinoic acid (RA) in cultured ES cells examined by Northern blot

analyses. In vivo, their expression was spatially restricted in embryos and in the adult

brain as determined by RNA in situ hybridization. We designated this family of

sequences as Dr (developrnentally regulated) repeats. The members of the Dr family,

identified by cDNA cloning and through database search are highly similar in sequence

and show peculiar structural features. Our results suggest the expression of Dr-containing

transcripts is part of an ES ceil differentiation program induced by RA.

INTRODUCTION

A substantial fraction of eukaryotic genomes is made up of repetitive DNA

sequences. Different ciasses of repeat sequences, ranging from a few copies per genome

to about one million Alu family members, are present in mamrnalian genomes

(Hellmann-Blumberg et al., 1993). The repeat units range from a few to a few thousand

nucleotides, and their genornic organization varies from tandemly repeated to widely

dispersed (Darne11 et aL, 1990). In mammals, interspersed sequences are classified as

SINES (short interspersed elements) ranging in size from 90 to 400 bp, and LINEs (long

interspersed elements) which can be up to 7.000 bp (Deininger et al., 1992). In the

mouse, SINE and LINE sequences account for about 15% of the genome. These elements

are dispened throughout the genome, and there is evidence to suggest that the dispersion

occurred by retrotransposition, i.e. by means of an RNA intermediate (Schmid and

Maraia, 1992). While LINEs may encode their own reverse transcriptase, SINEs are too

small to encode any proteins. Two major families of SINE sequences in the mouse, B1

and B2. have relatively short repeat uni& of -140 bp and -190 bp, respectively (Silver

1995). The majority of mouse middle-repetitive DNA sequences consist of numerous

families and are only a few hundred base pairs in length (Schmid and Deininger, 1975).

It has been proposed that repetitive sequences may provide a mechanism for

widespread and coordinated gene regulation (Britten and Davidson, 1969), although

evidence for their involvement in gene regulation remains limited. Alu repeat sequences

in humans are involved in transcriptional regulation (Brini et al., 1993; Saffer and

Thurstoa, 1989; Tomilin et al., 1990)- and they exhibit tissue specific patterns of

expression (Koga et al., 1988; Watson and Sutcliffe, 1987). In the mouse, Alu-like

repetitive sequences have k e n reported to confer growth- and transformation-dependent

regulation on transgene expression (Vidal et al., 1993). Repetitive elements upstream of

the murine erythropoietin receptor and the immunoglobulin Kappa light-chah genes

negatively regulate the activity of these genes (Saksela and Baltimore, 1993; Youssoufian

and Lodish, 1993).

Repetitive sequences may also be involved in gene rearrangements (Rudiger et

al., 1995; Wallace et al., 199 1). Alu repetitive elements have been frequentiy found at the

site of human gene rearrangements that occur by homologous recombination, such as in

sex chromosome exchange in XX males (Rouyer et al., 1987). rearrangernents in the

LDL-receptor (Hobbs et al., 1986; Rudiger et al., 1991), a-globin (Nicholls et al.. 1987).

P-globin (Henthom et al., 1986; Vanin et al., 1983). and apolipoprotein B (Huang et al.,

1989) genes, and in the c-sis protooncogene (Smidt et al., 1990). A short, weil conserved

region of the Alu repeat has been proposed to be involved in gene rearranpments

(Rudiger et ai., 1995).

Repetitive sequences have also been found at the site of transgene insertions

(Mark et al., 1992; W i k e and Palmiter, 1987), suggesting that these sequences may be

preferred sites for the integration of foreign DNA (Kato et al., 1986). Repeat-containing

sites may also be hotspots for mitotic recombination (Wilkie et al.. 1991). Finally,

instability of hypervariable minisateiiite repetitive sequences have been associated with

several human diseases, including carcinomas, leukemias, and insulin-dependent diabetes

(Krontiris 1995). Furthemore, these sequences bind transcription factors and activate

transcription (Krontiris 1995).

In this paper, we describe a novel class of expressed repetitive sequences in the

mouse genome whose expression is restricted during development and is induced in

response to retinoic acid (RA) in vitro. These sequences, which we have termed Dr repeat

sequences, were identified during gene trapping experiments while characterizhg fusion

transcripts between promoter-less lac2 transgenes, and unknown but transcriptionally

active endogenous genes (Forrester et al., 1996). The Dr element and its flanking

sequences contain palindromes, and direct and inverted repeats potentially capable of

fonning secondary structures in RNA and foldback structures in double suanded DNA.

MATERLhLS AND METHODS

Gene trapped embryonic stem (ES) ceil &es and tissue culture. Construction,

handling and characterization of the gene trapped mouse ES celI lines from the wild type

R 1 cells (Nagy et al., 1993b) used in this study have been described (Forrester et al.,

1 996).

5' RACE, PCR DNA amplifcation and cloning. 5' RACE-PCR on fusion

transcripts from the gene trapped ES ce11 lines was performed using a 5' RACE kit (Life

Technologies), essentially according to the manufacturer's instructions. with the

foliowing modifications: Reverse transcription was c&ed out at 4Z°C for 30 min in the

presence of [a32~]dCTP to monitor the synthesis of the fmt strand cDNA. Superscript

II (Life Technologies) and GTlacZ-1 [Sr-GCAAGGCGAmAAGTTGGGT-3'1 primer

were used for reverse transcription. DNA amplification of the RACE products was

performed in two rounds in 50-pi volumes containing 1X PCR Buffer II (Perkin Elmer),

1.25 mM MgC12, 200 mM deoxynucleoside triphosphates, 200 nM each primer, and 2.0

U of AmpliTaq DNA Polymerase (Perkin Elmer) in a DNA Thermal Cycler (Perkin

Elmer) (1 cycle of 94°C for 5 min; 1 cycle of 80°C for 8 min dunng which Taq

polymerase was added to the ceaction mix; 35 cycles of 94OC for 1 min, 60°C for 2 min,

and 7Z°C for 3 min; 1 cycle of 72°C for 5 min). The second round of amplification was

perfomed under the same conditions using nested primers (see below) and 5 ml of the

reaction product from the fmt round. The primes used for DNA amplification were as

f o I l o w s : f O r t he f i r s t r o u n d , GTZacZ-2A CS'-

CCGTCGACTCTGGCGCCGCTGCTCTGTCAG-3'1 and Anchor [Y-

GGCCACGCGTCGACTAGTACGGGLIGGGIIIGGGIIG-3'1 (Life Technologies); for the

second round, Nested2AU [5'-CAUCAUCAUCAUTTGTCGACCTGT

TGGTCTGAAACTCAGCCT-3'k and Universal Amplification Primer 15'-

CUACUACUACUAGGCCACGCGTCGACTAGTAC-3'1 (Life Technologies).

The RACE-PCR products were digested with Sa11 and cloned into pBlueScript II vector

(S tratagene) . Al tematively , they were cioned into the p AMP 1 plasmid using CloneAmp

System (Life Technologies). The cDNA inserts were sequenced on both strands using the

AutoRead Squencing kit (Pharmacia) on an A.L.F. DNA Sequencer (Phannacia).

cDNA cloning and sequencing. A mouse embryonic cDNA Library was screened

by standard procedures (Sarnbrook et al., 1989) using Dr-la as a DNA probe.

Hybridization was done in 50% formamide-5X SSPE ( l x SSPE is 150 mM NaCl, 10

mM NaH2P04 and 1 mM EDTA)-5X Denhardt's solution-100 mgfml, boiled and chilled

salmon sperm DNA-0.5% SDS at 42OC for 24 hours. Hi&-stringency washes were done

twice in 0.2X SSC, 0.2% SDS at 60°C for 15 min each. DNA was prepared from two

positive plaques (Dr-2 and Dr-3) and their cDNA inserts were sequenced as descnbed

above. GenBank search and sequence analyses were performed using Fasta, Bestfit.

Pretty and Pileup softwares in the Wisconsin Package, Version 8.0-Open VMS,

September 1994 (Genetics Cornputer Group, Inc., Madison, Wisconsin).

Ketinoic acid induction, RNA preparation and Northern blot analyses. In the

retinoic acid induction experiments, ES cells were maintained in ES ce11 medium

(without leukemia inhibitory factor - LE) containing 5% fetai calf semm (FCS) and 1 0 - ~

M dl-trans retinoic acid (RA) (Sigma). The medium was changed every 12 hours on al1

plates so that cells were in the continuous presence of RA for varying time periods (0, 12,

24 or 48 hours). Control samples (O hours) were in ES ce11 medium containing 5% FCS,

in the absence of LIF and RA throughout the expriment.

Total RNA was prepared from ES ce11 lines using a modified version of the

method described by Chomczynski and Sacchi (Chomczynski and Sacchi, 1987).

Confluent plates (60 mm) of cells were lysed with 5 ml guanidinium thiocyanate (4.4 M)

solution containing 25 m M sodium citrate (pH 7.0), 0.6% sarcosyI and 100 mM b-

mercaptoethanol. The lysate was mixed with 5 ml TE-saturated phenol, 1 ml chlorofonn

and 0.5 ml 2 M sodium acetate. m e r centrifugation, RNA was precipitated from the

aqueous phase with an equd volume of isopropanol. Northem blot analyses were carried

out according to standard procedures (Sambrook et al., 1989). The blots were hybridized

overnight with 32~-labelled probes, and were washed twice under stringent conditions in

0.2X SSC ( 1X SSC is 150 mM NaCl plus 15 rnM sodium citrate)-0.2% sodium dodecyle

sulphate (SDS) at 60°C for 15 min each.

RNA Ut situ hybridization. RNA in situ hybridization was performed as

described (Hogan et al.. 1994). Briefly, tissues were cryostat-sectioned at 10 Pm,

mounted on g l a s slides and refixed in 4% parafoddehyde. Pre hy bridization treatments

were performed as described (Hogan et al., 1994). 3%-labeled single stranded RNA

probes were prepared using T3 and T7 RNA polymerases (Boehringer Mannheim).

Adjacent sections were hybridized with Dr-2 sense and antisense probes. Post

hybridization washings included treatrnent with 50 pg/d RNase A (Sigma) at 37OC for

30 min. Following dehydration, the slides were dipped into NTB-2 film emulsion

(Kodak), exposed at 4OC for 6- 10 days, developed, and stained with toluidine blue.

Genomic screening and Southern blot analysis. A 129/Sv mouse genomic

library on 20 large LB plates was screened (1x106 plaques, representing four times the

haploid genome) using a Dr-la probe. For Southem blot, genomic DNAs from two

mouse species, C57BU6 and Mus spretus (SPRETEi, Jackson Laboratory) were

prepared and digested ovemight with either Sacl, ToqI or XbaI restriction enzymes.

Digested DNA (10pg per lane) was fractionated on a 0.9% agarose gel and transferred to

GeneScreen Plus nylon membrane (Dupont) and probed using standard procedures.

Hybridization and high-stringency washing conditions for genomic screening and

Southem blots were as described above for cDNA cloning.

FISH. Mouse chromosomes were prepared for fluorescence in situ hybridization

(FISH) according to the previously published procedure (Feng et al., 1994). Briefly.

lymphocytes were isolated fmm mouse spleen, cultured and then treated with 0.18 mg/d

bromodeoxyuridine for 14 hours. The synchronized cells were washed and recultured at

37°C for 4 hours. Chromosome siides were prepared as described previously (Heng et al.,

1992). The Dr-3 cDNA fiagrnent in p-Bluescnpt (Startagene) was used as probe and was

biotinylated with dATP ( W C , 1 hr) using the BioNick labeling kit (Life Technologies).

Hybridization was carried out ovemight at 37OC. The slides were washed and FISH

signals were detected and amplified as previously described (Heng et al., 1994). Mouse

chromosomes were stained with diamidino phenylindole (DAPI). A Leitz-Aristoplan

epifluorescent microscope with DAPI and fluorescein isothiocyanate (FITC) filters was

used for photographing the slides.

The experiments described in this report grew out of a study designed to identiQ

genes responsive to exogenous RA, by a novel gene trap approach in totipotent mouse ES

cells (Forrester et al., 1996). In this approach, we introduced a promoterless l a d

consbnict randody into the mouse genome by electroporation in ES cells. Because the

constmct contains a splice acceptor site 5' of l a d , a fusion transcnpt can be generated

that contains upstream exons of the trapped gene and the lac2 sequence. The lac2 can be

used as a marker to monitor the expression patterns of the trapped gene in ce11 culture and

in the embryo. We were therefore able to scmn the trapped ES cells for clones in which

the lac2 reporter was up- or down-regulated in response to RA. The IacZ transgene was

also used to clone the trapped gene. Using this approach. we isolated a number of ES ce11

clones which contain the gene trap lac2 constnict integrated within an RA-responsive

transcriptionai unit (Forrester et al., 1 996).

5' RACE-PCR cloning of fusion D r 4 transcripts from three independent

gene-trapped ES ceii lines induced with retinoic acid. To clone the trapped genes, we

empioyed a 5' RACE-PCR procedure to ampli@ fusion transcripts from several

independent gene trapped ES ce11 clones. The cDNAs obtained were subcloned and

sequenced. We repeated the RACE-PCR procedure on three clones. and repeated the

subcloning and sequencing of the RACE products. The cDNAs obtained were identical to

the ones isolated previously for each h e . Analysis of cDNA sequences indicated proper

splicing between an exon from an endogenous gene and the splice acceptor site of the

l a d reporter gene in dl cases. So far we have RACE products from three RA induced

and four RA repressed independent clones. The fusion transcripts from the three induced

clones contain at their 5' ends a common DNA sequence of 1 14-1 17 bp in addition to the

lac2 reporter and unique sequences. The RACE products from the four repressed clones

on the other hand do not contain this common DNA sequence (data not shown). This

fmding prompted us to study this common sequence hirther. We name these sequences

the Dr repeat famiiy, as their expression tunied out to be developmentally resuicted and

they are repetitive in the mouse genome. The three fardy members. Dr-la (1 17 bp), Dr-

1 b ( 1 15 bp) and Dr- l c ( 1 14 bp) were derived from the RA-responsive gene trapped ES

ce11 clones 1193, Il63 and 12 14, respectively. These three sequences are 92-974 identical

(Fig. 2.1). We described previously that the Luc2 expression pattern of the three ES ce11

clones, anaiyzed either in chimeric or normal embryos, are distinct during embryogenesis

(Forrester et al., 1996). In 1193-derived embryos, Lac2 is expressed in limb buds.

branchial arches, neural crest cells, and yolk sac endoderm. In 1163-derived embryos,

lac2 is expressed in the posterior spinal cord, adjacent mesodem. and yolk sac

rnesoderm. in IZlCderived embryos. l a d is expressed in the most rosval somites and

two lateral stripes dong the spinal cord.

Moiecular cloning and nucleotide sequence analysis of Dr-2 and Dr-3 cDNAs.

To obtain longer cDNAs containing Dr elernents. an 8.5 day post coitum (dpc) total

embryonic cDNA iibrary was screened ushg Dr- la as a probe. Two positive clones of 0.4

and 1.1 Kb were isolated, subcloned and sequenced (Dr-2 and Dr-3. respectively; Fig.

2.1). Dr-2 and Dr-3 are 95% similar to Dr- la and are themselves 95% similar throughout

the entire iength of the Dr-2 seqwnce (359 bp). The common Dr element and its flanking

regions in the Dr-2 and Dr-3 cDNAs have interesting sequence features, including long

and short palindromes, and direct and inverted repeats. These regions have the potential

to form stem-loop and other foldback structures, and the inverted repeat regions are

reminiscent of some transcription factor binding sites (Fig. 2.2). The 54-base long

palindrome close to the 3' end of the Dr-3 sequence is capable of fonning a large stem

loop structure. There is also a polyadenylation signal and a poiy A track. suggesting a

retrotransposition event in the history of these sequences. close to the 3' ends of the Dr-2

and the Dr-3 sequences (underlined in the case of Dr-3 in Fie. 2.2). In Dr-3, the long

paiïndromic sequence is immediately adjacent to the poiy A track. Al1 Dr sequences

obtained lack any significant open reading frame.

Figure 2.1. Alignment of the Dr cDNA sequences using the GCG Pretty software

program. Dr- la, Dr- l b, and Dr- Lc are fiom the 1193,1163, and I2 14 gene-trapped ES ce11

lines respectively. Dr-2 and Dr-3 were cloned from 8.5 dpc embryonic cDNA library (see

Materials and Methods). Deletions (-) are indicated based on the best dignment. The

numbenng is based on the Dr-3 sequence.

Dr-3 AGAATTTTGAGTTTGAGGAACACACCTTTPLATCTGGGCTACACCTTTCAT Dr-2 TG...C.CCGAG..A ......................... TG....... .

Dr-3 CTGGGATTAAAGGTGTGGTGGAACACACCTTTAA-TCTGGGCTACACCTT Dr-2 ............-..................... T.*.... .........

751 800 D r - 3 CTGCTGGAGACAATATAAGGACATTGGAAGAAGGGAGTCTAGCTCTTGCT .......................................... Dr-2 T.......

Dr-la ........ ........ Dr-lb . . Dr-lc A A . .

801 850 D r - 3 CTTGCTCCTTCGCCTGCTTGCTGCGTGAGAC--TGAGTAACTGCTAGATC D r - 2 ............................... CTGAG ..............

Dr-la A . . . . . . . . . . . . . . . . . . . . . . m . . . . . ~ a - - ................. Dr-lb ...........m.......... .Ta...... -- ................. Dr-lc ......... G..............C...... -- .................

851 900 D r - 3 CTTG---ACTCCATTCACAGCTGCGACTGAACCATTGTT-GGAATTGGGC D r - 2 .... GACCT .............................. G . . . . . . . . . .

Dr-la .... GA-CT ....................... A . . . . . . G .......... Dr-lb .... GA-CT..............,. .............. G.......... Dr-lc .... GA-CT. . . . . . ........-........ A . . . . . . G . . . . A . A . . .

901 950 Dr-3 TGCTGACTGTTAAGTCATCAATAAATTCCTTTACTATCTAAAAAAAAAAA D r - 2 ... C......A.-........................T..G.........

Dr-la .. . C . . . . C . A Dr-lb ... C..... Dr-lc . .AC ....

Figure 2.2. Dr-3 nucleotide sequence. Palindromes (thick arrows), direct repeats (thin

arrows), and inverted repeats (two-headed arrows) are indicated. Some of the small repeat

regions are not indicated for clarity. The presence of long and short palindromes is

indicative of the region's potential to fom foldback structures. The polyadenylation

signal and the poly A track are underlined.

In a GenBank search, three other mouse sequences were found to contain Dr

elements. Long mosaic repeated sequence (LMRS, Accession # X55036). KIF2

(Accession # D 12644) and Segment 4 of the transgenic mouse Myk- IO3 (Accession #

M16214) contain Dr elements that are respectively. 96%, 89%, and 98% sirnilar to the

Dr- 1 a sequence. No other sequences in the GenBank fkom other orgaaisms showed such a

degree of similarity to the Dr element. Best matches with other sequences in mouse or

sequences from other organisrns were in the 60-70% range. Such sequences may be more

distantly related to the Dr family. The nucleotide sequeace data reported here have been

submitted to GenBank and have k e n assigned the accession numbers US1725 (Dr-1),

US 1726 (Dr-Z), and US 1727 (Dr-3).

Dr-containhg transcripts are induced by reünoic acid in ES ceb. Because the

expression of the lac2 marker in the ES ce11 lines that were used to clone the Dr

sequences was induced in response to RA, we examined the RA responsiveness of Dr-

containing transcripts (Dr transcripts) by exposing wild type R1 and the gene-trapped

1193 ES ce11 lines to RA in the absence of LIF for up to 48 hours (see Methods). Northem

blot analyses of cytoplasmic RNA from these lines revealed a size range of Dr transcripts

(0.5- 18 Kb) whose expression did not alter in the presence of RA after 12 hours, but was

induced after 24 hours and was further overexpressed after 48 hours in RA (Figs. 2.3A &

2.3B). Induction of the Dr-facZ fusion transcript in the II93 line also confîîed its RA

responsivity (Fig. 2.3B).

Developmental expression of Dr repeats. We next examined the temporal and

tissue-specific expression pattern of the Dr element in embryos and adult tissues. RNA

transcnpts containing Dr sequences were observed at low levels in 8.5 dpc embryos and

their expression levels increased by midgestation (Fig. 2.3C). These hanscripts had a size

range similar to those observed in ES cells. Al1 adult tissues examined (testes, brain and

spleen) also express Dr trmscripts (Fig. 2.3C). To defme this expression pattern hirther,

we performed RNA in sim hybridization analysis on day 16 embryonic sections as well

as on adult brain (Fig. 2.4). Dr trax~scripts are widely expressed in day 16 embryos but the

highest expression is observed in the developing brain. specifically in the cortical areas.

In the adult brain, expression is widespread in the cerebral cortex with the strongest

expression in the hippocampus. Because of the high sequence homology between the Dr

family members, our probe was not specific to any particular member of this repeat

family. Thus, it is iikely that we are detecting multiple family members that rnight be

differentially regulated in different ceil lineages. Despite this lack of specificity of the

probe used for this RKA in situ analysis, it is interesting to note the developmentally

restricted pattern of expression.

Figure 2.3. RA induction of Dr expression in ES cells. and Dr expression in embryos and

adult tissues. Northern blot analyses were performed on tcn micrograms of total

cytoplasmic RNA using radiolabelid Dr-la probe and then a rnouse actin probe. (A)

Wild tvpe R1 ES ceiis, or (B) Il93 sene-trapped ceil Line were ueated with RA for O to

48 hours. In B. the sarne biot was reprobed with lac2 indicath_o the induction of its 4.5

Kb fusion transcript. (C) RNA from 8.5 to 15.5 days post coitum embryos, and adult

testes. brain and spleen tissues. The migrations of ribosomal RKAs are indicated as size

markers.

Figure 2.4. Developmental expression of Dr transcripts. RNA in situ hybridization was

perfonned in 16.0 days post coiturn rnouse embryo, sagittal sections (A-F), and in adult

rnouse brain, horizontal sections (G-J) using radiolabelled Dr-2 sense (B & E) and

antisense (C, F, H & J) RNA probes. Abbreviations: cb, cerebrum; cc, cerebral cortex; cp,

cortical plate; fc, frontal cortex; and hp. hippocampus.

Genomic representation and organization of Dr sequences. To gain

information about the genomic representation of the Dr elements, we screened a mouse

genomic library representing a fourfold equivalent of the haploid genome (106 PFU). A

significant number (0.12%) of the total clones screened contained Dr sequences,

suggesting a rough estimate of about 300 copies of Dr family members in the mouse

haploid genome. We then perfomied Southern blot analysis to examine in two related but

distinct mouse species the degree of conservation of the Dr loci (Fig. 2.5). At this level of

analysis, it was evident that there were both similarities and multiple differences between

laboratory mice and Mus spretus, hdicating a high degree of polymorphism in the Dr

loci. Similar polymorphism have previously been reported for several middle-repetitive

DNA sequences in two non-interbreeding strains of Drosophila (Young 1979).

The genomic organization of repetitive sequences range from tandemi y repeated

to widely dispersed across the genome. The Southern blot and genomic library analyses

described above both suggested that Dr sequences were widely dispersed thmughout the

genome. To test this further, we performed FISH analysis. This andysis revealed that Dr

sequences are widely dispersed across the whole genome on ai i the chromosomes (Fig.

2.5). A few of the signals were quite strong (Fig. 2.5. arrows) suggesting that some Dr

sequences rnay be clustered, a conclusion supported by the presence of several strong

bands of hybridization in the Southem blot analysis. The number of positive plaques from

the genomic screen and the number of signals from the FISH analysis were roughly in

correspondence. giving a minimum estimate for the number of Dr repeats in the rnouse

genome of 300. However, this number may be an underestimate of the Dr family size

since the strong signals in the FISH andysis may be the result of multiple copies of Dr

elements repeated in tandem. Furthemore, stringent conditions were used in these

analyses, suggesting that less highly related members of the Dr f d y might have been

missed in these experiments.

Figure 2.5. (A) Genomic representation of Dr sequences in two mouse species. Southern

blot andysis o f C57BU6 (lanes 1) and Mus spretus (lanes 2) genomic DNA (10 pg)

digested with three different restriction endonucleases and probed with radiolabelled Dr-

la cDNA. One Kb ladder (GibcoBRL) was used as size marker. (B-C) Localization of Dr

sequences in two sets of rnouse metaphase chromosome spreads by FISH. (B) Metaphase

chromosomes stained with DAPI. (C) PI-stained same chromosomes with F ï ïC signais.

Strong signais indicative of Dr clustering are rnarked by arrows.

Sacl Taql Xbal --- 1 2 1 2 1 2

DISCUSSION

We have identified a novel class of expressed repetitive sequences in the rnouse

genome, the Dr repeat farnily. We isolated these sequences from three independent gene

trapped ES ce11 lines (1 193,1163, and 12 14) responsive to RA. These lines were isolated

based on the insertion of a lac2 reporter into a gene that is upregulated in response to RA.

The three lines have distinct lac2 expression patterns during embryogenesis, indicating

the gene trapped in each line, and therefore the insertion locus, is also distinct (Forrester

et al., 1996).

The number of Dr elements per haploid mouse genome is of the order of 300

copies. This number is a rough estimate based on hybridization data from library

screening and RSH analysis. Dr sequences contain palindromes and direct and inverted

repeats. The repeat unit is at least 100 bp, and they are widely dispersed across the

genome. The Dr element can thus be considered as a low copy number SINE. The

presence of a polyadenylation signal and a poly A track in the Dr-2 and the Dr-3

sequences is indicative of a retrotransposition event in the history of these sequences. The

Dr-3 sequence also includes a 54-base long palindromic sequence close to its 3' end

which is capable of forming a stem loop structure in both DNA and RNA. Such

palindromic sequences have been found in most eukaryotic genomes and shown to be

capable of forming snap-back DNA structures (Hardrnan 1986). It is been suggested that

these sequences are mobile genetic elements (Perlman et al., 1976).

The three Dr-containing sequences identified in a GenBank search provide sorne

insights into the possible roles of these elements. The fmt Dr sequence (KM, # D 12644)

is in the 5' untranslated region of a kinesin-related gene (Aizawa et al., 1992). raising the

possibility that Dr elements may be included in RNA transcripts of functional genes

where the primary sequence or the secondary structure of the Dr sequence may affect

RNA stability or translation. Inclusion of other repetitive sequences in the RNA

transcripts of hnctional genes has already been reported. For example, there is a B2

repeat in the 3' untranslated region of the mouse muscle y-phosphorylase kinase gene

(Maichele et al., 1993), and a short interspened repetitive element in the 3' untranslated

region of four mamrnalian genes form the polyadenylation signal (Murnane and Morales,

1 995).

The second Dr sequence from GenBank (# M16214) is in a DNA fragment

displaced into the integration site of the MyK-103 transgenic mouse (Wilkie and

Palmiter, 1987). It is noteworthy that the Dr elements described here were also identified

on the basis of an insertion event, by a gene trap constmct. These results raise the

interesting possibility that Dr elements might bc preferred sites of transgenic insertions in

the mouse genome.

The third Dr sequence (# X55036) is part of a 15 Kb long mosaic repetitive

sequence (Decoville et al., 1992), suggesting that Dr elements may fuoction as part of a

long and complex repetitive sequence. The Iength and the structural features of the Dr-3

sequence, such as palindromic and inverted sequences, also support this notion. The high

degree of similarity between different Dr sequences in the mouse genome and their

absence in other organisms, indicated by GenBank search, suggests a role for these

sequences in the mouse.

niere are three possible ways in which Dr elements may have been introduced

into the 5' end of the lac2 fusion transcnpts that we cloned. First, they rnay have existed

downstream of the promoter within the 5' region of the genes where the lac2 transgene

inserted. In this case, the presence of the Dr element may render the locus favorable for

the integration of foreign DNA. It has been proposed that Alu repeats increase

susceptibility to gene rearrangements in hurnans (Rudiger et al., 1995). Other repeats

have also been found at the sites of integration of foreign DNA (Kato et al.. 1986). If

such preferred loci for integration do exist, gene trap insertions may not be as random as

it is generally thought.

Second, at the time of insertion, Dr elements may have cointegrated with the

transgene. There are precedents for the integration of repetitive sequences during gene

rearrangement. In characterizhg the insertional mutation locus in line 4 transgenic mice,

Mark et al. found insertion of DNA fragments similar to the rat LINE-1 repeat farnily at

the junction between host and foreign DNA (Mark et al., 1992). As noted above, Wilkie

and Palmiter also identified repetitive DNA fragments at the genomic integration site of

the MyK-103 transgenic mouse (Wikie and Palmiter, 1987).

Finally, Dr and lac2 transcnpts could have been brought together in Our gene

trapped lines by trans-splicing. Trans-spiicing has been well characterized in

trypanosomes (Agabian 1990; Matthews et al., 1994), and also occurs in organisms such

as C. elegnns (Blumenthal and Thomas, 1988) and A. lwnbricoides (Nilsen 1989). Tram-

splicing may also occur in rnammalian cells (Eu1 et al., 1995; Vellard et al., 1992).

At present, there is no evidence that would favor any of the above three models in

explaining the apparent involvement of Dr family members in the RA responsive gene

trapped lines. However, the presence of Dr elements within the lac2 transcripts in three

RA induced lines, and the activation of multiple Dr-containing an scripts in response to

RA. suggest the involvement of Dr sequences in an ES ceii differentiation prograrn that is

trïggered by RA. Evidence is now accumulating that repeat-containing vanscripts may

indeed be involved in regdatory pathways. Mouse BZ-containing (Alu type 2) transcripts

are strongly induced by mitogens and transforming agents (Edwards et al., 1985; Lania et

al., 1987; Singh et al., 1985). and theit expression is regulated by RA during the

differentiation of F9 embryonic carcinoma (Murphy et al., 1983) and PC 13 cells (Bennett

et al., 1984). Transcnpts from gamma satellite DNA repeats are differentially expressed

during mouse development and are strongly repressed by RA in Pl9 cells (Rudert et al.,

1995). Furthermore, the consensus sequence of a major Alu subfamiiy contains a

functional RA response eiement. raising the possibility that thousands of such sites in the

genome may be potential targets for RA regulatory pathways (Vansant and Reynolds,

1995).

The experiments reported here, as well as those descrïbed previously (Forrester et

al.. 1996) do not address whether the RA-responsive genes, and the Dr repeats contained

within them. are directly responsive to RA or are further downstrearn in a RA response

pathway. In these experiments, the kinetics of induction by RA is over a 2448 hour

penod. Such long kinetics of transcription activation by RA is not unusual (Simeone et

al.. 1990; Tini et al.. 1993). The HOXB 1 gene. for example, has an RA response element

in its promoter and is activated by RA after 24 hours (Ogura and Evans, 1995).

In summary, this paper describes the isolation of novel fusion transcripts

containing highly similar repetitive Dr elements that respond to RA. Although the

physiological significance of Dr repeats rernains to be elucidated, Our results suggest they

may be involved in developmenial and inducible patterns of gene expression.

Chapter 3: Aquarius, a novel gene isolated by gene trapping with a

RNA-dependent RNA poiymerase motif

This chapter is a version of the following publication:

Sam M., Wurst W., Klüppel M., Jin O., Heng EL, and Bernstein A. Aqzuzfius, a novel

gene isolated by gene trapping with a RNA-dependent RNA polymerase motif.

Developmenral DyMmics. (In Press)

ABSTRACT

In a retinoic acid gene trap screen of mouse embryonic stem cells, a novel gene, named

Aquarius (Aqr), was identified and characterized. A promoterless lac2 marker was used

to disrupt and trap the genomic locus and to determine the expression pattern of the gene.

Aqr transcripts are strongly induced in response to RA in vitro. During embryogenesis.

Aqr is expressed in mesodem. neural crest cells and their target tissues, as well as in

neuroepithelium. Expression was first detected at 8.5 dpc when neural crest cells are

visible at the lateral ridges of the neural plate. The gene trapped A q r locus was

transmitted through the mouse germ line in three genetic backgrounds. In the F2

generation, the expected rnendelian ratio of 1 :2: 1 was observed in al1 backgrounds

indicating that homozygous mice are viable. Homozygotes are normal in size and weight,

and breed norrnally. The Aqr ORF has weak similarity to RNA-dependent RNA

polymerases (RRP) of the murine hepatitis viruses and contains a RRP motif. Aqr was

maped to mouse chromosome 2, between regions E5 to F2 by FISH analysis.

Over the next decade, the complete DNA sequence of a mammal will become available.

This information will serve as a resource with which to decipher the genetic instruction^

required for embryological development, organogenesis, and the functioning of various

ce11 types in the adult. The challenge for biology will be to characterize functionally ail

the genes involved in these processes and in disease. Over the pst few years, various

strategies have been developed and refined to clone the genes correspondhg to existing

mutant or disease phenotypes, or to isolate genes that are the mammalian homologs of

genes in other organisms that are more amenable to genetic analysis. These approaches

can also be coupled with gene targeting in embryo-derived stem (ES) cells to generate

mice with mutations in any desired gene. However, these procedures are expensive, slow

and labor intensive, and hence are not yet sufficientiy robust to apply on a genome-wide

level,

Insertional mutagenesis, based on a stochastic insertion of exogenous DNA into

the genome of ES cells, provides another approach to generate novel mutations in the

mouse germ line. Furthermore, if the insertional construct includes features such as a

promoter-less lac2 cassette, then the inserted DNA provides an easy readout of the

normal expression pattern of the 'trapped' gene. and partial sequence information of

flanking exons can be derived by various PCR-based amplification strategies. This

process of gene trapping offers a feasible and potentially robust strategy to rnutagenize

and characterize the entire mouse genome.

Given the very large size of the marnmalian genome and large number of genes,

pseudogenes and non-coding regions, many of the ES clones isolated by a totally non-

selective gene trapping strategy will harbor insertions in regions of the genome that are

never expressed. Furthermore, a large portion of the remaining insertional events are

likely to be in genes that may not be of immediate interest. For example, in a recent test

for feasibility of large-scale gene trap mutagenesis, Wurst et al. recovered 279 gene-

trapped clones in a screen for developmentally regulated genes (Wurst et al., 1995).

Thirty-six clones (1396) exhibited resuicted patterns of gene expression in embryonic and

extraembryonic tissues, 88 (32%) showed widespread expression and 155 (55%) failed to

show detectable levels of expression at 8.5 days of gestation.

Two approaches have been described that are designed to enrich for particular

subsets of trapping events in vitro, prior to the generation of mice. Skarnes et al (1995)

developed a gene trapping construct that was designed to trap genes that encode secreted

and membrane-spanning proteins, such as receptor tyrosine kinases, phosphatases,

cadherin and laminin. The screening strategy was based oa an inframe insertion of the

CD4 trammembrane domain with the Pgeo (P-gaiactosidase-neomycine fusion) reporter

gene as the trapping vector, thereby capturing any insertion occming downstream of a

signal sequence (Skarnes et al., 1995).

In another approach, we have previously reported identification and preliminary

characterization of 20 gene trap integrations in ES cells that were screened in vitro for

response to exogenous retinoic acid (RA) (Forrester et al., 1996). Al1 but one of these

integrations subsequently showed unique and tissue-specific patterns of expression

during em bryogenesis. Sequence anal y sis from six integrations revealed five no vel genes

and one previously identified gene, the protooncogene c-fin. G e d i n e transmission and

breeding uncovered one homozygous embryonic lethal and three homozygous viable

insertions. A novel farnily of RA-responsive repetitive sequences was also identified in

this RA screen (Sam et al., 1996).

In other gene trap studies, novel genes with developmentaliy-restricted expression

patterns have k e n identified and it has been shown that their expression is not disturbed

by the insertion of the gene trap vector. Loss-of-function phenotype of homozygous

mutant embryos have also been characterized. Takeuchi et al. generated a gene trap

mouse mutation, jumonji, that leads to defective neural tube closure and embryonic

lethality before day 15.5 of gestation (Takeuchi et al., 1995). The gene is predominantly

expressed in the cerebellum and at the midbrain-hindbrain boundary and its deduced

amino acid sequence shares a portion of significant homology with human

retinoblastorna-binding protein RBP-2.

Chen et al. inactivated the murine Eck receptor tyrosine kinase by a retrovirai

gene trap insertion and showed that the expression of the Eck promoter was not affected

by provirus integration (Chen et al., 1996).

The function of a-E-catenin in mouse preimplantation development was also

defined by a gene trap approach (Torres et al., 1997). It was shown that a loss-of-function

mutation in a-E-catenin results in disruption of the trophoblast epithelium and

subsequent developmental block at the blastocyst stage.

To better understand B-ce11 development at the molecular level, the genetic

response of B-lineage cells to bacterial lipopolysaccharide (LPS. a potent stimulator of B-

cells) was also examined in a gene trap approach. Novel LPS-responsive genes were

identified and shown to have restricted expression within the B-lymphoid lineage (Kerr et

al., 1996).

The above experiments demonstrate the power of gene trapping to identifi and

analyze novel genes and their expression and function in the mouse. In this report, we

describe the isolation and characterization of the mamrnalian Aquarius gene, identified

by the RA induction gene trap strategy described above. and by using Intemet-based

database access and resources. We believe the combined gene trap and database approach

has considerable potential in increasing the Pace of gene discovery and functional

analyses, particularly as the sequence of the mouse genome cornes online.

MATERIALS AND METHODS

Gene trapped embryonic stem (ES) celi lines, lac2 staining and germ line

transmission. Construction, handling and characterization of the Il93 gene trapped ES

cell line from wild type R I cells (Nagy et al., 1993b), l a d staining and transmission of

the trapped locus through the mouse germ line have been described (Forrester et al..

1996).

Mouse genotyping and F2 breeding in difFerent backgrounds. Genomic DNA

was isolated from tails at the time of weaning, digested with EcoRl restriction enzyme

and separated by elecvophoresis on a 1% agarose gel. Southern blots were double

probed with both engrailed-2 (en-2) and 1ucZprobes. En-2 was used as an interna1 control

for quantitative Southem and lac2 probe detected the trapping vector inserted in the Aqr

locus. Densitometric analyses were used to measure the intensity of the l a d signal (4.0

Kb band) relative to the en-2 signal (1 1 Kb band). Three FI males and six FI females,

heterozygous for the disrupted locus, were bred to obtain F2 progeny in a 50% 129sv-

cp/50% C57BL6 outbred background. Because the chimeric animals generated by

blastocyst injection died shortly after germ line transmission, outbred F1 males were also

bred to CD-1 and 129 females to obtain 50% CD-1/25% 129/25% C57BL6, and 75%

129/25% C57BL6 backgrounds. Three males and six femaies from these backgrounds

were also bred to obtain F2 progeny. Heterozygous animals (+/-) would be expected to

have a IacWen-2 signal ratio of 1, while homozygotes (4) would be expected to yield a

lacZ/en-2 signal ratio of 2 on Southem blots. To confirm the genotyping, iwo

heterozygotes and two homozygotes were backcrossed to wild type animals and their

progeny were genotyped for the presence or absence of the transgene. In all four crosses,

half of the progeny from the heterozygotes and al1 the progeny from the homozygotes

carried the transgene, c o n f i n g the original genotypes.

RACE, PCR amplification, cloning of the 5' Aqr exon, and genomic PCR. 5'

RACE-PCR on fusion transcnpts from the gene trapped ES ce11 lines were perfomed

using a 5' RACE kit (Life Technologies, BurLington, On), essentialiy according to the

manufacturer's instructions, with the following modifications: Reverse transcription was

carried out at 42°C for 30 min in the presence of [a-32PIdCTP to monitor the synthesis of

the first strand cDNA. Superscript II (Life Technologies) and GTlacZ-1 [S-

GCAAGGCGATTAAGTTGGGT-3'1 primer were used for reverse transcription.

DNA amplification of the RACE products was performed in two rounds in 50-pl

volumes containing 1X PCR Buffer II (Perkin Elmer, Foster City, Ca), 1.25 mM iMgCI2,

200 rnM deoxynucleoside triphosphates. 200 nM of each primer, and 2.0 U of ArnpliTaq

DNA Polymerase (Perkin E h e r ) in a DNA Thermal Cycler (Perkùi Elmer) (1 cycle of

94°C for 5 min; 1 cycle of 80°C for 8 rnin during which Taq polymerase was added to the

reaction mix; 35 cycles of 94°C for 1 min, 60°C for 2 min. and 72°C for 3 rnin; I cycle of

72OC for 5 min). The second round of amplification was performed under the sarne

conditions using nested primers (see below) and 5 pi of the reaction product from the

first round. The primers used for DNA amplification were as follows: for the fmt round,

GTlacZ-2A [SI-CCGTCGACTCTGGCGCCGCTGCTCTGTCAG-3'1 and Anchor [SI-

GGCCACGCGTCGAmAGTACGGGI][GGGIIGGGIIG-3'1 (Life Technologies); for the

second round, Nested2AU [Y-CAUCAUCAUCAUTTGTCGACCTGT

TGGTCTGAAACTCAGCCT-3'1, and Universal Amplification Primer [S-

CUACUACUACUAGGCCACGCGTCGACTAGTAC-3'1 (Life Technologies,

B urlington, On).

The RACE-PCR products were digested with San and cloned into the vector

pBlueScript II (Stratagene. La lolla, Ca). Aitematively, they were cloned into the pAMPl

plasmid using the CloneArnp System (Life Technologies). The cDNA inserts were

sequenced on both suands using the AutoRead Squencing kit (Pharmacia, Baie dlUrfe,

Qc) on an A.L.F. DNA Sequencer (Pharmacia).

PCR amplification on genomic DNA from +/+, +/-, and 4- animais was

performed with pnmers flanking the 5' Aqr exon, 1 [SI-GCATGTAAATACTGGGCT-3 'l.

and 2 [SI-CTAAATTCCAGCAGCATT-3'1 and the GTlacZ-1 primer (above). PCR

conditions were as described above.

cDNA cloning, sequencing, GenBank and internet utiiitïes. A 12.5 days post

coitum rnouse embryonic cDNA library was screened by standard procedures (Sambrook

et al., 1989) using a non-repeat region of the RACE product (Aqr exon) as a DNA probe.

Hybndization was done in 50% fornamide. 5X SSPE (1X SSPE is 150 mM NaCl, 10

mM NaHzP04 and 1 mM EDTA), 5X Denhardt's solution 100 mg/mi, boiled and chilled

salmon sperm DNA, 0.5% SDS at 42OC for 24 houa. High-stringency washes were done

twice in 0.2X SSC. 0.2% SDS at 60°C for 15 min each. DNA was prepared from the only

positive plaque and its 2.4 Kb cDNA insert was sequenced as described above.

GenBank searches and sequence analyses were performed using Fasta. Translate,

Bestfit. Gap and Pileup sofiwares in the Wisconsin Package. Version 8.0-Open VMS,

September 1994 (Genetics Computer Group, Inc., Madison. Wisconsin). For the A q r

ORF sequence alignment with the viral RNA-dependent RNA Polymerases. I used the

GCG Pileup program with a gap weight and a gap weight Iength of 2. Only Aqr residues

that were in cornmon with at least two of the three other RRPs were considered

significant. To evaluate the alignment significance obtained by the Pileup prograrn,

Bestfit and Gap programs with LOO randomizations were used in painvise alignments of

Aqr with the three RRP sequences. Bestfit fin& the best region of similarity between two

sequences while gap generates the best alignment of the two sequences in their entirety.

In randomizations, the second sequence is repeatedly shuffled while its length and

composition maintained, and then it is realigned to the first sequence. The average

alignment quality, plus or minus the standard deviation, of al1 randomized alignments is

reported in the output file (Genhelp Besifit, Genetics Computer Group, Inc. 1992). z

score is a measure of the alignment significance and is calculated as: (actual alignment

quality - average random quaiity)/(standard deviation). z values above 3 are considered

possibly significant and above 10 as significant (Lipman and Pearson, 1985).

Aqr open reading frame was submitted to the National Center for Biotechnology

Information XREF intemet service (http:Ilwww.ncbi.nlanih.govB(REFdb/) where it was

searched against the most recent update of the EST database. Positive matches were

examined for homology with Aqr. Online sequence databases for yeast (100% of the

genome complete, http://speedy .mips. biochem.mpg.de/mips/yeast/) and C. elegans (60%

of the genome complete, http:l/www.sanger.ac.uW-sj/C.eleganssHomeehtrnl) were aiso

searched for homology with Aqr.

Retinoic acid induction, RNA preparation, Northern blot analyses, and RT-

PCR. In the retinoic acid induction experiments, ES cells were maintained in ES ce11

medium (without leukemia inhibitory factor - LF) containing 5% fetal calf semm (FCS)

and 10-6 M ail-tram retinoic acid (RA) (Sigma, St. Louis, Mo). The medium was

changed every 12 hours on ail plates so that ceiis were in the continuous presence of RA

for varying thne periods (0, 12, 24 or 48 hours). Control samples (O hours) were in ES

ce11 medium containing 5% FCS, in the absence of LIF and RA throughout the

experiment. In in vivo RA experiments, pregnant mice at E8.5-E9.5 were fed 2 mg of RA

per 100 gram of body weight and then embryos were surgically removed 24 or 48 hrs

after RA treatrnent and stained for IacZ histochemicd staiaing.

Total RNA was prepared from ES ce11 lines using a modified version of the

method described by Chomczynski and Sacchi (Chomczynski and Sacchi, 1987).

Confluent plates (60 mm) of cells were lysed with 5 ml guanidinium thiocyanate (4.4 M)

solution containing 25 mM sodium citrate (pH 7.0). 0.696 sarcosyl and 100 mM b-

mercaptoethanol. The lysate was mixed with 5 ml TE-saturated phenol, 1 ml chloroform

and 0.5 ml 2 M sodium acetate. After centrifugation, RNA was precipitated £Yom the

aqueous phase with an equal volume of isopropanol. Northem blot analyses were carried

out according to standard procedures (Sambrook et al., 1989). The blots were hybridized

ovemight with 32~-labelled probes, and were washed twice under stringent conditions in

0.2X SSC (lx SSC is 150 mM NaCl plus 15 mM sodium citrate)-0.2% sodium dodecyle

sulphate (SDS) at 60°C for 15 min each.

RT-PCR was done with the GeneAmp RNA PCR kit (Perkin Elmer, Foster City,

Ca) according to the manufacturer's instructions. The primers used for PCR amplification

were the Aqr exon Banking primers A & B. and the Dr-la repeat flanking primes C & D;

A [5'-GAGCTCTTGCTATTGmC-3'1. B [5'-TAGAGTCGGCAGCCCAAT-3'1, C [Y-

GCATGTAAATACTGGGCT-3'1, and D [5'-CTAAATTCCAGCAGCArlT-3'1. The PCR

products were mn on a 1 % gel. blotted and hybridized with 32~4abelled Dr 1 a-Aqr probe.

RNA and whole mount in silu hybridization. RNA in situ hybndization was

performed as described (Hogan et al., 1994). Briefly, tissues were cryostat-sectioned at

10 Pm. mounted on glas slides and refixed in 4 4 paraformaldehyde. Prehybridization

treatments were performed as described (Hogan et al., 1994). 3%-labeled single stranded

RNA probes were prepared using T3 and T7 RNA polymerases (Boehringer Mannheim,

Laval. Qc). Adjacent sections were hybridized with Aqr sense and antisense probes. Post

hybridization washes included treatrnent with 50 p g / d RNase A (Sigma, St. Louis, Mo)

at 37°C for 30 min. Following dehydration, the slides were dipped into NTB-2 film

emulsion (Kodak, Rochester. NY). exposed at 4OC for 6-10 days, developed, and stained

with toluidine blue. Whole mount in situ hybndization was performed as described

previously (Conlon and Hernnann, 1993). Single-stranded A q r RNA probes were

labeled with digoxygenin labeling kit according to the manufacturer (Boehringer

Mannheim Biochemicals). After RNA in situ hybridization, embryos were postfixed in

4% padormaldehyde at 4 'C ovemight. Sections were cut ai 5-6 pm and some sections

were counterstained lightly with eosin, and photographed using a Leitz Orthoplan

compound microscope and Nomarski optics.

FISH. A genomic clone for A q r was obtained and used as a probe for

fluorescence in situ hybridization (FISH) analysis. FISH was performed according to

published procedures (Heng et al., 1992; Heng and Tsui, 1993).

The Aquarius gene descnbed in this report was identified in a gene trap screen of

totipotent mouse ES cells, designed to capture genes responsive to exogenous RA

(Forrester et al.. 1996). In this approach, the gene trap construc,: carrying a splice

acceptor site 5' of the promoterless lac2 reporter gene was randody integrated into the

rnouse genome by electroporation. Twenty gene trapped ES clones were isolated that

were either induced or repressed after 48 hours of exposure to RA. This paper describes

the cloning of the Aqr gene from one such clone, 1193. This ES clone was of particular

interest to us since its lac2 was strongly induced in response to RA. Retinoic acid and its

derivatives regulate important biological processes. including ce11 proliferation,

differentiation and morphogenesis in a varïety of developmental systems (Mangelsdorf et

al., 1995).

Clonhg of the RA-responsive Aquarius gene

To examine RA responsivity, Il93 cells were grown in the presence or absence of

leukemia inhibitory factor (LIF) which maintains ES cells in an undifferentiated state.

and in the absence of LIF and in the presence of RA for 48 hrs, and then stained for lacZ

activity. As shown in Figure 3.lA-C, the lacZ reporter gene was strongly induced in the

1193 ce11 line in response to RA, indicating the RA responsivity of the trapped locus. TO

determine RA response NI vivo, we fed pregnant mothers with RA and stained embryos

for lac2 activity (see Materials and Methods). We did not detect any reproducibly

significant change in the extent of staining in the embryos from RA-treated mothers

versus those from untreated mothers.

To obtain sequence information about the trapped gene, 5' rapid amplification of

cDNA ends-polymerase chain reaction (RACE-PCR) was carried out twice in two

independent reactions on total RNA isolated from RA-treated 1193 cells. Both

experirnents yielded an identical 0.4 Kb cDNA, which was cloned and sequenced. One

hundred and fifty bases of the cDNA were derived fiom the gene trap vector. The 5' half

of the RACE product was the Dr-la repeat element that we have previously characterized

(Sam et al., 1996). Dr sequences have k e n subrnitted to GenBank (Accession numbers:

U51725-U5 1727). The 3' haif of the RACE product was novel, with an open reading

frame from a 5' exon of a novel gene which we have designaied Aquarius (Aqr). The

structure of the RACE product is shown in Figure 3.4A.

We used the 5' exon as a probe on Northern blot analysis of RNA isolated at

different time points from wiid type RI ES ceils treated with RA (Figure 3.1 D). Three

mRNA isoforms of 2.5, 6.0 and 8.5 Kb were induced after 24 hours of exposure to RA.

The 24-hour delay in induction may suggest that Aqr is further downstrearn of an RA

signaling pathway, however. such long kinetics of transcription activation by RA is not

unusud (Simeone et al., 1990; Tini et al., 1993).

To confirm the inclusion of the Dr repeat element within Aqr transcripts, we

designed two pairs of PCR primes, one fiom the Dr element and one from the Aqr ORF

(Figure 3.1 E) and used them in RT-PCR reactions on total RNA from the parental R 1 and

gene trapped 1193 lines (Figure 3. IF). Amplification of the expected 250 bp fragment

with A & D pnrners c o n m e d the presence of a Dr element on the Aqr transcript in both

ceil lines. We have previoirsly reported the inclusion of Dr repetitive elements on many

RA responsive transcripts ( S m et al., 1996).

Fipre 3.1 Luc2 and Aqr expression are induced in vitro in response to retinoic acid

(RA).

Il93 gene trapped ES cells stained for Luc2 activity, (A) in the presence of LIF and in the

absence of RA, (B) in the absence of LIF and the absence of RA, (C) in the absence of

LIF and the presence of RA for 48 hrs. (D) RA induction of Aqr expression. Wild-type

R I ES cells were treated with RA for O to 48 hours. Northern blot analyses were

performed on ten micrograrns of total cytoplasmic RNA using a radiolabelled Aqr probe

and then a mouse actin probe. The migrations of ribosomal RNA (1.9 & 4.9 Kb) are

indicated as size markers. (E) Diagrammatic representation of the Aqr transcript and Dr-

la repeat sequence deduced from the RACE-PCR cDNA product. Pnmers A-D were

used in RT-PCR experiment to confirm the presence of Dr-la sequences on the Aqr

transcript (F).

I l 93 CONA. 5' RACE produet

- R.T. +kt.

Aq-us has weak simiiarity to RNA-dependent RNA polymerases (RRPs)

To determine the sequence of the coding region comsponding to the A q r ORF,

we screened a 12.5 days post coitum embryonic cDNA library using the A q r 5' exon as a

probe. A single 2.4 Kb cDNA was obtained and sequenced. This cDPlA contained a

novel552 amino acid open reading frame (ORF) with weak sirnilarity to RNA-dependent

RNA polymerases (RRP) of the murine hepatitis and avian bronchitis viruses (GenBank

accession: VFIHJH, S 15760 and VFMBZ) . There is 25% identity between Aqr and the

viral RRP genes over the entire deduced amino acid sequence of A q r (Figure 3.2A). The

2.4 Kb cDNA that we have cloned is simiiar in size to the 2.5 Kb isoform of the A q r

transcript on Northem blots. However our cDNA is not cornplete as it does not contain an

initiation codon or 5' UTR.

To evaluate the significance of Aqr and RRP sequence alignment, Gap and Besdit

sequence alignments were performed with 100 randomizations to calculate z scores as

described in the Materials and Methods (Table 3.1). Since Pileup, Gap and Bestfit

programs use different algorithrns for aligning sequences, the actual alignments obtained

in each case are different even if in some cases they were in the sarne region (asterisks in

Table 3.1 ). The sequence identities in these alignments were between 22 to 3 1 percent. In

al1 cases reported in Table 3.1, the sequence similarity is dong the whole length of the

A q r ORF although Besdit f i d s only the best region of similarity. z scores above 3 are

considered possibly significant (Lipman and Pearson, 1985) and is obtained by Besdit in

three alignments (Table 3.1). However, since the z score obtained in the other cases is

below 3, the sequence aiignment between A q r and the RRPs must be taken with caution

and as tentative.

Two sets of values for gapweight and gaplength weight parameters were used for

comparison in pairwise alignments. As indicated in Table 3.1, the gapweight and

gaplength weight values have a notable effect on z scores, again suggesting caution in

evaluating sequence alignments. In such cases of weak sequence similarity with

questionable signincance, other biological and biochemical data such as sequence motifs

and biochernicai assays, or three dimensional structural information about the protein

product are highly desirable.

The 25% sequence identity with RRPs rnay nevertheless be significant because

the Aqr ORF also contains the four segment RRP motif (A-D) corresponding to the

polymerase active site (Figure 3.2B) (Poch et al., 1989; Sousa 1996). The RRP motifs

within the viral RRPs are in a different region of their sequence from the ones that are

shown in Figure 3.2A. Three dimensional structure of RRPs or their active site is not yet

determined and therefore the signifcance of the RRP motif alignment in the figure 3.2B

is not clear. Having the active site motif, Aqr may possess R N A polymerase activity.

ALtematively, it may have descended from a common ancestor shared with RRP genes

but which later lost its original function.

The Aqr ORF also contains two putative nuclear localization signals (several

lysine and arginine residues) similar to those present in known nuclear proteins such as

p53 and D M (Figure 3.2C), suggesting nuclear localization of the Aqr gene product

(Robbins et al., 1991). Several lysine and arginine residues in a small stretch of amino

acids constitute a nuclear localization signal (Robbins et al., 1991). The nucleotide

sequence (GenBank: U90333) includes a 5' palindrornic region within the ORF capable

of forming a 27 bp stem and a 5 base-long loop structure. There is also an RNA

instability motif (AUUUA) in the 3' untranslated region (Ross 1996). suggesting a shoa

haif-life for the Aqr transcript.

We searched several online sequence databases to identify other genes with

significant homology to Aqr. No significant homology was detected in the yeast or C.

elegans databases. However, there were four EST sequences. two from human and two

from the mouse EST databases, that were highly similar to the Aqr ORE The four ESTs

show 88-10046 amino acid sequence identity with Aqr in overlapping regions. Human

EST 56423 1 (GenBank: AM2 1582) was cloned from NT2 neuronal precursor cells and

is 9496 identical to Aqr over 167 amino acid residues; human EST 626472 (GenBank:

A M 8 8 145) was cloned fiom HeLa cells and is 88% identical to Aqr over 92 amino acid

residues; mouse EST 765493 (GenBank: AA274735) was cloned from mouse lymph

node and is 100% identical to Aqr sequence over its full length of 157 amino acids, and

rnouse EST 535 17 1 (GenBank: AA073476) was cloned from retinoic acid induced Pl 9

embryonic carcinoma ceils and is 89% identical to Aqr over 123 amino acid residues.

This Iast EST was cloned from an RA induced ceii Line, suggesting it may also be

responsive to RA.

Gerrn iine transmission of the gene trapped Amus locus

The trapped Aqr allele was transmitted through the germ line of mouse chimeras to

produce heterozygous I l 9 3 mice. Heterozygotes were identified and bred to obtain F2

progeny in three different outbred backgrounds (see Materials and Methods). Al1

genotyping was perforrned using a quantitative Southem blot analysis (Figure 3.3A &

3.3B). The F2 progeny results generaily indicate a 1 :2: 1 Mendelian distribution of +/+.

+/- and -1- animals (Figure 3.3C). The distribution in the 75% 129 background was

slightly skewed but not statistically significant (Chi-Square Goodness of Fit Test. P =

0.09). suggesting that a few homozygous mutant embryos may not have suMved to birth.

However, when embryos were examined from mid- to Late-gestation, no gross

abnormalities were observed in -/- embryos. At 10 months of age. +/- and -1- animais

were the same weight and size as their wild type (+/+) littermates, bred normally and did

not manifest any discemible abnormalities. Newbom skeletons were also prepared and

stained with alizarin red and alcian blue stains for skeletal and cartilage structures. NO

abnonnalities were detected in craniofacial bone or cartilage structures where A q r is

expressed.

To confirm genomic disruption of the A q r locus by the gene trap vector, genomic

DNA from +/+, +/- and -1- animals was amplified by PCR using primers from the 5' A q r

exon and the l a d sequences (Figure 3.3D). We obtained arnpiîfied recombinant DNA

products from +/- and -/-, but not from +/+, animals. confimiing disruption of the Aqr

gene by lac2 sequences (Figure 3.3E).

In order to determine if Aqr transcripts are present in the homo y gous animais.

we performed Northern blot analysis on total RNA from genotyped midgestation

embryos (Figure 3.4B). Aqr transcripts were present in the homozygous, as well as

heterozygous and wild type embryos, indicating that the disniption of the Aqr locus by

the gene trap vector does not lead to a nul1 mutation. The gene trap vector employed in

these expenments were designed such that transcription is terminateci after the l a d

polyadenylation site, thus leading to a transcription disruption of the endogenous trapped

gene. It is possible however that RNA polymerase proceeds with transcription

downstream of the polyadenylation site. Polyadenylation site selection by RNA

polymerase II and RNA transcript cleavage are not weli understood (Alberts et al., 1994;

Dameu et al., 1990). and rernains a concem in gene trapping experirnents. In the case of

a run-through transcription, splicing across the gene trap vector sequences could produce

a wild type mRNA. The presence of a wild type Aqr transcript may indicate why we did

not detect any abnomialities in the homozygous mice.

Figure 3 3 Aqr is similar to viral RNA-dependent RNA polymerases (RRPs).

(A) Alignment of the Aqr deduced arnino acid sequence with the murine hepatitis virus

RRPs (Vfihjh & S 15760). and the avian infectious bronchitis virus RRP (Vfihb2). Gaps

(.) are introduced for the best alignrnent but are not counted in numbenng of residues.

Numbering is based on the Aqr amino acid residues. Amino acid identities are shaded.

(B) Alignment of the RRP conserved motifs for the A q r and the three viral RRP

sequences in (A).The strictly conserved residues are in bold and the conserved

hydmphobic residues are indicated by a star. Numbers indicate the arnino acid residues

behlreen the fragments or from the beginning of the sequence. (C) Aqr contains two

Nuclear Localizacion Signals (NLSs) aligned here with known NLSs from four other

genes. Nurnbers indicate the residues from the beginning of the sequence.

A- Vf ihjh SI5760 Vf i&b2

Aqr V f i h j h SIS760 Vf a b 2

Aar

A w V f i h j h SIS760 Vfihù2

41 7 5 162 V f i h j h 717 89 44 515760 717 89 44 Vfihb2 729 27 0SRL.m 189 92

m e * O 0 r

Table 3.1. Measurements of Aqr aügnment sigrMcance by randomization wïth Gap

and Bestfit.

GCG programs, Gap and Bestfit were used with 100 randomizaticns on painvise

sequence alignments between Aqr and RRP sequences. z scores above 3 are considered

as possibly significant and were calcdated as descnbed in the Materials and Methods.

Two sets of values for gapweight and gaplength weight parameters were used for

comparison. Asterisks indicate sequence similarity in the same region as in Figure 3.2.

Aqr aligned gapweight, w ith gaplength

Vfihjh 1,4 2, 2

Figure 33. Cenotyping of F2 progeny and the genomic disruption of the Aqr locus.

(A) Southem blot. The upper band is the endogenous en-2 gene used as intemal control

for DNA loading. The lower two bands contain the lac2 transgene. The ratio of the

intensity of the faster lac2 band (peak 3 in B) to the en-2 band (peak 1 in 8) indicates the

genotype of the animal: +/+ (ratio = O, lane 5), il- (ratio = 1, lanes 2 & 4), and -/- (ratio =

2, lanes I & 3). (B) The densitometric measurements of the band intensities. (C)

Genotypes of F2 progeny in three mixed genetic backgrounds. i) 50% 129svcp/50%

C57-Bi6, ii) 25% 129sv-cp/25%C57-B16/50% CD-1, fi) 75% 129svcp/25% C57-B16.

(D) Lac2 disrupts Aqr genomic locus. Aqr exon was cloned by 5' RACE-PCR. The

primers 1-3 were used for PCR on genomic DNA to confvm disruption of the Aqr locus

by the gene trap vector. (E) PCR on genomic DNA from +/+, +/- and -1- animals and

Southem blot probed with Aqr exon.

Figure 3.4 Aqr RACE product and Northern anaiysis of Aqr embryonic expression.

(A) The structure of the RACE product. Dr- la: 1 19 bp Dr repeat element, Aqr: 133 bp

Aqr exoo, and the Lac2 reporter sequeoces are indicated. (B) Northern blot aaalysis on

total rnidgestation embryonic RNA from wild type (+/+), heterozygous (+A) and

homozygous (-/-) embryos. The blot was hybridized with the 2.4 Kb Aqr cDNA probe.

Three Aqr transcript isofoms are present in the embryos of the three genotypes.

Ribosomal RNA are indicated as size markers.

Dr-la LacZ

Embryonic expression in neural crest and mesodermd tissues

The developmental pattern of A q r gene expression was examined during embryogenesis

by staining for 8-galactosidase activity (Figure 3 3 , and by RNA and whole mount in

situ hybridization techniques using the Aqr cDNA as a probe (Figures 3.6 & 3.7). At day

8.5 of gestation, expression was detected in the cephalic neural folds and mesoderm,

paraxial mesoderm, somites, and throughout the lateral ridges of neural fol& including

the fused regions in between the caudal and rostral extremities of neural tube closures

(Figure 3.5A). The lateral ridges of neural fol& are where premigratory neural crest ceils

emerge and subsequently migrate in a rostral-to-caudal sequence at the cranid, trunk and

tail levels (Serbedzija et al., 1992; Serbedzija et al., 1990). Trunk neural crest cells

migrate through the somites dong well-characterized pathways in a segmentai fashion

and give rise to numerous derivatives including the sensory and sympathetic ganglia

(Km11 et al., 1995).

At day 10.5, expression is detected in forebrain, posterior midbrain, hindbrain,

craniofacial structures, limb buds and intemaiiy dong the length of the body (Figure

3.5C), indicating that Aqr expression is maintained in neural crest target tissues after their

migration is terminated. To examine the expression pattern in more detail, histological

sections of the lacZ-stained embryos at day 11 were prepared (Figures 3.5E-3-51).

Ventricular zone of neuroepithelium throughout the nervous system was positive for

lacZ. Ventricular expression in neuroepithelium was observed in all staining embryos and

was also confirmed by whole mount in situs (Figure 3.7B). Figures 3.5E to 3.5G present

the expression in the neural tube, telencephaion and diencephalon. Widespread

expression was also detected in limb mesoderm (Figure 3.5G), in the mesenchyme within

the branchial arches (Figure 3.5H), and in paraxial mesodenn at the lower lumbar region

(Figure 3.51).

To determine whether the gene trap vector insertion deregulated Aqr expression.

wild type E9.5 embryos were stained by whole mount RNA in situ hybridization with

Aqr cDNA and compared with lac2 staining of embryos (Figure 3.6, A & B, D & E).

Both lac2 and in situ hybridization patterns recapitulate the expression patterns for Aqr in

forebrain. trigeminal and facioacoustic neural crest tissues, olfactory placode. nasal

process, branchial arches 1 to 3, limb buds and the lateral ridges of neural folds. These

results indicate that the pattern of lac2 expression faithfully reproduces the endogenous

Aqr gene expression pattern. The expression surrounding the otic vesicle and in olfactory

placode was more pronouaced in the whole mount in situs than in the lad-stained

embryos, perhaps due to an unspecific staining rather than a deregulation in gene

expression.

Figure 35. LocZ expression in whde mount embryos and sections.

(A) E8.5: lac2 staining in the cephalic neural fol& (cd). cephalic mesodem (cm), at the

laterd ridges of neural folds including the fused regions in between the caudal (cc) and

rostral (rc) extremities of neural tube closures, somites (s), and paraxial mesodem (pm).

E 10.5: non-staining control (B) and la&-expressing (C) Littermate embryos. Expression

in forebrain, posterior midbrain, hindbrain. craniofacia! structures, Limb buds and

intemaily dong the length of the body. Note the expression in both anterior and postenor

regions of the Limbs (arrowheads). (D) E13.0: forefoot plate, expression in the

circumference of the cartilaginous anlagen of digits. (E-1) Expression in El 1 embryonic

sections. (E) Ventricular zone in the dorsal region of neuroepithelium (ne) in the neural

tube. O Ventx-icular cells of neuroepithelium in telencephalon (tl) and diencephalon (dn).

(G) Widespread expression in limb mesoderm (lm). and in ventricular zone (vz) of

neuroepithelium. (H) Widespread expression in mesenchyme within the fmt branchial

arch. (I) Expression in paraxial mesoderm at lower lumbar region.

Figure 3.6. RNA in süu hybridization recapituiates laeZ expression.

RNA in situ hybridization of whole mount embryos (B & E) recapitulates lac2

expression in E9.5 embryos (A & D). (A & B) Expression in forebrain (fb), trigeminal

neural crest tissue (tg), region surrounding the otic vesicle (ot), facioacoustic neural crest

tissue (fa), olfactory placode (op), nasal process (np), branchial arches 1 to 3 (al&). and

limb bud (lb). @ & E) Expression in the laterai ridges of neural fol& where premigratory

neural crest cells originate (amows). RNA in situ of embryonic sections, (C) E12.5.

sagittal section, expression in neopalliai cortex (neo) , midbrain (mb), infundibulum of

pituitary (ip), Rathke's pouch (rp), cartilage primordium of lumbar vertebral body (cv),

and neural-crest derived structures: intrinsic muscle of tongue (imt) and Meckei's

cartilage (mc). Expression ais0 in liver (li) and lung (ig). (F) E14.5, sagittal section.

expression in neopalliai cortex (neo), infundibulum of pituitary (ip), lung (lg), Liver (li),

cartilage primordium of ribs (cr), thyroid and cricoid cartilages (tc & cc), dorsal root

ganglion (drg), metanep h m (met). and skin (sk).

A q r expression was further delineated by examining day 9 and 9.5 embryonic

sections from whole mount in situs (Figure 3.7). Aqr RNA was expressed in trigeminal

neural crest, branchial arch mesenchyme (Figure 3.7A & 3.7D). laterally migrating neural

crest cells and the ventricular region of the neuroepithelium (Figure 3.7B). Furthemore,

Aqr was expressed in a dynamic and spatiaily regulated rnanner in the limbs, in the zone

of polarking activity (ZPA) at the posterior end of limb bud at day 9.5 (Figure 3.7C).

The ZPA constitutes a group of mesenchyme cells that establish the anteroposterior

patterning of the limb a d express Sunic hedgehog, a key signal involved in limb

patterning (Cohn and Tickle, 1996). At day 10.5, Aqr was expressed in both anterior and

posterior ends of the Lunb bud (Figure 3.5C). and at day 13 expression of Aqr in the foot

plate was restricted to the circumference of the cartilaginous d a g e n of digits (Figure

3.5D). This distinct limb expression suggests a role for A q r in the determination or

patterning of limb structures.

To observe Aqr expression at later stages of embryonic development, we

performed RNA in situ hybridization on embryonic sections at days 12.5 (Figure 3.6C)

and 14.5 (Figure 3.6F). Aqr was expressed in the neopallial cortex, midbrain, the

infundibulum of the pituitary, Rathke's pouch, intrinsic muscle of the tongue, Meckel's

cartilage, thyroid and cricoid cartilages, cartilage primordium of ribs and Iumbar

vertebral body, liver, lung, metanephros, dorsal root ganglion and skin. Some of these

structures, such as Meckel's cartüage and ganglions are derived from neural crest cells

indicating the persistence of Aqr expression during the differentiation of these celis. Aqr

may potentially be a useful marker in following the development of these lineages.

Chromosome Iocalization by FISH analysis

To localize Aqr in the mouse genome, we isolated a phage clone from a genomic mouse

library and used it as a probe in ASH analysis (Figure 3.8). The assignment between

signal from the probe and mouse chromosome 2 was obtained by DAPI banding and

superimposing FISH signals with DAPI banded chromosomes. The detaiied position on

chromosome 2, region E5 to F2, was based on the analysis of 10 mitotic chromosome

spreads. There are huo semidominant mutations, strong's luxoid and tight-&in, in the ES

to F2 region of rnouse chromosome 2 that are potentiaiiy interesting since they exhibit

phenotypes in Aqrexpressing tissues. Strong's luxoid ( k t ) mice present limb phenotypes

including polydactyly, and reductions or duplications of the radius (Lyon et al., 1996).

Tight-skin (Tsk) mice have increased growth of cartilage and bone, tight skin and

hyperplasia of the subcutaneous comective tissue (Lyon et aL, 1996).

Figure 3.7. Expression in sections from whole mount üz sïîus.

Aqr expression in E9-9.5 embryonic sections. (A) Expression in trigeminai neural crest

tissue (arrowheads) just above the rostral extension of dorsal aorta, and in the

mesenchyme within the f m t branchial arch (arrow) around the fmt branchial arch artery.

(B) Expression in the neural crest cells migrating Iaterdly next to surface ectoderm

(arrowheads), and in the venuicular zone of neuroepithelium (mows). (C) Expression in

the posterior end of limb bud (arrow). (D) Expression in the mesenchyme within the fmt

branchial arch, stronger expression in the rostral regions.

Figure 3.8. FISH localization of the Aqr gene.

Localization of the Aqr gene was determined in a set of mouse metaphase chromosome

spread by FXSH. (a) Pi-stained chromosomes with FITC signals. (b) Chromosomes

stained with DAPI to assign signal from the probe with the mouse chromosome 2 bands.

Under the conditions used, the hybridization efficiency was 93% for the probe (93 out of

100 mitotic figures showed signais on one pair of the chromosomes).

DISCUSSION

W e have identified a novel mammalian gene, named Aquuriw (Aqr), which may be

related to RNA-dependent RNA polymerases (RRPs). A q r shares 25% arnino acid

identity over the length of its ORF with murine hepatitis virus RRPs and it has the

conserved active site RRP motif. Our sequence analysis and measurement of alignment

significance indicates that the RRP sequence similarity by itself may or may not be

significant. In addition, very Little is known about the structural and functionai domains

of RRPs and thus the signifcance of amùio acid identities and the presence of the RRP

motif is not clear. Following entry into the host cell and uncoating in the cytoplasrn.

positive-sîranded RNA viruses replicate their genetic matenal by using their (+) RNA

genome as a template to synthesize a compiementary (-) RNA molecule which in tum

serves as a template for the synthesis of progeny genomic (+)-strand RNA. There have

been a few reports suggesting the presence of RNA-dependent RNA synthesis activity in

eukaryotic cells (Schiebel et al., 1993a; Schiebel et al., 1993b; Volloch et al., 199 1). In

addition, although hepatitis delta virus does not encode any polymerase, it can stiIl

replicate autonomously via an RNA-dependent RNA synthesis mechanisrn in animal

cells, raising the possibility that eukiuyotic celis have the enzymatic machinery necessary

to replicate RNA molecules (Lai 1995).

The polymerase subunit of the Tetrahymena telomerase, p95, ais0 yields an

alignment with the polymerase active site of a family of viral RNA-dependent RNA

polymerases (Collins et al., 1995). Telomerase is a specialized polymerase that

synthesizes long repetitive telomeric DNA (Shay 1996). The mouse homolog of the p80

subunit of Tetrahymena telornerase has been identified but the mouse p95 homolog,

which is thought to carry the catalytic polymerase domain, has not been identified

(Harington et al., 1997). Clearly, it WU be of interest to determine whether the Aqr gene

product possesses RNA-dependent RNA polyrnerase activity. The Aqr ORF also carries

two putative nuclear localization signals, suggesting that the Aqr protein is localized to

the cell nucleus.

Dunng embryogenesis, A q r is predominantly expressed in neural crest (NC)

tissues from early on in their ontogeny to their differentiated denvatives in cartilaginous

head structures such as the hyoid bone and Meckel's cartilage. Expression was first

detected in the dorsal edges dong the whole length of the neural groove and in its

adjacent mesenchyrne in the headfold which contains the earliest migrating NC ceils. The

NC is a transient embryonic celi population which arises from the dorsal region of the

neural tube, extensively migrates and gives nse to a number of tissues and structures

including ectomesenchyme which forms most of the skeleton, dermis and connective

tissue of the vertebrate head (Couly et al., 1993; Schilling 1997). Cranid mesoderm and

NC celis are codistributed in the craniofacial mesenchyrne but are distinctly segregated in

the branchial arches, although the interactions of these ceil populations are not yet fuliy

understood (Trainor and Tam, 1995). Aqr expression in NC and craniofacial and

branchial mesoderm suggests a possible role for its gene product in craniofacial

development. Aqr is also expressed in paraxial and branchial mesodenn and other

mesodemal tissues such as the zone of polarizing activity in the limb buds where Sonic

hedgehog signaiing is thought to establish the anteroposterior limb pattern (Pemmon

1995). There is also evidence that Sonic hedgehog cooperates with a RA inducible

cofactor to establish ZPA-like activity (Ogura et al., 1996). The dynarnic expression of

Aquarius in the limbs and its in vitro RA-inducibility suggests a possible role for Aqr in

the development of Limb structures or their patterning.

Through germ line transmission of the trapped locus. we created anirnals

homozygous for the disrupted Aqr gene. Homozygous mutant mice did not present any

obvious abnonnalities. At 10 months of age, they were of the sarne size and weight as

their heterozygous and wild type siblings, and they bred normaily. Newborn skeletons

from homozygotes also did not present any abnonnalities in craniofacial structures. The

absence of any discemible phenotype in mice homozygous for the gene trap allele does

not nile oui the possibility that Aqr plays an essential role in the cells that express i t The

gene trap allele described here may not be a nul1 mutation as Aqr RNA is present in

homozygous animais. The lack of a discemible phenotype is likely due to read-through

transcription and splicing across the trapping vector.

Aqr was identified in a retinoic acid gene trap screen of mouse ES cells. In vitro.

Aqr expression is induced nine-fold in response to RA (Forrester et al., 1996) and its

transcnpt includes an RA-induced Dr repeat element in its 5' end (Sam et al., 1996). We

did not detect RA responsivity of the A q r locus when we examined embryonic lac2

expression in vivo (E9.5-E11). This lack of in vivo RA-responsivity might be due to the

inability to expose embryos to a concentration of RA in vivo that corresponds to the

optimal dose in vin0 (Momss-Kay 1993).

RA is known to induce both aiterations in neural crest cell migration as well as

craniofacial defects (Gale et al., 1996; Lee et al., 1995). Retinoic acid receptor a (RARu)

is expressed in migrating crest cells and in facial mesenchyme (Ruberte et al., 1991).

Also, mutant mice with targeted mutations in both RARa and other RARs have severe

defects in the craniofacial complex (Lohnes et al., 1994). The transcription factor AP-2 is

induced in response to RA, is expressed in neural crest cell lineages and its disruption in

mice also leads to anencephaly and craniofacial defects (Zhang et al., 1996). Aqr

induction in response to RA and its expression in ectomesenchyme presents Aqr as a

candidate gene in an RA signaling pathway that may also play a role in craniofacial

development.

In a search of EST databases, we identified Aqr itself and three other genes

similar to Aqr, two from human and one from the mouse. These genes are highly similar

in sequence to Aqr (88-945 aa identity). The high degree of similarity suggests that they

may be homologous genes in the two species. We did not detect any other sirnilar -

sequences in the GenBank. yeast or C. elegam databases, suggesting thatAqr may be the

founding member of a mammalian-specific gene family.

The availability of a significant number of sequence databases for different

organisms, the utility of analytical software products and the ready access to these tools

through the intemet has led to an increase in gene identification by cornputer (Fickett

1996). Approxirnately 50% of al1 human genes are already represented in the Expressed

Sequence Tag (EST) databases (http://www.ncbi.nlm.nih.gov/dbEST)(Gerhld and

Caskey, 1996). Up to half of ESTs are sequences from the 5' end of cDNAs, and gene

trapping experiments c m also be designed to trap genes in their 5' exons, as was the case

in the present study. Thus, one can envision rapidly obtaining sequence information

about the trapped gene by RACE-PCR cloning of 5' exons, searching the EST databases

for the 5' exon sequences and obtaining the cornrnerciaily available EST clones. This

combination of gene trap experimentation and EST databases provides ready access to

the sequence of trapped genes and has the potential to increase several fold the Pace of

gene discovery and hnctional characterization. At the same time, the gene trapped mice

d o w for expression and functional analyses of the trapped genes.

In sumrnary, in an RA gene trap screen of mouse ES cells, we have identified and

characterized the RA-responsive gene Aquarius. Aqr is developmentally expressed and is

a novel mamrnalian gene with some similarity to RNA-dependent RNA polymerases.

Our results demonstrate the feasibility of in vitro manipulation and screening of ES cells

to rapidly identifi and characterize genes that lie dong a cellular response pathway.

Modification of this gene trap strategy in ES cells should make it possible to trap and

study genes that are either induced by other biological, chernical or physical agents or

that are subject to developmental regulation.

Chapter 4: The Novel Gene Trapped mym Locus 1s Required for

Midgestation Heart Development

Chapter 4 is in preparation for publication as: Sam M and Bernstein A. The Novel Gene Trapped mym Locus 1s Required for Midgestation Heart Development.

ABSTRACT

In a gene trap screen of mouse ernbryonic stem (ES) ceiis, a novel locus named myocyte

maintenance (mym) was identified and characterized. The promoterless lac2 gene was

used to disrupt and trap the genomic locus and to detemine the expression pattern of the

gene. During development, mym is expressed in heart, lirnbs, gut, meninges and the

choroid plexus. Expression was first detected at day 9.5 of gestation in the heart,

branchial arches and craniofacial mesoderm. The dismpted myrn locus was transmitted

through the mouse germ h e in three genetic backgrounds. No homozygous animals were

born in the F2 generation indicating embryonic lethality of the mutant embryos.

Heterozygous males had reduced fertility in a pure genetic background. Homozygous

mutant embryos died during midgestation due to heart failure. Myocardial wall and

trabeculae were diminished in thickness and size, and were hypoplastic due to apoptotic

ce11 death of myocytes. In mutant myocytes, the organization of actin-myosin filaments

and ce11 junctions were disrupted and the expression of myosin light chah isoforms was

deregulated. Our results indicate that mym is required for the maintenance and

organization of heart myocytes during midgestation development.

INTRODUCTION

In vertebrates, the first organ to develop and function in the fetus is the heart. Cardiac

development involves the coordinated interaction and development of three Lineages,

myocytes, endothelial and neural crest cells, to establish the main structures of the heart

by midgestation. Heart morphogenesis has been studied in greatest detail in chick. where

at stage 4. the anterior lateral plate mesoderm ceUs are committed to the cardiogenic

lineage. The bilateral cardiac primordia then fuse and give nse to the primitive heari tube

at stage 10 when the rhythrnic heart beating begins. The heart tube then undergoes

looping where atria and ventricles subsequentiy appear (Litvin et al., 1992). A variety of

genes have been recently implicated in vertebrate heart development (reviewed in Olson

and Srivastava, 1996; and Rossant 1996). A number of these genes are required for the

development of myocardium and heart uabeculae, as determined by gene targeting in

mouse embryonic stem ceiis, and by antisense or dominant negative mutant studies.

Many of the mutant embryos in these studies die during midgestation due to heart failure

at the t h e yolk sac-based circulation is replaced by a cardiovascular circulation. Here we

provide a summary of gene mutations that lead to embryonic lethality as a result of a

defect in the myocardium.

N-Myc is a member of the well-characterized myc nuclear protooncogene family,

and has transcriptionai regulatory hinction. N-myc is expressed at its highest levels in the

expanding primitive streak and embryonic mesoderm, and subsequently in the

myocardium of the cardiac ventricles (Downs et al., 1989; Kato et al.. 199 1 ). Targeted

disruption of N-myc results in embryonic lethality between embryonic day E10.5 and

E12.5 with a primary underdeveloped defective heart retaining the S-shape more typical

of E9 embryos (Charron et al., 1992; Sawai er al., 1993; Stanton et al., 1992). Mutant

hearts contain four chambers but vulvular, septal and trabecular tissues are

underdeveloped, suggesting a defect in endothelium-myocardium interaction which

nomaily lead to epithelium-mesenchymal transformation. Compound heterozygous

fetuses with a leaky and a nul1 mutation in the N-myc locus also die around E13.5 due to

similar heart defects (Moens et al., 1993).

Transcriptional enhancer factor 1 (TEF-1) regulates the cardiac-specific genes,

including troponin C. T. I, and the myosin heavy chah genes (Kariya et al., 1993). TEF-

1 disruption by gene trapping leads to embryonic lethality between El 1 and El2 with an

abnorrnally thin ventricular wall and a reduced number of trabeculae in the heart (Chen et

al., 1994). An enlarged pericardial cavity and a pale yok sac, indicative of poor blood

circulation. are also detected. Transcription of TEF-1-regulated genes in the heart,

however, appears normal and heart development is not otherwise compromised in the

TEF- 1 mutant fetus.

The homeobox gene Nkr2-5, the murine homologue of the Drosophila gene

tinmnn, is the earliest known rnyogenic transcription factor (Lints TI et al., 1993). It is

expressed in myocardiogenic progenitor cells and continues to be expressed in

embryonic, fetal and adult cardiornyocytes. It is also expressed in the future pharyngeal

endoderm, the tissue with presumptive heart inducing activity. Murine embryos lacking

NkcZ-5 exhibit abnormal heart morphogenesis, growth retardation and die during

rnidgestation apparently from hemodynamic insufficiency (Lyons et al., 1995). The

process of looping morphogenesis is not initiated in these embryos and trabeculation and

endocardial cushion formation are also blocked. The myosin light-chain 2V gene, the

earliest known molecular marker of ventricular differentiation, is not expressed in N M - 5

mutant hearts, while myosin heavy-chain B and cyclin D2, two other ventricle-specific

genes, were expressed normally. Nkr2-5 is thus a component of a geoetic pathway

required for myogenic specialization of the ventricles.

The Wilms' tumor-associated gene, W-1, is a zinc fmger transcription factor

expressed as early as E9 in intermediate mesoderm and subsequently in differentiating

mesothelium, spinal cord, brain and urogenital ndge derivatives. Targeted mutation of

WT-I causes thin ventricular walls, frequent pericardial bleeding and systemic edema

(Kreidberg et al., 1993). Mutant fetuses die ai day 14.5 of gestation likely due to cardiac

dysfunction. Similarly, hypoplastic development of the ventricular chambers was

observed in homozygous mutant fetuses lacking the ligand-dependent nuclear receptor

transcription factor, retinoid X receptor (RXRa), and dying between E13.5 and E 16.5

(Sucov et al., 1994). RXRa mutant fetuses also have a poorly developed ventricular

septum.

Fibroblast growth factor (FGF) and the FGF receptor (FGFR) are expressed at

high levels early in cardiogenesis but their expression later declines as the mitotic activity

of myocytes decreases (Consigli and Joseph-Silverstein, 1991; Engelmann et al., 1993;

Parlow et al., 199 1; Peters et al., 1992). Antisense RNA and dominant negative mutants

of FGFR type 1 (FGFRI) inhibit myocyte proliferation or survival dunng the f m t week

of chicken embryonic development, but had much less effect after the second week,

suggesting that FGF signaling rnay regulate myocyte growth during tubular stages of

cardiogenesis (Mima et al., 1995). Neuregulin growth factors and their receptors ErbB2

and ErbB4 are involved in the paracrine mechanism for intercellular communication in

embryonic cardiac and neural development (Canaway 1996). Hornozygous mutations in

neuregulin, erbB2, and erbB4 genes lead to embryonic lethality before E 1 1 due to heart

malformations identical to each other (Gassmann et al., 1995; Lee et al., 1995; Meyer

and Birchmeier, 1995). Ln these three gene mutant mice, ventricle morphology is

abnormal due to a complete absence of rnyocardial trabeculae. Trabeculae are the first

anatomical feature of ventricle differentiation and are responsible for the maintenance of

blood circulation during early stages of heart development before expansion of the

compact zone. The endocardial cushion is also hypoplastic in al1 three homozygous

mutant embryos.

The neurofibromatosis type 1 gene, NFI, is a member of the GAP f d y of Ras

regulatory proteins involved in a Ras-dependent signal transduction pathway. NF1 is

highly expressed in the fetai heart (Gutrnann et al., 1993). A targeted nul1 mutation of

mouse NF2 leads to fetal lethality at E13.5 in homozygotes, with generalized edema and

abnormal cardiac development where both great vessels, aorta and pulmonary artery,

emerge from the right ventncle instead of the aorta exiting normdly from the left

ventricle (Brannan et aL, 1994; Jacks et al., 1994). The myocardium, partkulariy in the

ventricles, is lacy in appearance and hypoplastic. and there is a significant venvicular

septal defect.

Tyrosine hydroxytase is an enzyme in the catecholamine (neurotransmitters and

hormones like dopamine and noradrenaiine) biosynthetic pathway. The inactivation of

both its alleles in the mouse results in lethality between El 1.5 and El55 due to

cardiovascular failure (Zhou et al., 1995). The atria of mutant fetuses at E12.5 are

markedly dilated, and their walls are reduced to one or two cell layers. Ventricular

cardiornyocytes are less organized, heterogeneous in size and more vacuolated. Targeted

mutation in noradrenaline also leads to fetal lethaüty at E13.5 with a heart phenotype

very similar to tyrosine hydroxylase mutant fetuses (Thomas et al., 1995). indicating a

requirement for catecholamines for early fetal development and that the heart may be an

important target for their action. The absence of striking morphological defects in these

mutant fetuses suggest that the primary effect may be physiological such as an inability

to sustain sufficient heart rate or contractility.

Plukuglobin is a component of desrnosomal plaques and cadherin-catenin ce11

adhesion complexes in ce11 junctions. Its inactivation by homologous recombination in

mice leads to embryonic death after E10.5 due to severe heart defects including

hypoplastic ventricular wall, trabeculae. and endocardiai cushion, and bleeding into the

pericardial cavity (Bierkamp et al., 1996). Ultrastnictudy, the number of desmosomes

is greatly reduced and their intercellular cement (desmogleas) is atypical in appearance.

Vascular celi adhesion molecule 1 (VCAM- 1) is a type 1 trammembrane protein

belonging to the immunoglobulin superfamily. Half of the VCAM- 1-deficient embryos

die before E 1 1.5 fiom a severe placental defect, while the other half survive to E 12.5 and

die from heart failure due to a reduction of the compact layer of the ventricular

myocardium and intraventricular septum (Kwee et al., 1995). The mutant hearts also

lacked an epicardium. Mutant homozygous embryos Iack the ce11 surface adhesion

protein N-Cadherin, and die at El0 with a dramatic cell adhesion defect in myocardial

tissue. Mutant myocardium initially forms but subsequently dissociate and as a result the

heart tube f i l s to develop normally (Radice et al., 1997). Targeted mutation in the zinc

finger Evil proto-oncogene also leads to embryonic lethality around E10.5 with abnormal

heart development (Hoyt et al.. 1997). The mutant hearts are poorly developed with a

definite looping defect and a retarded constriction between atria and venûicle.

As surnmarized above, functional mutations in several transcription factors,

growth factors and their receptors, signaling regdatory proteins, metabolic enzymes and

structural proteins lead to abnormal midgestation myocardium development and fetal

lethality. The complexity of ceii lineage interactions and tissue morphogenesis required

for cardiogenesis, and the incompatibility of the absence of cardiovascular function with

Life in midgestation development explains the myocardial defects and the lethality of

mutations discovered in such a wide variety of genes.

Using a gene trap strategy in mouse ernbryonic stem (ES) cells. we have disrupted

and identified the myocyte maintenance (mym) locus that is essential for the formation of

the myocardial wail and the heart trabeculae. mym mutant embryos have disorganized

hearts with hypoplastic myocardiurn which is diminished in thickness and size. In mutant

myocytes, actin-myosin Naments and cell junctions are disrupted and the cells undergo

apoptotic ce11 death.

MATERIALS AND METHODS

Gene trapped embryonic stem (ES) ce11 lines, hcZ staining and germ 1ine

transmission. Construction, handling and characterization of the R140 gene trapped ES

ce11 line, lacZ staining and the transmission of the trapped locus through the mouse germ

line have been descnbed (Forrester et al., 1996).

Mouse and embryo genotyping and F2 breeding in àiff'erent backgrounds.

Genomic DNA was isolated from tails or yolk sac, digested with the EcoRl restriction

enzyme and separated by electrophoresis on a 1% agarose gel. Southem blots were

double probed with en-2 and lac2 probes. En-2 was used as an internai control for

quantitation and the l a d probe detected the transgeae in the mym locus. Densitometric

analyses were used to mesure the intensity of the lac2 signal relative to the en-2 signal.

Three F1 males and six F1 females, heterozygous for the disrupted locus. were bred to

obtain F2 progeny in 50% 129sv-cp/50% C57BU6 outbred background. The germ line

chimeric males were also bred to CD-1 and 129 wild type females to obtain 50% CD-

1150% 129, and 100% 129 backgrounds. Three males and six femaies from these

backgrounds were dso bred to obtain F;! progeny. Heterozygotes (+/-) had a LacZen-2

signal ratio of 1, and homozygotes (-1-) had a laczen-2 signal ratio of 2 on Southem

blots. To c o n f m the genotyping, four heterozygotes were backcrossed to wild type

animais and their progeny were genotyped for the presence or absence of the transgene.

In al1 four cases, haif of the progeny carried the transgene confirming heterozygosity of

the mym locus in al1 tested animais carrying the transgene.

RACE, PCR amplification, cloning of the 5' rnym exon. 5' RACE-PCR on

fusion transcnpts from the gene trapped ES ce11 line R140 was performed using a 5'

RACE kit (Life Technologies, Burlington, On), according to the manufacturer's

instructions, with the following modifications: Reverse transcription was carried out at

42OC for 30 min in the presence of [ ~ - ~ Z P ] ~ C T P to monitor the synthesis of fmt strand

cDNA. Superscript II (Life Technologies) and GTl a cZ-1 [5'-

GCAAGGCGATTAAGTTGGGT-31 primers were used for reverse transcription.

DNA amplification of the RACE products was perfonned in two rounds in 50-pi

volumes containing 1 X PCR Buffer 11 (Perkin Elrner, Foster City, Ca), 1.25 mM MgC12,

200 mM deoxynucleoside triphosphates, 200 n M of each primer, and 2.0 U of AmpliTaq

DNA Polymerase (Perkin Elmer) in a DNA Thermal Cycler (Perkin Elmer). Cycle

rotations were as follows, 1 cycle of 94°C for 5 min; 1 cycle of 80°C for 8 min during

which Taq polymerase was added to the reaction mix; 35 cycles of 94OC for 1 min, 60°C

for 2 min, and 72°C for 3 min; 1 cycle of 72OC for 5 min). The second round of

amplification was performed under the same conditions using nested primers (see below)

and 5 ml of the reaction product from the first round. The pnmers used for DNA

amplification were as follows: for the first round, GTlacZ-2A [S-

CCGTCGACTCTGGCGCCGCTGCTCTGTCAG-3'1 and Anchor 15'-

GGCCACGCGTCGACTAGTACGGGIIGGGLIGGGIIG-3'1 (Life Technologies); for the

second round, Nested2AU (5'-CAUCAUCAUCAUTTGTCGACCTGT

TGGTCTGAAACTCAGCCT-3'1, and Universal Amplification Primer [5'-

CUACUACUACUAGGCCACGCGTCGACTAGTAC-3'1 (Life Technologies,

Burlington. On). The RACE-PCR products were digested with San and cloned into

pBlueScript II vector (Stratagene, La lolla, Ca). Altematively, they were cloned into the

pAMPl plasmid using CloneArnp System (Life Technologies). The cDNA inserts were

sequenced on both strands using the AutoRead Squencing kit (Pharmacia. Baie d'Urfe,

Qc) on an A.L.F. DNA Sequencer (Pharmacia).

Genomic PCR, and FISH. PCR amplification on genomic DNA from +/+, +/-.

and -1- embryos was performed with primen flanking the 5' mym exon, a [5'-

GCTCTTGTCCTAGAAGCC-3'1, and b [5'-CAAAGGGCAAACCCAAAA-3'1 and the

GTlacZ- 1 primer (above). PCR conditions were as described above. A genomic clone for

mym was obtained and used as a probe for fluorescence in situ hybridization (FISH)

analysis. FISH was performed according to published procedures (Heng et al., 1992;

Heng and Tsui, 1993).

RNA in situ hybridization. RNA in situ hybridization was performed as

described (Hogan et al., 1994). Bnefly, fixed tissues were cryostat-sectioned at 10 pm

thickness, mounted on glas slides and refixed in 4% pacafonnaldehyde. Prehybridization

treatrnents were performed as described (Hogan et al., 1994). 3%-labeled single stranded

RNA probes were prepared using T3 and T7 RNA polymerases (Boehringer Mannheim,

Laval. Qc). Adjacent sections from wild-type and homozygous mutant hearts were

hybridized with different myocyte and endothelid marker probes (obtained from J. Cross

and J. Rossant). The probes were a-cardiac actin (Kumar et aL, 1997), atrïd myosin light

chah I & 2 (MLCIA & -2A, Kubalak et al., 1994; Lyons et al., 1990). ventricle myosin

light chah 2 (MU72V, Lee et ai., 1992), muscle-specific isoforrn of adenylosuccinate

synthetase (Adssl, Lewis et al., 1996), and Flk-I (Yamaguchi et al., 1993). Post

hybndization washings included treatment with 50 pg/ml RNase A (Sigma, St. Louis.

Mo) at 37°C for 30 min. Following dehydration, the slides were dipped into NTB-2 film

emulsion (Kodak, Rochester, NY), exposed at 4OC for 6 days, developed, and stained

with toluidine blue.

Scanning electron microscopy. Whole embryos were fixed in 2% glutaraldehyde

in O. 1 M sodium cacodylate buffer, post-fixed in 1% osmium tetroxide in 0.1 M sodium

cacodylate buffer, dehydrated in a graded ethanol senes followed by propylene oxide,

and embedded in Spurr epoxy resin. Sections 100 nm thick were cut on an RMC MT6000

ultramicrotome, stained with lead citrate and uranyl acetate. and viewed on a Philips

EM430 transmission electron microscope.

RESULTS

The mym gene was originally identified in a gene trap screen of totipotent mouse ES ceus

bat was designed to capture genes responsive to exogenous retinoic acid (RA) (Forrester

et al., 1996). The gene trap constnict used in this study carried a splice acceptor site 5' of

a promoterless lac2 reporter gene randomly integrated into the mouse genome by

electroporation. The R140 insertion characterized here is one of 20 gene trapped ES

clones that were isolated and characterized iùrther for their patterns of lac2 expression

during embryonic development.

RI40 gene trap ES cells and RACE-PCR products. The R140 ES clone was

originally thought to be repressed in response to RA (Forrester et al., 1996). We

examined lac2 staining of cells in the presence and absence of RA and leukemia

inhibitory factor (Lm, and observed that repression of lac2 activity was minimal and

most Likely due to the absence of LIF in the media, and not the presence of RA (data not

shown). Nevertheless, due to the restncted expression pattern of the trapped locus in the

embryonic heart, we were interested in characterizing the R140 clone hirther.

To obtain sequence information about the trapped gene, the 5' rapid amplification

of cDNA ends-polymerase chah reaction (RACE-PCR) technique was employed on total

cellular RNA isolated from the RI40 clone. A O 4 Kb cDNA was cloned and sequenced.

The 3' 200 nucleotides of the cDNA came from the gene trap vector itself. The 5' half

was novel, had the beginning of an open reading frame and was used as a probe for

cDNA and genornic library screening and used to design primers for further 5' and 3'

RACE-PCR. Further 5' RACE on total cDNA from El2 embryos provided us with an

extra 200 bases of sequence for a total of 400 bp (Figure 4.1, GenBank accession:

AF019615). The partial cDNA contains initiation codon and the begining of an open

reading frarne (ORF), 14 amino acid residues in length. The amino acid stretch is novel

and does not match any sequences in the database. To date, we have not obtained the full

length mym cDNA.

Genomic disruption and chromosome localization by FISH analysis. To

confinn genomic disruption of the mym locus by the gene trap vector, genomic DNA

from +/+, +/- and -/- embryos was ampiified using primes from the mym exon 5' of the

IacZ insertion and the l a d sequences (Figure 4.2C). We obtained amplified recombinant

DNA frorn +/- and -1-, but not from +/+ embryos, indicating the disruption of the mym

gene by the lac2 sequences (Figure 4.2D).

To localize the trapped locus in the mouse gemme, we isolated a genomic clone

and used it as a probe for FISH analysis (Figure 4.2E-2G). The gene was localized on

chromosome 5, region B3. A spontaneous recessive lethal mutation (15Hl) has also been

mapped to this region close to the W locus (Lyon and Glenister, 1982). wlgX a radiation

induced allele of W aiso leads to preimplantation embryonic lethality in homozygotes

(Lyon et al., 1984). W19H is a large deletion covering Kit, the Ph locus, and a recessive

lethai located 2 CM distal to W. ~ 1 9 H mice may be useful in crosses with mym mice for

complementation and mapping expenments.

Lac2 expression pattern of the mym trapped locus. The pattern of mym

expression was examined during embryogenesis by staining for lac2 expression (Figure

4.3). At days 9.5 to 1 1.5 of gestation (E9.5- 1 1 S), expression was observed in the heart,

gut, craniofacial structures, otic vesicle and the apical ectodermal region of the lirnb buds

(Figures 4.3A-E & 4.4). At E12.5-14.5, expression was detected in the leptomeninges

(Figure 4.3D), atrium and ventricle of the kart (Figure 4.3F), choroid plexus (Figures

4.3G & 4.4C) and in the undifferentiated mesoderm of the limbs (Figure 4.3H). mym was

also expressed in the adult hippocampus and the Purkinje celi layer of the cerebellar

cortex (Figures 4.31-J). At El 1.5, lac2 expression in the heart was locdized in the

ventricle and atriai myocardium as well as in the trabeculae (Figures 4.4A-B), in the

ependymai cells of the choroid invagination (Figure 4.4C), and in the rnidgut. hindgut

and urogenital ridge (Figure 4.4D).

Figure 4.1. Nucleotide sequence of the partial mym cDNA. Partial cDNA sequence

was obtained in two rounds of 3' RACE-PCR. Partial open reading frame starting with a

methionine codon is translated to amino acid residues towards the 3' end of the sequence.

The splice junction where vector sequences began is indicated by an arrow.

400 * * * * *

TGGATCTGCACTCTCCCGGATGTCCTCTGGTGCTCAAAGACC~ M S S G A Q R P F W

Figure 4.2. Southem genotyping, gene trap insertion in the mym locus and FISH

analysis. (A) Southem blot. The upper band, derived from the en-2 gene, is used as

internai control for DNA Ioading. The lower band is derived from the lac2 transgene.

The ratio of intensity of the lac2 band (peak 2 in B) to the en-2 band (peak 1 in B)

indicates the genotype of the animal: +/+ (ratio = O, lanes 1, 6 & 9), +/- (ratio = 1, lanes

3-5 & 7), and -/- (ratio = 2, lanes 2, 8 & 10). (B) An example of densitometric

measurements of the band intensities for +/+, +/- and -/- genotypes. Peak 1 : en-2 band,

peak 2: lac2 band. (C) Lac2 disrupts the rnym genomic locus. mym 5' exon was cloned

by 5' RACE-PCR. The primers a. b, and c were used for PCR on genomic DNA to

c o n f m disruption of the myrn locus by the gene trap vector. (D) PCR on genomic DNA

from +/+, +/- and -/- embryos and Southem blot probed with the myrn exon. (E-G)

Localization of the myrn gene in a set of mouse metaphase chromosome spread was

performed by FISH. (E) PI-stained chromosomes with FITC signals, (F) same

chromosomes stained with DAPI, (G) diagrammatic view of the banding pattern of the

mouse chromosome 5 with the corresponding mym signals in the B3 region.

Figure 4.3. Whole mount embryonic and adult lac2 expression. (A) Expression at

E9.5 & (B) El0.5 in the hem, gut, craniofacial structures and the otic vesicle, (C) E l 1.0,

expression in the heart and gut, (D) E13.5 and E14.5, expression in the leptomeninges

covering the whole central nervous system, (E) E10.5, expression in the apical

ectodermal region of limb buds (arrowhead), (F) El 1.5, expression in the heart, (G)

E 11 .S. expression in the choroid plexus (arrowhead), (H) E13.5, in the undifferentiated

mesoderm of the limbs (arrow), (1) expression in the adult hippocarnpus (hp), and (J) in

the Purkinje ceil layer (pcl) of the cerebeilar cortex.

G e m line transmission and analysis of F2 progeny. The mym trapped locus

was transmitted through the germ line of mouse chimeras to produce heterozygous

animals. Heterozygotes were identified by Southern analysis and bred to obtain F2

progeny in three different genetic backgrounds (see Materials and Methods). Genotyping

was perfomed using a quantitative Southem blot analysis (Figure 4.2A & 4.2B). Out of

263 animals in the F2 progeny of al1 three backgrounds, no homozygotes were detected

(Table 4.1). Therefore, the mym gene is required to complete ernbryogenesis.

129 males show reduced ferolity. Heterozygous males in the pure 129 genetic

background were sterile or had reduced fertility resulting in few progeny. The penetrance

of sterility/reduced fertility was 82% (ntotal=17). Mating behavior of the 129

heterozygous males was however normal, as they were able to plug fernales. To

understand the cause of reduced fertility of heterozygous males in the 129 background,

we examined histological sections of their sexual organs and found that spermatogenesis

was severely affected (Figure 4.5). In the testis, the seminiferous tubules structure was

loose and disorganized. and spermatid flagella were ruptured. In the epididymis where

sperms mature, there were few or no spermatozoa present.

Figure 4.4. Lac2 expression in embryonie sections at El15 (A & B) LocZ staining in

the ventrïcle (v) and atrial (a) myocardiurn, in the myocardial wall (mw) and in the

trabeculae (tb); note the organization of endothelid cells (en) surrounding the trabeculae.

(C) Expression in the ependymal ceils (ec) of the choroid invagination and (D) in midgut

(mg), hindgut (hg) and urogenitd ridge (ur).

Figure 4.5. Bistotogical sections of heterozygous male sexual organs.

Spematogenesis is affected in the 129 heterozygous males. (A & C) +/+ males, and (B &

D) +/- males. (A & B) Testis. Seminiferous epithelium (se), cornposed of Sertoli and

spermatogenic cells, and flagella (0. In the +/- males. the seminiferous tubules structure

are loose and disorganized, and spermatid flagella are ruptured. (C & D) Epididymis.

Epithelium (e), spermatozoa (sp), connective tissue (ct), and stereocilia (SC). No

spermatozoa is present in the epididymis of +/- males.

Bomozygous embryos die during midgestation. To understand the time and the

cause of embryonic lethality more precisely, we undertook systematic dissection of

embryos at various gestation t h e points (Table 4.1). Due to reduced fertility in the 129

background, we performed al1 embryonic phenotypic analyses in the two mixed genetic

backgrounds. Upon dissection of rnidgestation uteruses from the heterozygote breedings,

we noticed some of the conceptuses were about half the size of the others (Figure 4.6A).

We dissected out the embryos at E l 0 and used their yolk sac for isolathg genomic DNA

for genotyping. The retarded conceptuses were homozygous for the gene trapped locus

and were of two distinct phenotypes. About one third of the homozygous ernbryos were

very small, grossly abnormai in appearance and stalled at the headfold stage (Figures

4.6C-D). Histological sections revealed that these embryos had rather well-developed

organs and structures such as yolk sac, head folds, neural tube and heart tube (Figures

4.6E-H). Such grossly abnormai homozygous embryos were detected as early as E9.0,

but never after E 10.5, most likely due to resorption.

The remaining two thirds of the homozygous embryos were about half the size of

their wild type littermates, and presented a characteristic heart phenotype with differing

degrees of severity (Figures 4.6B-C & 4.7). The less severely affected embryos had a

beating har t at the time of dissection. We recovered three embryos which had the heart

phenotype and were scored as heterozygotes, but these were likeiy due to misgenotyping

by quantitative Southem. We did not detect any live homozygous embryos beyond

E10.5, suggesting embryonic lethality during this stage of development (Table 4.1).

Table 4.1. Genetic aaalysis of the mym gene trap insertion. Genotyping o f embryos

was done between days 9 and 11.5 of gestation (E9-E11.5). Genotyping of animais was

done at 4 weeks of age (P28).

Stage Total

-1- Embryos Genotype

Grossly Heart +/+ +/- abnomala defectb Resorbedc

aThe embryos were highly deformed and deveIopmenta1Iy delayed. b ~ h e embryos were reduced in size and presented the typical heart phenotype. CIncludes deteriorated embryos.

Figure 4.6. Phenotypes of n?ym homozygous embryos. (A) Midgestation uterus

dissected out frorn heterozygous breeding showing the smdler mutant conceptuses

(mwheads). (B) A wild type embryo at E10, and its mutant Littemate (on the nght). (C)

One delayed embryo with heart defect on the left and a grossly abnormal mutant embryo

on the right. These two mutant embryos were also Littermates. E10. (D) Two other

grossly mutant embryos. headfolds are indicated by arrowheads. (E-H) Histological

sections of the grossly abnormal embryos, indicated are: headfold (arrowheads), neural

tube (nt), y o k sac (ys). heart tube (hi). foregut (fg) and dorsal aorta (da).

mym is required for kart development. To examine the general rnorphology of

mutant hearts, wild type and mutant hearts at E1O.O were prepared for and micrographed

by scanning electron microscopy. The mutant hearts were of a variably irregular shape

(Figures 4.7A-B). Histological analyses were performed on haematoxylin and eosin-

stained paraffin sections of wild type and mutant hearts and micrographed under W for

higher resolution (Figures 4.7C-E). Atrium and ventricle, as well as the three ce11

lineages, endocardium, myocardium and pericardium were present in the mutant hearts;

however, their organization was af5ected. The myocardial waii was only one or two cell

layers thick, and the trabeculae, a spongiform network of myocytes infdterated by the

endocardium, were also thin and not well-organized throughout the heart including the

bulbus cordis. There was also no clearly distinguishable endocardium surrounding the

trabeculae. The endocardial cavity was almost absent and there was blood trapped in both

atrium and ventricle indicating a failure in cardiovascular circulation (Figure 4.7D).

To understand the progression of the phenotype in the heart, systematic

heterozygote test breeduigs were set up and embryos were dissected out at embryonic

days 9-9.5, 10.0 and 10.5. Wild type and homozygous mutant embryos were sectioned

and their hearts examined and compared under light microscopy (Figure 4.8). At E9.0,

wild type hearts are highly Looped and their wails consist of myocardial and endocardial

Layers separated by cardiac jelly (Viragh and Challice, 1977). Mutant hearts were as well-

formed and normal in appearance as their wild type littermates at E9.0, except for some

disorganization of the trabeculae and a larger space between the interacting endocardium

and myocardium (Figure 4.8A, 4.8D & 4.8E). At E9.5, however, a distinct phenotype

was clear in the myocardium. The trabeculae and myocardial wall were disorganized and

the pericardial layer was loosely attached to the underlying myocardium (Figure 4.8H).

At ElO, the disarray in myocardial wall and trabeculae was more severe and the heart

retained its less developed tube-like appearance. The endocardial cells also did not form

the orderly structures around the trabeculae and endocardial cushion formation was

underdeveloped (Figure 4.8F). in some instances, part of the heart tube had cornpletely

coilapsed (Figure 4-81). At EL0.5, the severity of the myocardial defect affected the

integrity of the myocardial wall, leading to bleeding in the pericardial cavity and

disruption of blood circulation (Figure 4.8G). Some mutant embryos had already started

deteriorating at this stage (Figure 4.8J).

Figure 4.7. Embryoaic heart phenotype. General morphology of +/+ and 4- hearts at

E 10. Scanning electron microscopy of: (A) a +/+ wild type heart, myocardial wall (mw),

trabeculae (tb), endothelium (en), and endocardial cavity (ec); and (B) a -/- mutant heart

in which tb, en and ec are not easily recognizable. UV transiliumination of histological

sections of: (C) a wild type heart; and (D & E) mutant hearts, (v) ventricle. Note the

b l d trapped in the atrium (a), and the disorganization of the trabeculae (tb) in D & E.

Figure 4.8. Time series analysis of homoygous embryonic hearts. Progression of the

phenotype in mutant hearts between E9.0 to El05 (A-C) +/+ conml hearts, (D-J) -1-

mutant hearts. Atrium (a), ventricle (v), pericardium (p), myocardium (m). endocardium

(e), pericardial cavity (pc), tnincus arteriosus (ta), trabeculation (tb). and endocardial

cushion (ec) are indicated. Note the underdeveloped endocardial cushion and the tube-

like structure in the E 10 heart in F. See the text for M e r description.

Expression of myocyte-spitic genes in nzym mutant hearts. To understand

the defect in the myocardium and the stage of myocyte differentiation in the mutant

hearts. we examined the expression of myocyte-specific genes at E l 0 by RNA in situ

hybridization (Figure 4.9). Mutant myocytes, Like their wild type counterparts, expressed

differentiated myocyte marken such as a-cardiac actin, myosin light chain LA and the

muscle-specific isoform of adenylosuccinate synthetase (Figures 4.9E-I), but they failed

to express myofilarnent genes, myosin light chains 2A and 2V (MLCZA & -2V) (Figures

4.9A-D). MLC2A was expressed in both ventricles and atrium, while MLC2V expression

was restricted to ventricles in wild type hearts. A few cells in the mutant ventricles may

stiil be expressing MLC2V (Figure 4.9D), while MLC2A expression was totaiiy absent

(Figure 4.9B). Thus, in the absence of mym function. the MLC2A & MLCZV genes were

not activated. Expression of the endothelid-specific receptor tyrosine kinase, Flk-1. was

also examined to visualize the organization of the endothelium in the mutant hearts

(Figures 4.9K-L). The endothelid cells in the mutant ventricles were dispersed and less

orderly than their counterpart in the wild type heart, likely due to the disorganization of

the myocardium and trabeculae.

Failure in myocyte maintenance. We next examined the structural organization

of myocytes at El0 by transmission electron microscopy (Figure 4.10). Cardiac myocyte

ultrastructural features include abundant mitochondria, extensive deposition of

myofilarnent arrays and specialized intercellular junctions, desmosomes and facia

adherens with anchoring rnyofibrils (Hirakow and Sugi. 1990). Mutant myocytes failed

to organize actin-myosin and thin filaments in their cytoplasm and there was no evidence

of ce11 junctions between mutant myocytes. The less severely affected embryos with

beating hearts at the time of dissection had evidence of ce11 junctions and some actin-

rnyosin contractile elements in their cardiac myocytes (not shown). It seems that myocyte

ultrastructural defects are the underlying cause of cardiac failure in mutant embryos,

however, its severity is likely influenced by modifier genes in the mixed genetic

backgrounds used in these phenotypic analyses. We also detected the presence of

condensed cytoplasm, apoptotic bodies and electron dense nuclei in the mutant

myocardium indicative of apoptotic ce11 death (Figures 4.lOC-D).

Figure 4.9. Heart-speclfic gene expression by RNA in silu hybridization. E 10 sagittai

sections of +/+ control, and -1- mutant hearts. Antisense probes used: (A & B) myosin

Light chah 2A, (C & D) rnyosin light chah 2V. (E & F) myosin light chain 1A, (G & H)

acardiac actin, (1 & J) muscle-specific isoform of adenylosuccinate synthetase, (K & L)

flk-l. Myosin Iight chains 2A and 2V are downregulated in -/- mutant hearis (B & D).

Figure 4.10. Structural organization of myoeytes analyzed by transmission eiectron

microscopy. (A & B) +/+ control rnyocytes and (C & D) -/- mutant myocytes. hdicated

are: myofibrillae (rn), Z lines of actin-myosin filaments (arrowheads), nucleus (n),

cytoplasm (c) , sarcoplasmic reticulum (sr), thin filament (tf). desmosome (d), apoptotic

nucleus (an), apoptotic body (ab), and condensed cytoplasm (cc). Note the absence of ce11

junctions between myocytes in C and in between the two arrows in D where membranes

of neighboring cells are in close proximity. Magnifications: (A) X6,500; (B) X50,OOO;

(C) X6,ûûû; (D) X 1 5,000.

DISCUSSION

In a gene trap screen of rnouse embryonic stem cells, we have identified a novel gene.

myrn, that is required for mpcyte maintenance during embryonic heart development.

mym is expressed in the heart, gut, craniofacial structures and h b buds. Expression in

the heart is first detected at day 9.0 of gestation in the myocardium. At day 10, myrn is

expressed in the myocardial wall and the trabeculae where the phenotype of the

homozygous mutant embryos cm fmt be detected during development. The gene trapped

mym locus is a recessive lethal mutation. Two thirds of the homozygous mutant embryos

die during midgestation due to cardiovascular failure. In the mutant hearts the

myocardium is hypoplastic resulting in myocardial wall and trabeculae that are thin,

reduced in size and disorganized. Myocardial defects often Iead to trapping of the blood

in the endocardial cavity and bleeding into the pericardial cavity. Both of these defects

were observed in the myrn mutant hearts. At E10.5, necrosis is evident in some mutant

embryos indicating a progression toward resorption. No live mutant embryos were

observed after E 1 1 .O, indicating day 1 1 .O of gestation as the time of embryonic arrest.

AU cardiac morphogenetic processes such as looping, formation of the chambers,

vaive, and endocardial cushion, and trabeculation are initiated in myrn mutant hearts.

However, their further development is generally arrested due to a defect in myocyte

maintenance. The earliest signs of myocardial defects were seen at day 9.0 when sorne

disorganization in the trabecule structure first became apparent. The impairment

progressively worsened until day 10.5 of gestation when some mutant embryos were

already dead and had started deteriorating. At this stage of development, when the yolk

sac-based circulation is being replaced by cardiovascular circulation, the heart is under

considerable mechanical load to supply the embryo with blood (Kaufman 1992). Any

structural defects that may compromise the integrity of cardiovascular circulation at this

stage are going to be detrimental to embryonic life. This explains why mutations in many

genes involved in cardiovascular development or function lead to embryonic Iethaiity

during the midgestation (see Introduction).

At day 10.25 to 10.5 of gestation, two obvious morphogenic changes take place

within the heart; i) left and right components of the cornmon atrial chamber begin to

separate as the fmt clear indication of the septum primum is seen, and ii) there is marked

trabeculation in the bulbus cordis (Kaufman 1992). Both of these changes are directly

related to myocardial proliferation which is severely compromised in the rnym mutant

embryos. In the mym mutant hearts, we observed that myocytes are disorganized, their

ceIl numbers are reduced and trabeculation is severed, precluding the landmark

midgestation developmental changes in the heart to take place.

mym mutant rnyocytes express differentiated iineage gene markers, myosin light

chain 1 A (MLC 1 A), a-cardiac actin and the muscle-specifc isoform of adenyiosuccinate

synthetase, indicating that they have initiated and progressed dong their differentiation

pathway. In contrast, myrn myocytes failed to express myosin Light chahs 2A and 2V

(MLCZA & -2 V), demonstrating that mym is required for a component of the cardiac

myogenic pathway that is not essential for the expression of the other myosin and actin

genes. The expression of different myosin isoforms during cardiac development is

suspected to be under different regulatory mechanisms as their differential and dynamic

patterns of expression change at different rates during development (Lyons et al., 1990).

MLClA and -2A expression are nonnaily restricted to the atria, while MLC2V

expression is rzstricted to the ventricular chambers (Kubalak et al., 1994; Lee et al.,

1992; Lyons et al., 1990). The fact that MLC2A and -2 V, but not MLClA expression is

deregulated indicates that rnym is not affecting myogenesis in a chamber-specific manner.

Different myosin genes perform different physiological roles in cardiac muscle cells

(Malhotra et al., 1979). Lack of MLC2A and -2V expression in the myrn hearts has Wtely

compromised myocardial function, and thereby blood circulation, leading to the death of

mutant embryos. Interestingly, the anthracycline antibiotic doxorubicin, used in cancer

therapy, decreases in a dose-dependent manner the levels of MLC2 gene expression in

cardiac muscle and as well leads to a myofibrillar loss in a characteristic cardiac

myopathy (Ito et al., 1990). The deregulation of MLC2 gene expression and its

concomitant myocardial defect in doxorubicin treatment is similar to the absence of mym

gene hnction presented here.

MLC2V expression is also inactivated in Nkx2-5 mutant hearts, while the

expression of other myofilarnent genes remain intact (Lyons et al., 1995). NkrZ-5 is a

homeobox-containing transcriptional regulator and a marker of the early stages of

cardiogenesis (Harvey 1996). It is homologous to the Drosophila tinrnan gene and its

targeted interruption leads to midgestation embryonic lethality due to abnormal heart

morphogenesis. The myocardial and trabecule defects in the NkxZ-5 mutant hearts are

very similar to the defects in mym mutant hearts presented here. Myofib~ogenesis and

the ultrastnictural organization of the myocytes are not, however, compromised in the

Nkx2-5 mutant h e m . The embryonic lethality in Nk*2-5 mutant embryos is at 9-10 days

of gestation which is at least a day earlier than mym mutants. On the other hand, Nkx2-5

mutant hearts present an early looping defect which we did not observe in the mym

hearts. Similar myocardial defects and deregulation of myosin genes in both mym and

Nkr2-5 mutants, and earlier more extensive phenotype in the NkrZ-5 suggest that myrn

may be downstream of Nkx2-5 in the myocardial regulatory pathway.

A gene targeted mutation in Plakoglobin, a ce11 adhesion molecule involved in

cell junctions, also leads to embryonic death after E10.5 due to heart defects very similar

to those in Nkr2-5 and myrn mutant h e m . The Plakoglobin heart phenotype includes

hypoplastic ventricular wall, trabeculae, and endocardial cushion, and defects in

myofibrillar organization (Bierkamp et aL, 1996). suggesting that Plakoglobin may also

be in the Nkr2-5 regulatory pathway.

The mym gene is located on mouse chromosome 5, region B3, close to the W

locus where a recessive lethal mutation (15H1) has also been mapped (Lyon and

Glenister, 1982). ~ 1 9 ~ a deletion in the W locus also leads to embryonic lethality in

homozygotes (Lyon et al., 1984). ~ 1 9 H and 15H1 mice may be used in crosses with mym

mice for complementation and rnapping experiments to understand the genetic relations

between these three lethal mutations. N M - 5 and plakoglobin mutations present v e r -

sirnilar phenotypes to mym gene trap mutation in embryonic heart but map t~ different

chromosomes (17 and I l , respectively, Himmeibauer et al. 1994; Guenet et al. 1995). At

present we have only a partial mym cDNA with a lirnited ORF which is novel and is not

similar to any sequences in the database. Further clonllig is required to obtain complete

sequence information.

In the mym mutant myocytes. the degree of myofibrillogenesis seen was severely

reduced. We examined mutant myocytes from embryos with severe heart phenotypes at

day 10 of gestation. The actin-myosin and thin filaments in the cytoplasm were absent

and the desmosornal ce11 junctions were not present between cells. Some of the less

severely af5ected embryos, however, had beating hearts at the time of dissection, and had

evidence of myofibril contractile elements. It is possible that such embryos lose their

myofibrils sorne time later than their more severe mutant littermates and eventually

succumb by day 1 1 of gestation. We have not yet confirmeci the deletion of myocytes by

apoptotic celI death in a more direct investigation of the phenornenon; however, myocyte

ce11 death was detected by electron microscopy and likely contributeci to the myocardial

disorganization and trabecule disruption. In mutant embqos, such defects incite heart and

circulation failure when rapid proliferation and rearrangement of cardiac myocytes are

the primary contnbuting factor to heart development (Fishman and Chien, 1997).

It is quite likely that the absence of MLC2A and -2V in mutant myocytes is a

major reason for the absence of contractile elements in these cells. Interactions between

myosin light chahs and actin are known to regulate cardiac contractility (Morano et al.,

1995). In the less severe mutant embryos, it is possible that other rnyosin isofonns c m

initially replace the function of misshg myosin light chains. Later, when the mechanical

load on cardiac tissue becomes more demanding. functionai replacement by other

i s o f o m may no longer be sufficient for the maintenance of myocytes.

Because the heterozygous males in the pure 129 background were mostly sterile,

we performed al1 the phenotypic analyses in mixed genetic backgrounds. We therefore

cannot exclude the action of modifier genes affecting the severity or presentation of the

phenotype in the myocardiurn. An earlier embryonic lethality would be paaicularly likely

since as early as day 9.0 of gestation cardiovascular circulation is required to supplement

the yolk sac circulation, and therefore a mutation affecthg myocyte structural integrity

and their cell junctions would be expected to be detrimental to the embryo shortly afler

day 9 of gestation. However, if no modifier genes tum out to be involved in the observed

phenotype, the severity or time of presentation. then the late effect of the mym mutation

may refiect functional redundancy of the myosin isoforms.

In summary, our results with the gene trap mutation in the mym locus demonstrate

an essential role for this gene in the maintenance of myocytes during midgestation

embryonic heart development. The disruption of structural organization of the myocytes

and deregulation of myosin light chah gene expression suggests a key role for mym in

proper execution of cardiac myogenic processes.

Chapter 5: Concluding Remarks

Snmmary and Generai Discussion

Over the last few years. gene trap strategies using mouse embryonic stem (ES) cells have

been developed to perform genetic screens in mammals for novel genes and mutations.

The appeal of gene trapping is in its power to characterize sirnultaneously three different

aspects of developmentaüy regulated geoes: i) expression patterns by the use of a reporter

gene; ü) cloning by RACE-PCR and sequence analysis; and iii) transmission through the

germ line and phenotypic anaiysis of the heterozygous and homozygous disruptions.

Furthermore, as ES cells can be grown and manipulated in vitro, it is convenient to

prescreen for featws of interest in culture before further characterization in vivo.

The gene trap strategy, combined with other transgenic and gene targeting

technology now available in mice, provides a powerful array of experimental approaches

to dissect genetic pathways involved in rnammalian development. The RA induction gene

trap screen in which If 93 and R140 clones were isolated indicate the feasibility of this

approach to identiw genes that lie downstream of ügmd/receptor-mediated signaling

pathways. The RA induction screen significantly e ~ c h e d for integration events into

genes that respond to an inducer in vitro and that present restncted expression patterns

during development. For instance, 19 of 20 (95%) RA-responsive ES ceii lines displayed

a restncted expression pattern between E8.5 and E l 1.5 (Forrester et al.. 1996). In

contrast, gene trap screens in which ce11 lines were selected solely on the basis of l a d

expression in undifferentiated ES cells. only 30% displayed specific expression patterns

at these stages (Wurst et al., 1995). Potentially, by scaling up an induction gene trap

screen, it may be possible to identify most genes involved in any given genetic pathway

that are active in ES cells.

As describeci in Chapter 1, other genes involved in developmental processes have

been identified in gene trap approaches. In addition, other gene trap experiments have

identified genes that are specific to certain developmental lineages such as hematopoietic

lineages. as well as genes that are responsive to other physical, chernical or molecular

agents (W. Stanford. G. Carnana, and W. Wurst, unpublished data).

Dr repeats

In Chapter 2. Dr repeats. a novel class of expressed repetitive DNA elements in the

mouse genome. were described. Dr expression is spatially restricted in embryos and in

the adult brain. and in vitro is induced by RA. The memben of the Dr famiiy are highly

similar in sequence and show peculiar structural features. Repetitive DNA sequences

form a substantial portion of eukaryotic genomes. Although there is evidence that some

repetitive sequences rnay participate in gene regulation and other cellular functions, due

to their repetitive nature and global presence in the genome and transcnptome (the sum

total of al1 transcripts in a ceIl), functionai studies on these nucleic acid sequences have

been difficult. A few experimental approaches may however be suggested to further

elucidate the cellular role of these sequences.

Our results raise the possibility that the expression of Dr-containing transcripts is

part of an ES ce11 differentiation program induced by RA. Employing a transgenic

approach, the precise length of Dr repeats required for RA response can be determined

and their functional domains can be dissected. In particular. since Dr sequences rnay be

part of a much larger repetitive sequence, it is conceivable that other sequence domains

exist within these elements that are essential for other possible structural or regulatory

functions such as interaction with regulatory proteins. Using transgenic experiments. the

spatial and temporal domains of Dr expression and function can also be addressed.

Transgenic expression of Dr repeats in heterologous tissues may also shed some light on

their function. We know that Dr sequences are highly dispersed in the genome. As the

mouse genomic sequences become available, the exact length, location and organization

of these elements in the genome rnay further indicate some functional or structural roles

for them. Finally, it rnay be possible to inactivate these sequences by overexpressing their

antisense RNA,

Dr sequences include a polyadenylation signal, a poly A track and long 3'

palindromic sequences indicating a retrotransposition event in the history of these

sequences. Dr repeats rnay be part of a mobile genetic element and as such rnay have

played a role in the construction of the genome organization.

Aquurius is a novel developmentally regulated gene descrïbed in Chapter 3. Durhg

ernbryogenesis, it is expressed in mesodem. neural crest cells (NC) and their target

tissues. and in neuroepithelium. There are many genes with restricted expression in

subsets of NC ceils. For instance, the transcription factors AP-2, AP-2.2, DLr-2 , Hm-2,

P d , Pax7 and twist are dl expressed in cranial NC cells (Chazaud et al., 1996; Chen

and Behringer, 1995; Gendron-Maguire et al., 1993; Mansouri et al., L996; Qiu et al.,

1995; Tremblay et al., 1995; Zhang et al.. 1996). In a ce11 lineage marker analysis, it

would be of interest to compare Aqr expression pattern with other NC-expressed genes.

Such analysis would clariq the subset of NC cells in which Aqr rnay play a role.

In vitro, Aquarius is responsive to RA and its transcript carries the RA-inducible

Dr repeat element. By isolating genomic sequences and the Aqr promoter region, it will

be possible to see whether Aquarius carries RA responsive elements or whether its

activation by RA is indirect. Aitematively, there rnay be a novel RA induction pathway

through the Dr elements as was suggested in Chapter 2. The purpose of the onginal RA

screen was to identify genes downstream of RA signaling. Future expenments will be

needed to indicate whether Aqr is responsive to RA in vivo. and if so, where in the RA

signaling pathway, Aquurius rnay play a role.

AquariuF cDNA sequence shows weak similarity to RNA-dependent RNA

polymerases. To understand whether the sequence similarity to RNA-dependent RNA

polymerases has a functional relevance, the protein product of the gene produced in an in

vitro expression system could be examined for RNA binding and polymerase activity. At

present, little is known about the structure and function of RNA-dependent RNA

poly merases.

The Aquarius protein product may also be used to raise antibody for

immunohistochemical studies for cellular localization of the gene product and to

precipitate from cellular extracts other proteins that rnay interact with the Aquarius

protein. Cloning and sequence analysis of the other Aquarius isoforms identified by

Northem analysis may also provide more information about other possible functional

domains in this gene. The presence of nuclear locaiization signal in the Aquarius

sequence suggests the presence of its gene product in the cell nucleus.

The confinnation of an RNA-dependent RNA polymerase in the mouse, would

not be that surprising, as several reports have already indicated the presence of such

activity in eukaryotic organisms (Schiebel et al., 1993a; Schiebel et ai., 1993b; Volloch

et al., 1991; Lai 1995). The Tetrahymena telomerase is also similar to viral RNA-

dependent RNA polymerases (Collins et al., 1995). It wiil then be of interest to know,

through a biochemicd assay, if the Aquarius protein possesses telomerase activity.

Mice homozygous for the Aqr gene trap insertion are viable and normal.

However, the gene trap allele described is not a null mutation as Aqr RNA is present in

hornozygous animais. The lack of a discemible phenotype may be due to read-through

transcription and splicing across the trapping vector. To understand the function of

Aquarius during embryogenesis, it is essential to mutate it by standard gene targeting

techniques.

Using the Cre-loxP recombination system, it is possible to mutate a gene in a

temporal ou tissue-specific manner to more accurately dissect gene function during

specific times or in specific tissues. This technique requires the presence of loxP sites in

the genomic locus, and transient or tissue specific expression of Cre recombinase, both of

which are now possible in the mouse using embryonic stem cells to generate the h e s

(Smith et al., 1995). To study different domains of the Aquarius protein, subtle mutations

can now be introduced through the 'knock in' version of gene targeting, where a

replacement vector introduces the mutated version of the gene into its genomic locus

(CasMa et al., 1996).

The gene trapped locus in the RI40 ES line was named myocyte maintenance (rnym) as it

is required for the development of cardiac myocytes as described in Chapter 4. During

midgestation ernbryonic development, mym is expressed in cardiac myocytes, limbs, gut,

meninges and the choroid plexus. Homozygous mutant embryos died during

midgestation due to heart failure. Mutant hearts presented thin myocardial walls and

defective trabeculae. The myocardium, where myrn is expressed, was hypoplastic due to

myocyte ce11 death. Apoptotic cell bodies and electron dense nuclei indicative of

apoptotic ceii death were detected by electron microscopy.

To undentand the myocyte defects better, myoblasts from early homozygous

mutant embryos should be cultured and further investigated in vitro. For instance, as

mysoin light chah genes are deregulated in these myocytes, it would be informative to

document in vitro the deregulation of these and other myocyte markers. The apoptotic

response needs to be confmed in vitro and in vivo, and investigated further at the

molecular level. Ce11 division as weil as ce11 differentiation with its ensuing cellular

changes, such as myofilament expression and organization, can also be examined in

mutant myocytes in vitro.

To bypass rnidgestation embryonic iethality and to examine other possible

functions of rnym later in development, chimeric analysis c m be undertaken in which

homozygous mutant ES cells are selected in high G418 concentration and used to

generate aggregation chimeras. Homozygous mutant myocytes will then be amongst wild

type cells which, depending on the degree of chimensm, might compensate for

myocardial defects. This analysis should determine whether the rnym defect is cell

autonomous (the gene function is required in the ceiis that express the gene), or ce11 non-

autonomous (the gene hinction is required in neighboring ceus).

About one third of homozygous mutant embryos were blocked at the headfold

stage (E8.5) with no further development up to El05 These early defective embryos

suggest a more general cell division or biological defect due to rnym disruption. Lineage

marker analysis and in vitro culture of embryonic cells may shed some light on the cause

of this early developmental defect. Ideally this defect should be examined in a pure

genetic background as the two phenotypes (E10.5 mayocardial defect & E8.5

developmental block) and their penetrance (66% & 33%) are likely due to modifier gene

effects.

Heterozygous rnym males had reduced fenility in a pure 129 genetic background

due to a defect in spermatogenesis. This defect, although a disadvantage for studying

rnym in a pure genetic background, indicates the likelihood of phenotypic variation in

different genetic backgrounds. There are indeed other cases of phenotypic variations in

mice in different genetic backgrounds (Ramirez-Solis et al., 1993; Threadgill et al.,

1995).The rnym defect in spermatogenesis also needs further investigation perhaps by

lineage marker analysis and by in vitro culture of spermatocytes.

S imilar myocardial defects and deregulation of my osin genes have k e n repo rted

in NkrZ-5 mutant hearts (Lyons et al.. 1995). Embryonic death after E10.5, hypoplastic

ventricular walls, trabeculae and endocardial cushion, and defects in myofibriiiar

organization have aiso been detected in mice homozygous for a nul1 mutation in the

Plakoglobin gene (Bierkamp et ai-. 1996). The remarkable similarities in time and tissue-

specificity of these mutant phenotypes suggest functional roles in the sarne pathway.

Since genetic interactions are of great value in unraveling the rnoIecular b a i s of

developmental processes, gene interactions among these three loci cm be investigated by

breeding mice carrying these mutations, generating double mutant animais and

examining their phenotype. This is possible because each mutation manifests phenotypic

features that are not shared by the other two. Therefore, by scoring for the presence or

absence of the exclusive phenotypic features and the severity of the phenotypes in double

mutants, it will be possible to gain some insights into whether the products of these genes

functionaiiy interact during heart developrnent.

Conclusions

Developmentai genetics seeks a moIecuIar understanding of the developmental

processes that perpetually produce progeny in each successive generation. Such an

understanding at the level of gene function is currentiy limited in mammaiian embryonic

development. The development of transgenic technology, encompassing gene targeting

and trapping methodologies. has initiated a flourishing era of molecular investigations

into mamrnalian developmental genetics. This approach is promising to narrow the

'phenotype gap,' the gap in the available mutant resources (2% of the full range of

phenotypes possible), that now exist in the rnouse as a mode1 organism (Brown and

Peters, 1996)-

The work described here demonstrates the feasibility of in vitro manipulation and

screening of ES cells to identiw and rapidly characterize genes and genetic elements that

are developmentally regulated or lie dong a cellular response pathway. The generation of

gene trapped mice ailows for expression and functional analysis of the trapped genes.

Sequence databases, analytical softwares and ready access to these tools through

the internet have led to an increase in gene identification by cornputer (Fickett 1996).

Approximately 50% of al1 human genes are already represented in the Expressed

Sequence Tag (EST) databases (Gerhold and Caskey, 1996). A similar EST project is

now underway at the Washington University, St Louis USA, to develop 400,000 mouse

ESTs. The combination of gene trap experimentation and EST databases provides ready

access to the sequence of trapped genes and has the potential to increase several fold the

pace of gene discovery and functional characterization.

Literature Cited

Agabian N. ( 1990). T m splicing of nuclear pre-mRNAs. Ce116 1, 1 157-60.

Aizawa H, Sekine Y, Takemura R, Zhang 2, Nangaku M, and Hirokawa N. (1992). Kinesin family in murine central nervous system. J Ce11 Bio 119, 1287-1296.

Alberts B. Bray D, Lewis J, Raff M. Roberts K, and Watson JD. (1994). Molecular Biology of the Cell, pp. 456. Garland Publishing, New York.

Men ND, Cran DG, Barton SC, Hettle S. Reik W. and Surani MA. (1988). Transgenes as probes for active chromosomal domains in mouse development. Nature 333,852-855.

Ashburner M. (1989). "Drosophila, A laboratory handbook." Cold Spring Harbor Laboratory, Cold Spring Harbor.

Azpiroz-Leehan R, and Feldrnann KA. (1997). T-DNA insertion mutagenesis in Arabidopsk going back and forth. Trends Genet 1 3, 152-6.

Baker RK, Haendel MA, Swanson BJ, Shambaugh JC, Micales BK. and Lyonr GE. (1997). In vitro preselection of gene-trapped ernbryonic stem cell clones for characterizing novel developmentdly regulated genes in the rnouse. Dev Bio 185, 201- 2 14.

Beddington RSP. (1992). Transgenic strategies in rnouse ernbryology and development. In "Transgenic Animals" (Grosveld F and Koiiias G. Eds.), pp. 47-77. Academic Press, San Diego.

Bedell MA, Jenkins NA, and Copeland NG. (1997). Mouse models of human disease. Part 1: Techniques and resources for genetic andysis in mice. Genes & Oevelopment I l , 1-10,

Bennett KL, Hill RE, Pietras DF, Woodworth-Gutai M, Kane-Haas C, Houston JM, Heath JK, and Hastie ND. (1984). Most highly repeated dispersed DNA families in the mouse genome. Mol Cell Biol4, 156 1-7 1.

Bierkamp C, McLaughlin KJ, Schwarz H, Huber 0, and Kemier R. (1996). Embryonic heart and skin defects in mice lacking plakoglobin. Developrnental Biology 180,780-785.

Blumenthal T, and Thomas J. (1988). CLr and tram mRNA splicing in C. elegans. Trends Genet 4,305-308.

Brandon EP, Idzerda RL, and McKnight GS. (1995). Targeting the mouse genome: a compendium of knockouts (part 1). Curr Bio 5,625-634.

Brannan CI, Perkins AS, Vogel KS, Ratner N, Nordlund ML, Reid SW, Buchberg AM, Jenkins NA, Parada LF, and Copeland NG. (1994). Targeted dismption of the neurofibromatosis type- 1 gene leads to developmental abnormaiities in heart and various neural crest-derived tissues [published erratum appears in Genes Dev 1994 Nov 15; 8(22):2792]. Genes Dev 8, 10 19-29.

Brini AT, Lee GM, and Kinet EP. (1993). Involvement of A h sequences in the cell- specific regulation of transcription of the g chain of Fc and T cell receptors. J Bi01 Chem 268, 1355-1361.

Brinster RL, Chen HY, Trumbauer ME, Senear AW, Warren R, and Palmiter RD. (198 1). Somatic expression of herpes thymidine kinase in mice following injection of a fusion gene into eggs. Ce11 27,223-23 1.

Britten RJ, and Davidson EH. (1969). Gene regulation for higher cells: a theory. Science 165, 349-357.

Brown SDM, and Peters J. (1996). Combining mutagenesis and genornics in the mouse - closing the phenotype gap. Trends in Genetics 12,433-435.

Capecchi MR. (1989). Altering the genome by homologous recombination. Science 244, 1288-92.

Carraway-III KL. (1996). Involvement of the neuregulins and their receptors in cardiac and neural development. Bioessays 18,263-266.

Castilla LH, Wijmenga C, Wang Q, Stacey T, Speck NA, Eckhaus M, and et al. (1996). Failure of embryonic hematopoisis and lethal hemorrhages in mouse embryos heterozygous for a knocked-in leukernia gene CBFB-MYHI 1. Cell 87.

Chant J. (1996). Generation of ce11 polarity in yeast. Curr Opin Cell Biol8,557-65.

Charron J, Malynn BA, Fisher P. Stewart V, Jeannotte L, Goff SP, Robertson EJ, and Ait FW. (1992). Embryonic lethality in mice homozygous for a targeted disruption of the N- myc gene. Genes & Dev 6,2248-2257.

Chazaud, C., Oulad-Abdelghani M. Bouillet P, Decimo D, Chambon P, and Dolle P. (1996). AP-2.2, a novel gene related to AP-2, is expressed in the forebrain, limbs and face during mouse embryogenesis. Mech Dev 54,83-94.

Chen J, Gorman JR, Stewart V, Williams B, Jacks T, and Alt FW. (1993). Generation of normal lymphocyte populations by Rbdeficient embryonic stem cells. Current Biology 3, 405-413.

Chen J. Nachabah A, Scherer C, Ganju P. Reith A, Bronson R and Ruley HE. (1996). Germ-line inactivation of the murine Eck receptor kinase by gene trap retroviral insertion. Oncogene 12,979-88.

Chen 2, Friedrich GA, and Soriano P. (1994). Transcriptional enhancer factor 1 disruption by a retroviral gene trap leads to heart defects and embryonic lethality in rnice. Genes Dev 8,2293-30 1 .

Chen 2-F, and Behringer RR. (1995). îwist is required in head mesenchyme for cranial neural tube morphogenesis. Genes Dev 9,686-699.

Chomczynski P. and Sacchi N. ( 1987). Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162, 156- 159.

Chowdhury K, Bonddo P, Torres M, Stoykova A, and Gruss P. (1997). Evidence for the stochastic-integration of gene trap vectors into the mouse germline. Nucleic Acids Research 25, 1531-1536.

Cohn MJ, and Tickle C. (1996). Limbs: a mode1 for pattern formation within the vertebrate body plan. Trends in Genetics 12,253-257.

Collins K, Kobayashi R, and Greider CW. (1995). Purification of Tetrahymena telornerase and cloning of genes encoding the two protein components of the enzyme. Ce11 8 1,677-686.

Codon RA, and Herrmann BG. (1993). Detection of messenger RNA by in situ hybridization to postimplantation embryo whole mounts. In "Methods in Enzymology, Guides to Techniques in Mouse Development" (Wassarman PM and DePamphiLis ML, Eds.), Vol. 225, pp. 373-383. Academic Press, New York.

Consigli SA, and Joseph-Silverstein I. ( L 99 1). Immunolocalization of basic fibroblast growth factor during chicken cardiac development. J Cell Physiol 146,379-85.

Costantini F, and Lacy E. (1981). Introduction of a rabbit beta-globin gene into the mouse germ line. Nature 294,92-94.

Couly GF, Coltey PM, and Le Douarin NM. (1993). The triple origin of skuU in higher vertebrates - A study in quail-chick chimeras. Development 1 17,409-429.

Darne11 J, Lodish H, and Baltimore D. (1990). Molecular Ce11 Biology, pp. 296. Scientific American Books, New York.

Decoville M, Moreau P, Viegas-Pequignot E, and Locker D. (1992). Genomic organization and nucleotide sequence of a long mosaic repetitive DNA in the mouse genome. Mammalian Genome 2, 172- 185.

DeGregori J, Russ A, von Melchaer H, Raybum H, Priyaranjan P, Jenkins NA, Copeland NG, and Ruley HE. (1994). A murine homolog of the yeast RNA1 gene is required for pos timplantation development. Genes Dev 8,265-76.

Deininger PL, Batzer MA, Hutchison III CA, and Edgell MH. (1992). Master genes in mammaiian repetitive DNA amplification. Trend Genet 8,307-3 1 1 .

Delseny M, Cooke R, Raynal M, and Grellet F. (1997). The Arabidopsis thaliana cDNA sequencing projects. FEBS Len 405, 129- 132.

Dietrich W, Katz H, Lincoln S, Shin HS, Freidman J, Dracopoli NC, and Lander E. (1 992). A genetic rnap of the mouse suitable for typing intraspecific crosses. Genetics 13 1,423-447.

Doetschman T, Gregg RG, Maeda N, Hooper ML. Melton DW, Thompson S, and Srnithies 0. (1987). Targeted corrections of a mutant HPRT gene in mouse embryonic stem ceus. Nature 330,576-578.

Downs KM, Martin GR, and Bishop JM. (1989). Contrasting patterns of myc and N-myc expression during gastnilation of the mouse embryo. Genes & Dev 3,860-869.

Dnever W, and et al. (1996). A genetic screen for mutations affecting embryogenesis in zebrafish. Development 123.37-46.

Edwards DR, Parfett U, and Denhardt DT. (1985). Transcriptional regulation of two serurn-induced RNAs in mouse fibroblasts: equivalence of one species to B2 repetitive elements. Mol Cell Biol5,3280-3288.

Engelmann GL, Dionne CA, and Jaye MC. (1993). Acidic fibroblast growth factor and kar t development. Role in myocyte proliferation and capillary angiogenesis. Cire Res 72,7-19.

Eu1 J, Graessmann M. and Graessmann A. (1995). Experimental evidence for RNA trans- splicing in mamrnaiian cells. EMBO J 14,3226-3235.

Evans MJ. Carlton MBL, and Russ AP. (1997). Gene trapping and functional genomics. Trends in Genetics 13,370-374.

Evans MJ, and Kaufman MH. (1981). Establishment in culture of pluripotential cells from mouse embryos. Naare 292. 154-156.

Felsenfeld AL. (1996). Defining the boundaries of zebrafish developrnental genetics. Nature Genetics 14,258-263.

Feng G-S. Shen R. Heng HHQ, Tsui L-C. Kazlauskas A, and Pawson T. (1994). Receptor-binding, tyrosine phosphorylation and chromosome localization of the mouse SH2-containing phosphotyrosine phosphatase S yp. Oncogene 9, 1545-50.

Fickett W. (1996). Finding genes by computer: the state of the art. TIG 12.3 16-320.

Fishman MC, and Chien KR. (1997). Fashioning the vertebrate heart: earliest embryonic decisions. Development 124,2099-2 1 17.

Forrester L. Nagy A, Sam M. Watt A, Stevenson L, Bernstein A, Joyner AL, and Wurst W. (1996). An induction gene trap screen in embryonic stem ceus: Identification of genes that respond to retinoic acid in vitro. Proc. Nd. Acud. Sci. 93, 1677- 1682.

Friedrich G, and Soriano P. (1991). Promoter traps in embryonic stem cells: a genetic screen to identiQ and mutate developmentai genes in mice. Genes Dev 5, 15 13- 1523.

Gaiano N, Amsterdam A, Kawakami K, Allende M. Becker T, and Hopkins N. (1996). Insertional mutagenesis and rapid cloning of essentiai genes in ~ e b ~ s h . Nature 383. 829-832.

Gale E, Prince V, Lumsden A, Clarke J, Holder N, and Maden M. (1996). Late effects of retinoic acid on neural crest and aspects of rhombomere identity. Development 122, 783- 793.

Gardner RL. (1968). Mouse chimaeras obtained by the injection of cells into the blastocyst. Nature 220,596-7.

Gasca S, Hill DP, Klingensmith J, and Rossant J. (1995). Characterization of a gene trap insertion into a novel gene, cordon-bleu, expressed in axial structures of the gastrulating mouse embryo. Dev Genet 17, 141-54.

Gassmann M, Casagranda F, Orioli D, Simon H, Lai C, Klein R, and Lernke G. (1995). Aberrant neural and cardiac development in mice lackng the ErbB4 neuregulin receptor. Nature 378,390-394.

Gendron-Maguire M, Mallo M. Zhang M. and Gndley T. (1993). Hoxa-2 mutant mice exhibit homeotic transformation of skeletal elements derived fiom cranial neural crest. Ce11 75, 1317-1331.

Gerhold D, and Caskey Ci'. ( 1996). It's the genes! EST access to human genome content. Bioessays 18,973-98 1.

Gordon IW, and Ruddle FH. (1981). Integration and stable germ line transmission of genes injected into mouse pronuclei. Science 2 14, 1244- 1246.

Gossler A, Doetschrnan TT Kom R, Serfling E, and Kemler R. (1986). Transgenesis by means of blastocyst-derived embryonic stem cell lines. Proc Natl Acad Sci 83, 9065- 9069.

Gossler A, Joyner AL, Rossant J, and Skarnes WC. (1989). Mouse embryonic stem celis and reporter constmcts to detect developmentally regulated genes. Science 244,463-465.

Gu K. Zou YR, and Rajewsky K. (1993). Independent contml of immunoglobulin switch recombination at individual switch regions evidenced through Cre-loxP-mediated gene targeting. Cell73, 1 155- 1 164.

Guenet JL, Simon-Chazottes D. Ringwald M, Kemler R. (1995) The genes coding for alpha and beta catenin (Catnal and Catnb) and plakoglobin (Jup) map to mouse chromosomes 18,9, and 1 1. respectively. Mamm Gemme 6(5), 363-6.

Guthrie C, and Fink GR. (1991). "Guide to yeast genetics and molecular biology." Acadernic Press, San Diego.

Gutmann DH, Andersen LB, Cole L, Swaroop M, and Collins FS. (1993). An altematively-spliced mRNA in the carboxy terminus of the neurofibromatosis type 1 m l ) gene is expressed in muscle. Hwnan Molecular Genetics 2,989-992.

Haffter P, and et al. (1996). The identification of genes with unique and essential functions in the developrnent of the zebrafish, Danio rerio. Oevelopment 123, 1-36.

Harbers K, Jahner D, and Jaenisch R. (1981). Microinjection of cloned retroviral geaomes into mouse zygotes: Integration and expression in the animal. Nature 293.540- 542.

Hardrnan N. ( 1986). Structure and function of repetitive DNA in eukaryotes. Biuchem J 234, 1-11.

Harrington L. McPhail T, Mar V, Zhou W. Oulton R, Amgen EST, Bass MB, Amda 1, and Robinson MO. ( 1997). A marnmalian telornerase-associated protein. Science 275. 973-977.

Harvey RP. (1996). Review: NK-2 homeobox genes and heart development. Dev Bi01 178,203-2 16.

Heilxnann-Blumberg U, McCarthy Hintz MF, Gatewood JM, and Schmid CW. (1993). Developmental Differences in Methylation of Human Alu Repeats. Mol Ce11 Bio 13, 4523-4530.

Heng HHQ, Squire J, and Tsui L-C. (1992). High resolution mapping of mammalian genes by in situ hybridization to free chromatin. Proc Natl Acad Sci USA 89,9509-95 13.

Heng HHQ, and Tsui L-C. (1993). Modes of DAPI banding and simultaneous in situ hybridization. Chromosoma 102,325-332.

Heng tMQ, Xiao H, Shi X-M, Greenblatt J, and Tsui L-C. (1994). Genes encoding general initiation factors for RNA polymerase II transcription are dispersed in the human genome. Hum Mol Genet 3.6 1-64-

Henthom PS, Mager DL, Huisman TH, and Smithies 0. (1986). A gene deletion ending within a cornplex array of repeated sequences 3' to the human beta-globin gene cluster. Proc Natl Acad Sci USA 83,s 194-5 198.

Hicks GG, Shi EG, Li XM, Li CH, Pawlak M, and Ruley HE. (1997). Functional genomics in mice by tagged sequence mutagenesis. Nature Genetics 16,338-344.

Himmelbauer H, Harvey RP, Copeland NG, Jenkins NA, and Silver LM. (1994) High- resolution genetic analysis of a deletion on mouse chromosome 17 extending over the fused, tufied, and homeobox NkxS-5 loci. Mamm Genorne 5(l S), 8 14-6.

Himen A, Hicks JB, and Fink GR. (1978). Transformation of yeast. Proc Nad Acad Sci USA 75, 1929-1933.

Hirakow R, and Sugi Y. (1990). Intercellular junction and cytoskeletal organization in embryonic chick myocardial cells. In "Developmental cardiology: Morphogenesis and hinction" (Clark EB and Takao A, Eds.), pp. 95-1 13. Futura hiblishing, Mount Kisco.

Hobbs HH, Brown MS, Goldstein JL, and Russell DW. (1986). Deletion of exon encoding cysteine-rich repeat of low density Iipoprotein receptor alters its binding specificity in a subject with familial hypercholesterolemia. J Biol Chem 26 1, 13 1 14- 13 120.

Hogan B, Beddington R, Costantini F, and Lacy E. (1994). "Manipulating the Mouse Embryo, A Laboratory Manual." Cold Spring Harbor Laboratory. Cold Spring Harbor, New York.

Hope IA. (1 99 1). Promoter trapping in Caenorhabditis elegans. Development 1 13, 399- 408.

Hoyt PR, Bartholomew C, Davis AI, Yutzey K, Garner LW, Potter SS, Ihle JN, and Mucenski ML. (1997). The Evil proto-oncogene is required at rnidgestation for neural, heart, and paraxial mesenchyme development. Mechanism of Development 65.55-70-

Huang LS, Ripps ME, Korman SH, Deckelbaum RJ, and Breslow JL. (1989)- Hypobetaiipoproteinernia due to an apolipoprotein B gene exon 21 deletion denved by Nu-Alu recombination. J Biol Chem 264, 1 1394- 1 1400.

Ito HT Miller SC, Billingham ME. Akimoto H, Torti SV, Wade R, Gahlmann R Lyons G, Kedes L, and Torti FM. (1990). Doxombicin selectively inhibits muscle gene expression in cardiac muscle cells in vivo and in vitro. Proc Natl Acad Sci USA 87,4275-9.

Jacks T, Shih TS, Schmitt EM. Bronson RT, Bemards A, and Weinberg RA. (1994). Tumour predisposition in mice heterozygous for a targeted mutation in ~ f l . Nat Genet 7, 353-6 1 .

Jaenisch R. ( 1976). Genn Line integration and Mendelian transmission of the exogenous Molony leukemia virus. Proc Natl Acad Sci 73, 1260- 1264.

Johnson SL, and Weston JA. (1995). Temperature-sensitive mutations that cause stage- specific defects in zebrafish fin regeneration. Genetics 14 1 , 1583- 1595.

Jorgensen RA, Rothstein SJ, and Reznikoff WS. (1979). A restriction enzyme cleavage map of Tn5 and location of a region encoding neomycin resistance. Mol Gen Genet 177, 65-72.

Kariya K, Farrance MT and Simpson PC. (1993). Transcriptional enhancer factor-1 in cardiac myocytes interacts with an alpha1 adrenergic- and a beta-protein kinase C- inducible element in the rat beta-myosin heavy chah prornoter. J Biol Chem 268,26658- 26662.

Kato KT Kanarnori A, Wakamatsu Y, Sawai S. and Kondos H. (1991). Tissue distribution of N-myc expression in the early organogenesis period of the mouse embryo. Dev Growth Differ 33.29-36.

Kato S , Anderson RA, and Camerini-Otero RO. (1986). Foreign D N A introduced by calcium phosphate is integrated into repetitive DNA elements of the mouse L ce11 genome. Mol Ceil Bi01 6, 1787-1795.

Kaufman MH. (1992). 'The Atlas of Mouse Development." Academic Press, London.

Kerr WG, Heller M. and Herzenberg LA. (1996). Analysis of lipopolysaccharide- response genes in B-lineage cells demonstrates that they can have differentiation stage- restricted expression and contain S H 2 domains. Proc N d Acad Sci 93,3947-3952.

Knapik EW, and et al. (1996). A reference cross DNA panel for zebrafish (Danio rerio) anchored with simple sequence length polymorphisms. Development 123,45 1460.

Koga Y, Lindstrom ET Fenyo EM, Wigzell H, and Mak TW. (1988). High levels of heterodisperse RNAs accumulate in T ceils infected with human irnmunodeficiency virus and in normal thymocytes. Proc Natl Acad Sei USA 85,452 1-4525.

Kothary R, Clapoff S, Brown A. Campbell R, Peterson A, and Rossant J. (1988). A transgene containing lac2 inserted into the dystonia locus is expressed in neural tube. Nature 335,435-437.

Kothary R, Clapoff S, Darling S. Perry MD, Moran LA, and Rossant J. (1989). Inducible expression of an hsp68-lad hybrid gene in transgenic mice. Development 105,707-7 14.

Kreidberg JA, Sariola H, Loring JM, Maeda M. Pelletier J, Housman D, and Jaenisch R. ( 1993). WT- 1 is required for early kidney development. Ce21 74, 679-69 1 .

Krontiris TG. ( 1995). Minisatellites and human disease. Science 269, 1682-1683.

Krull CE, Collazo A, Fraser SE, and Bronner-Fraser M. (1995). Segmental migration of tmnk neural crest: time-lapse analysis reveals a role for PNA-binding molecules. Development 12 1,3733-3743.

Kubalak SW, Miller-Hance WC, O'Brien TX, Dyson E, and Chien KR. (1994). Chamber specification of atrial rnyosin light chain-2 expression precedes septation during murine cardiogenesis. J Bi01 Chem 269, 1696 1-70.

Kucherlapati US, Eves EM, Song KY, Morse BS, and Srnithies 0. (1984). Homologous recombination between plasmids in mamrndian cells can be enhanced by treatment of input DNA. Proc Natl Acad Sci USA 81,3153-3 157.

Kumar A, Crawford K, Close L, Madison M, Lorenz I, Doetschman T, Pawlowski S, D u w J, Neumann J, Robbins J, Boivin GP, OToole BA, and Lessard JL. (1997). Rescue of cardiac a-actin-deficieut mice by entenc smooth muscle g-actin. Proc Natl Acad Sci USA 94 ,44064 1 1.

Kwee L, Baldwin HS, Shen HM, Stewart CL, Buck C, Buck CA, and Labow MA. (1995). Defective development of the embryonic and extraembryonic circulatory systems in vascular ce11 adhesion molecule (VCAM- 1) deficient mice. Development 12 1, 489- 503.

Lai MMC. (1995). The molecular biology of hepatitis delta virus. Annu Rev Biochem 64, 259-86.

Lania L, Pannuti A, La Mantia G, and Basilico C. (1987). The transcription of B2 repeated sequences is regulated during the transition from quiescent to proliferative state in cultured rodent cells. FEBS Lett 2 l9,4oO-~O4.

Leberer E, Thomas DY, and Whiteway M. (1997). Pheromone signailhg and polarized morphogenesis in yeast. Curr Opin Genet Dev 7,59-66.

Lee K-F, Simon H, Chen H, Bates B, Hung M-C, and Hauser C. (1995). Requirement for neuregulin receptor erbB2 in neural and cardiac development. Nature 378,394-398.

Lee KJ, Ross RS, Rockman HA, Harris AN, O'Brien TX, van Bilsen M, Shubeita HE, Kandolf R, Brem G, and Pnce J. (1992). Myosin light chain-2 luciferase transgenic mice reveal distinct regulatory programs for catdiac and skeletal muscle-specific expression of a single contractile protein gene. J Bi01 Chem 267, 15875-85.

Lee YM, Osumi-Yamashita N, Ninomiya Y, Moon CK, Eriksson U, and Eto K. (1995). Retinoic acid stage-dependently alters the migration pattern and identity of hindbrain neural crest cells. Development 12 1,825-37.

Lewandoski M, and Martin GR. (1997). Cre-mediated chromosome loss in mice. Nature Genetics 17,223-225.

Lewis AL, Guicherit OM, Datta SK, Hanten SK, and Kellems RE. (1996). Structure and expression of the murine muscle adenylosuccinate synthetase gene. J Biol Chem 271, 22647-56.

Lints TJ, Parsons LM. Hartiey L, Lyons 1, and Harvey RP. (1993). Nkx-2.5: a novel murine homeobox gene expressed in early heart progenitor cells and their myogenic descendants. Development 1 19.4 19-43 1 .

Lipman DJ, Pearson WR. (1985). Rapid and sensitive protein simüarity searches. Science 227, 1435- 144 f .

Lis JT. Simon JA. and Sutton CA. (1983). New heat shock puffs and b-galactosidase activity resul ting from transformation of Drosophila with an hsp 70-lac2 hybrid gene. Cell35,403-410.

Litvin J, Montgomery M, Gonzalez-Sanchez A. Bisaha JG, and Bader D. ( 1992). Trends Cardiovasc Med 2-27.

Lohler J, Timpl R, and Jaenisch R. (1984). Embryonic lethal mutation in mouse coiiagen I gene causes rupture of blood vessels and is associated with erythropoietic and mesenchymal celi death. Ce11 38,597-607.

Lohnes D, Mark M, Mendelsohn C, Dolli P, Dierich A, Gorry P, Gansmuller A, and Chambon P. (1994). Function of the retinoic acid receptors (RAR) during development.

Craniofacialand skeletal abnormaiities in RAR double mutants. Devehpment 120. 2723-2748.

Lyon MF. and Glenister PH. (1982). A new allele sash (wsh) at the W-locus and a spontaneous recessive lethal in mice. Genet. Res. 39, 3 15-322.

Lyon MF, Glenister PH. Loutit JF, Evans EP, and Peters J. (1984). A presumed deletion covering the W and Ph loci of the mouse. Genet. Res. 44, 16 1 - 168.

Lyon MF. Rastan S. and Brown SDM. (1996). "Genetic variants and strains of the laboratory mouse." Oxford University Press, Oxford.

Lyons GE, Schiaffino S. Sassoon D, Barton P, and Buckingham M. (1990). Developrnental regulation of myosin gene expression in mouse cardiac muscle. J Cell Bi01 1 1 1,2427-36.

Lyons 1, Parsons LM, Hartley L, Li R, Andrews JE, Robb L, and Harvey RP. (1995). Myogenic and morphogenetic defects in the heart tubes of murine embryos lacking the homeo box gene Nkx-2.5. Genes & Development 9, 1654- 1666.

Maichele AJ, Farwell NJ, and Chamberlain JS. (1993). A B2 repeat insertion generates alternate structures of the mouse muscle g-phosphoryl&e kinase gene. ~enomic f 16, 139- 149.

Malhotra A, Huang S, and Bhan A. (1979). Subunit hinction in cardiac myosin: Effect of removal of Lc2 ( l8OOO molecular weight) on enzymatic properties. Biochemistry 18.46 1 - 467.

Mangelsdorf DJ, Thummel C, Beato M. Herdich P. Schutz G. Umesono K, Blumberg B, Kastner P, Mark M, Chambon P, and Evans RM. (1995). The Nuclear Receptor Superfamily: The Second Decade. Ce11 83,835-839.

Mansour SL, Thomas KR. and Capecchi MR. (1988). Disruption of the proto-oncogene int-2 in mouse embryo-denved stem cells: a general strategy for targeting mutations to non-selectable genes. Nature 336,348-352.

Mansouri A, Stoykova A, Toms M. and Gmss P. (1996). Dysgenesis of cephalic neural crest derivatives in P d - l - mutant mice. Development 122.83 1-8.

Mark WH, Signorelli K, Blum M. Kwee L. and Lacy E. (1992). Genornic structure of the locus associated with an insertional mutation in line 4 transgenic mice. Genomics 13, 159- 166.

Martin GR. ( 198 1). Isolation of a pluripotent ce11 line from early mouse embryos cultured in medium conditioned by teratocarcinorna stem cells. Proc Nat1 Acad Sci USA 78,7634- 38.

Matthews KR. Tschudi C, and UUu E. (1994). A common pyrimidine-rich motif govems tram-splicing and polyadenylation of tubulin polycistronic pre-mRNA in trypanosomes. Genes Develop 8.49 1-50 1 .

Melchner H von, DeGregon JV, Raybum H, Reddy S, Friedel C, and Ruley HE. (1992). Selective disruption of genes expressed in totipotent embryonal stem cells. Genes Dev 6, 9 19-27.

Meyer D, and Birchmeier C. (1995). Multiple essential functions of neuregulin in development. Nature 378,386-390.

Mima T, Ueno H, Fischan DA, Williams LT, and Mikawa T. ( 1995). Fibroblast growth factor receptor is required for in vivo cardiac myocyte proliferation at early embryonic stages of heart development. Proc Nat1 Acad Sei USA 92,467-47 1.

Mintz B. (1962). Formation of genotypicdly rnosaic mouse embryos. Am Zoo1 2,432.

Moens CB, Stanton BR, Parada LF, and Rossant J. (1993). Defects in heart and lung development in compound heterozygotes for two different targeted mutations at the N- mye locus. Development 1 19,485-499.

Morano 1, Ritter O, Bonz A, Timek T, Vahl CF, and Michel G. (1995). Myosin light chain-actin interaction regulates cardiac contractility. Circ Res 76,720-5.

Momss-Kay G. ( 1993). Retinoic Acid and Craniofacial Development : Molecules and Morphogenesis. Bioessays 15,9- 15.

Mumane JP, and Morales IF. ( 1995). Use of a marnmalian interspersed repetitive (MIR) element in the coding and processing sequences of mammalian genes. Nucleic Acids Res 23,2837-2839.

Murphy D, BrickeII PM, Latchrnan DS, Willison K, and Rigby PWJ. (1983). Transcripts regulated during normal embryonic development and oncogenic transformation share a repetitive element. Cell35, 865-87 1.

Nagy A, Rossant J, Nagy R, Abramow-Newerly W. and Roder JC. (1993a). Derivation of completely ceii culture-derived mice fiom early-passage embryonic stem cells. Proc Nuti Acad Sci USA 90,8424-8428.

Nagy A, Rossant J, Nagy R, Abromow-Newerly W, and Roder JC. (1993b). Viable ce11 culture-derived mice from early passage embryonic stem cells. Proc Nat1 Acad Sci USA 90,8474-8477.

Nichoils RD, Fischel-Ghodsian N, and Higgs DR. (1987). Recombination at the human alpha-globin gene cluster: sequence features and topological constraints. Cell49, 369- 378.

Nilsen W. ( 1 989). Trnns-splicing in nematodes. Exp Parasitoi 69,4 1 3 4 1 6.

Nusslein-Voihard C, and Weischaus E. (1980). Mutations affecting segment number and polarity in Drosophila. Nature 287,795-801.

O ' h e C, and Gehring WJ. ( 1987). Detection in situ of genomic reguiatory elements in Drosophila. Proc Natl Acud Sci 84,9 1 23-9 1 27.

Ogura T, Alvarez IS, Vogel A, Rodriguez C, Evans RM, and Belmonte JCI. (1996). Evidence that Shh cooperates with a retinoic acid inducible CO-factor to establish ZPA- like activity. Development 122,537-542.

Ogura TT and Evans RM. (1995). A retinoic acid-triggered cascade of H O X B l gene activation. Proc Natl Acad Sci USA 92,387-39 1.

Olson EN, and Srivastava D. ( 1996). Molecular Pathways Controlling Heart Development. Science 272,67 1-676.

Orban PC, Chui D, and Marth JD. (1992). Tissue- and site-specific DNA recombination in transgenic mice. Proc Natl Acad Sci USA 89,6861-6865.

Palmiter RD, and Brinster RL. (1986). Germ-Iine transformation of mice. Ann Rev Genet 20,465-99.

Parlow MH, Bolender DL, Kokan MN, and Lough J. (1991). Locaiization of bFGF-iike proteins as punctate inclusions in the preseptation myocardium of the chicken embryo. Dev Bi01 146, 139-47.

Perlman S, Phillips CA, and Bishop ID. (1976). A study of foldback DNA. Cell8,33-42.

Perrimon N. (1995). Hedgehog and beyond. Ce12 80,s 17-520.

Peten KG, Werner S, Chen G, and Williams LT. (1992). Two F G F receptor genes are differentiaily expressed in epithelial and mesenchymal tissues during lirnb formation and organogenesis in the mouse. Development 1 14,23343.

Petenon KR, Clegg CH, Li Q, and Starnatoyannopoulos G. (1997). Production of transgenic mice with yeast artificial chromosomes. Trends in Genetics 13,6 1-66.

Poch O, Sauvager 1, Delarue M, and Tordo N. (1989). Identification of four conserved motifs among Fhe RNA-dependent polymerase -encoding elements. EMBO 1 8, 3867- 3874.

Qiu M, Bulfone A, Martinez S, Meneses JJ, Shimamwa KT Pedersen RA, and Rubenstein LR. (1995). Nul1 mutation of DLx-2 results in abnormal morphogenesis of proximal f i t

and second branchial arch denvaüves and abnormal differentiation in the forebrain. Genes Dev 9,2523-2538.

Radice GL, Rayburn H, Matsunami H, Knudsen KA, Takeichi M, and Hynes RO. (1997). Developmental defects in mouse embryos lacking N-Cadherin. ~evel&mental ~ i o l o b 18 1,6478.

Ramirez-Solis R, Zheng Et, Whiting I, Knimlauf R, and Bradley A. (1993). Hoxb-4 (Hox- 2.6) mutant mice show homeotic transformation of a cemicd vertebra and defects in the closure of the stemal rudiments, Ceil 73,279-94.

Riddle DL, Blumenthal T, Meyer BJ, and Priess IR. (1997). "C. elegans II." Cold Spring Harbor Laboratory, Cold Spring Harbor.

Robbins J, Dilworth SM, Laskey RA, and Dingwall C. (1991). Two interdependent basic domains in nucleoplasmin nuclear targeting sequence: Identification of a class of bipartite nuclear targeting sequence. Cell64.6 15-623.

Robertson E, Bradley A, Kuehn M, and Evans M. (1986). Germ-line transmission of genes introduced into cultured pluripotential cells by retroviral vector. Nature 323,445- 48.

Robertson El. (1986). Pluripotential stem celi lines as a route into the mouse germ iine. Trends Genet 2,9- 13.

Ross 1. (1996). Control of messenger RNA stability in higher eukaryotes. TIG 12, 171- 175.

Rossant 1. (1987). Ceii lineage analysis in mamrnalian embryogenesis. Curr Top Dev Bio 23, 115-146.

Rossant J. (1996). Mouse mutants and cardiac development: new molecular insights into cardiogeriesis. Circ Res 78,349-53.

Rossant J, Zimgibl R, Cado D, Shago M, and Giguere V. (1991). Expression of a retinoic acid response element-hsplacZ transgene defines specific domains of transcriptional activity during mouse embryogenesis. Gen Dev 5, 1333- 1344.

Rouyer F, Sirnmler MC, Page DC, and Weissenbach J. (1987). A sex chromosome remangement in a human XX male caused by Alu-Alu recombination. Ce115 1,417-425.

Ruberte E, Dolle P, Chambon P, and Morriss KG. (1991). Retinoic acid receptors and cellular retinoid binding proteins. II. Their differential pattern of transcription during early morphogenesis in rnouse embryos. Development 11 1,45-60.

Rudert F, Bronner S, Garnier JM, and Dolie P. (1995). Transcnpts fiom opposite strands of gamma satellite DNA are differentially expressed during mouse developmeni. Momm Genome 6,76-83.

Rudiger NS, Gregersen N, and Kielland-Brandt C. (1995). One short wel1 conserved region of Alu-sequences is involved in human gene rearrangements and has homology with prokaryotic chi. Nucleic Acidr Res 23,256-260.

Rudiger NS, Hansen PS, Jorgensen M, Faergeman O, Bolund L, and Gregersen N. (1991). Repetitive sequences involved in the recombination leadhg to deletion of exon 5 of the low-density-lipoprotein receptor gene in a patient with familial hypercholesterolernia. Eur J Biochem 198. 107- 1 L 1.

Russell ES. ( 1985). A History of Mouse Genetics. Ann Rev Genet 19, 1-28.

SafEer JD, and Thurston SJ. ( 1989). A negative regdatory element with properties similar to those of enhancers is contained within an Alu sequence. Mol Cell Biol9,355-364.

Saksela K. and Baltimore D. (1993). Negative regulation of immunoglobulin kappa ligth- chain gene transcription by a short sequence homologous to the murine B 1 repetitive element. Mol Ce11 Biol 13,3698-3705.

Sam M, Wurst W. Forrester L. Vauti F, Heng H, and Bernstein A. (1996). A novel family of repeat sequences in the mouse genome responsive to retinoic acid. Mallunalian Genome 7,74 1-748.

Sambrook J, Fritsch EF, and Maniatis T. (1989). "Molecular cloning, A laboratory manual." Cold Spring Harbor Laboratory, Cold Spriog Harbor. New York.

Sanes IR, Rubenstein JL, and Nicolas JF. (1986). Use of a recombinant retrovirus to study post-implantation cell lineage in mouse embryos. EMBO J 5, 3 133-3 142.

Sawai S, Shimono A, Wakamatsu Y, Palmes C, Hanaoka K, and Kondoh H. (1993). Defects of embryonic organogenesis resulting from targeted disruptioo of the N-myc gene in the mouse. Development 1 17, 1445-55.

Schiebel W. Haas B. Marinkovic S. Klanner A. and Sanger HL. ( 1993a). RNA-directed RNA polymerase from tomato leaves. 1. Purification and physical properties. J Bi01 Chem 268, 1 185 1-7.

Schiebel W. Haas B. Marinkovic S, Klanner A. and Sanger HL. (1993b). RNA-directed RNA polymerase from tomato leaves. II. Catalytic in vitro properties. J Bi01 Chem 268, 1 1858-67.

Schilling TF. (1997). Genetic analysis of craniofacial development in the vertebrate embryo. Bioessays 19,459-468.

Schrnid C, and Maraia R. ( L 992). Transcriptionai regulation and transpositional selec tion of active SINE sequences. Curr Opin Genet Develop 2,874-882.

Schmid CW. and Deininger PL. (1975). Sequence organization of the human genome. Cell6,345-358.

Serbedzija GN, Bronner-Fraser M, and Fraser SE. (1992). Vital dye analysis of cranial neural crest ce11 migration in the mouse embryo. Development 116,297-307.

Serbedzija GN, Fraser SET and Bro~er-Fraser M. (1990). Pathways of trunk neural crest ce11 migration in the mouse embryo as revealed by vitai dye labelling. Development 108, 605-6 12.

Shay W. ( 1996). Telornerase and cancer. TIG 12,486-487.

Shirai M, Miyashita A, Ishii N, Itoh Y, Satokata 1, Watanabe YG, and Kuwano R. (1996). A gene trap strategy for identifying the gene expressed in the embryonic nervous system. Zoolog Sci 13,277-83.

Silver LM. (1995). "Mouse genetics, concepts and applications." Oxford University Press, New York,

Simeone A, Acarnpora D, Arcioni L, Andrews PW, Boncinelli E, and Mavilio F. (1990). Sequential activation of HOX2 homeobox genes by retinoic acid in human embryonal carcinoma cells- Nature 346,763-766.

Singh K, Carey M, Saragosti S, and Botchan M. (1985). Expression of enhanced levels of small RNA polymerase III trmscnpts encoded by the B2 repeats in simian virus 40- transformed mouse cells. Nature 3 14,553-556.

Skarnes WC, Auerbach BA, and Joyner AL. (1992)- A gene trap approach in rnouse embryonic stem celis: the lac2 reported is activated by splicing, reflects endogenous gene expression, and is mutagenic in mice. Genes Dev 6,903- 18.

Skarnes WC, Moss JE, Hurtley SM, and Beddington RSP. (1995). Capturing genes encoding membrane and secreted proteins important for mouse development. Proc Nat2 Acad Sci USA 92,6592-6596.

Slack M W . (1991). "From Eggs to Embryo, Regional Specification in Early Development." Cambridge University, Cambridge.

Small J, and Scangos G. (1983). Recombinaiion during gene uansfer into mouse cells can restore the function of deleted genes. Science 2 19, 174- 176. . Smidt M, Kirsch 1, and Ratner L. (1990). Deletion of N u sequences in the fifth c-sis intron in individuals with meningiomas. J Clin Invest 86, 115 1- 1157.

Smith AG, Heath JK, Donaldson DD, Wong GG, Moreau J, Stahl M, and Rogers D. ( 1988). Inhibition of pluripotential embryonic stem ce11 differentiation by purified polypeptides. Nature 336,688-690.

Smith AIH, De Sousa MA, Kwabi-Addo B. Heppell-Parton A, Impey H, and Rabbitts P. ( 1995). A site-directed chromosomal translocation induced in embryonic stem cells by Cre-2oxP recombination. Nature Genetics 9,376-384.

Sousa R. (1996). Structural and mechanistic relationships between nucleic acid polymerases. TlBS 2 1, 186- 190.

Stanton BR, Perkins AS, Tessarollo L, Sassoon DA, and Parada LF. (1992). Loss of N- myc function results in embryonic Iethality and failure of the epithelial component of the embryo to develop. Genes di Development 6,2235-2247.

Steller H. and Pirrotta V. (1985). A transposable P vector that confen selectable G418 resistance to Drosophila larvae. EMBO J 4, 167- 17 1.

Sucov HM, Dyson ET Gumeringer CL, Pnce J, Chien KR, and Evans RM. (1994). RXR alpha mutant mice establish a genetic basis for vitamin A signaling in heart morphogenesis. Genes Dev 8, 1007- 18.

Sundaresan V, Springer P, Volpe T, Haward S, Jones JD, Dean C, Ma H, and Martienssen R. ( 1995). Patterns of gene action in plant development revealed by enhancer trap and gene trap transposable elements. Genes Dev 9, 1797-8 10.

Takeuchi T, Yamazaki Y, Katoh FY, Tsuchiya R, Kondo S, Motoyama I, and Higashinakagawa T. (1995). Gene trap capture of a novel mouse gene, jumonji, required for neural tube formation. Genes Dev 9, 12 1 1-22.

Tarkowski AK. (1961). Mouse chimeras developed from fused eggs. Nature 190, 857- 860.

Tarkowski AK. (1965). Embryonic and postnatal development of mouse chimeras. In "Ciba Foundation Symposium". Ciba Foundation.

Thomas KR, and Capecchi MR. (1987). Site-directed mutagenesis by gene targeting in mouse embryo-denved stem ceils. Ce11 5 1,503-5 12.

Thomas SA, Matsumoto AM, and Palmiter RD. (1995). Noradrenaiine is essentiai for mouse fetal development. Naiure 374,643-646.

Threadgiii DWT Dlugosz AA, Hansen LA, Temenbaum T, Lichti U, Yee D, LaMantia C, Mourton T, Hemp K, Harris RC, Barnard JA, Yuspa SH, Coffey RI, and Magnuson T. (1995). Targeted disruption of mouse EGF receptor: Effect of genetic background on mutant phenotype. Science 269,230-234.

Tini M, Otulakowski G, Breitman ML, Tsui LC, and Giguere V. (1993). An everted repeat mediates retinoic acid induction of the gF-crystallin gecrl: evidence of a direct role for retinoids in lens development. Genes Dev 7,295-307.

Tomilin NV, Iguchi-Ariga SMM, and Anga H. (1990). Transcription and replication silencer elernent is present within conserved region of human Alu repeats interacting with nuclear protein. FEBS Lea 263,69-72.

Torres M, Stoykova A, Huber O, Chowdhury K, Bonaldo P, Mansouri A, Butz S, Kemler R, and Gmss P. (1997). An a-E-catenin gene trap mutation defines its function in preimplantation developrnent. Proc N d Acad Sci 94,90 1-906.

Townley DJ, Avery BJ, Rosen B, and Skarnes WC. (1997). Rapid sequence analysis of gene trap integrations to generate a resource of insertional mutations in rnice. Genome Research 7,293-298.

Trainor PA, and Tarn PPL. (1995). Cranial mesoderm and neural crest cells of the mouse embryo: CO-distribution in the craniofacial mesenchyme but distinct segregation in branchial arches. Developrnent 12 1,2569-2582.

Tremblay P. Kessel M. and Gruss P. (1995). A transgenic neuroanatornical marker identifies cranial neural crest deficiencies associated with the Pax3 mutant Splotch. Dev Biol 17 1,3 17-29.

Vanin EF, Henthorn PS, Kioussis D, Grosveld F, and Smithies 0. (1983). Unexpected relationships between four large deletions in the human beta-globin gene cluster. Ce11 35, 70 1-709.

Vansant G, and Reynolds W. (1995). The consensus sequence of a major Alu subfamily contains a functional retinoic acid response element. Proc Natl Acad Sci USA 92, 8229- 8233.

Velculescu VE, Zhang L, Zhou W, Vogelstein J, Basrai MA, Bassett DE(Jr), Hieter P, Vogelstein B. and Kinzler KW. (1997). Characterization of the yeast transcriptome. Cell 88,243-25 1.

VeIlard M, Sureau A, Soret J, Martinerie C, and Perbol B. (1992). A potential splicing factor is encoded by the opposite strand of the trans-spliced c-myb exon. Proc Natl Acad Sei 89,251 1-2515.

Vidal F, Mougneau E, Glaichenhaus N, Vaigot P, Dannon M, and Cuzin F. (1993). Coordinated posttranscriptional control of gene expression by rnodular elements including Alu-like repetitive sequences. Proc Natl Acad Sci USA 90,208-2 12.

Viragh S, and Chailice CE. (1977). Ongin and differentiation of cardiac muscle cells in the mouse. J Ultrastruct Res 42, 1-24.

Voiloch V, Schweitzer B, Zhang X, and Rits S. (1991). Identification of negative-strand complements to cytochrome oxidase subunit III RNA in Trypanosoma bmcei. Proc Nat1 Acad Sci USA 88, 10671-5.

Wagner EF, Stewart TA, and Mintz B. (198 1). The human beta-globin gene and a functional thymidine kinase gene in developing mice. Proc Natl Acad Sci 78,50 1 6-5020.

Wagner TE, Hoppe PC, Joilick JD, Scholl DR, Hodinka RL, and Gault JB. (1981). Microinjection of a rabbit beta-globin gene into zygotes and its subsequent expression in adult mice and their offspring. Proc Natl Acad Sci 78,6376-6380.

Wallace MR, Andersen LB. Saulino AM, Gregory PE, Glover TW, and Collins FS. (199 1). A de vovo Alu insertion results in neurofibromatosis type 1. Nature 353,864-866.

Watson JB, and Sutcliffe IG. (1987). Rimate brain-specific cytoplasrnic transcript of the Alu repeat family. Mol Cell Bi01 7,3324-3327.

Wiles MV. (1993). Embryonic stem ce11 differentiation in vitro. Methods in Enrymology 225,900-9 18.

Wilkie TM, Braun RE, Ehrman WJ, Palmiter RD, and Hammer RE. (1991). Germ-line intrachromosomaI recombination restores fertility in transgenic MyK-103 male mice. Genes Devel5,38-48.

Wilkie TM, and Palmiter RD. (1987). Analysis of the integrant in the MyK-103 transgenic mice in which males fail to transmit the integrant. Mol Cell Biol7, 1646-1655.

Wilkins AS. ( 1993). "Genetic analysis of animal development." Wiley-Liss, New York.

Wood SA, Allen ND, Rossant J, Auerbach A, and Nagy A. (1993). Non-injection methods for the production of embryonic stem cell-embryo chimaeras. Nature 365, 87- 89.

Woychik RP, Stewart TA, Davis LG, D'Eustachio P. and Leder P. (1985). An inhented limb deformity created by insertional mutagenesis in a transgenic mouse. Nature 3 18.36- 40.

Wurst W, Rossant J, Prideaux V, Kownacka M, Joyner A, Hill DP, Guillemot F, Gasca S, Cado D, Auerbach A, and et, a. 1. (1995). A large-scale gene-trap screen for insertionai mutations in developmentally regulated genes in mice. Genetics 139.889-99.

Yamaguchi TP. Dumont DJ, Conlon RA, Breitman ML, and Rossant J. (1993). flk-1, a flt-related receptor tyrosine kinase is an early marker for endothelid ce11 precursors. Development 1 1 8,489498.

Young MW. (1979). Middle repetitive DNA: a fluid component of the Drosophila genome. Proc Nat1 Acad Sci USA 76,6274-6278.

Youssoufian H, and Lodish HF. (1993). Transcriptional inhibition of the murine erythropoietin receptor gene by an upstrearn repetitive element. Mol Cell Biol 13,98404.

Zachgo J, and Gossler A. (1996). Gene trapping strategies: Screens for developmentdy regulated genes and insertional mutagenesis. In "Mammalian Development" (Lonai P, Ed.). Harwood Academic, London.

Zhang J, Donaldson SH, Serbedzija G, Elsemore J, Plehn-Dujowich D, McMahon AP, Flavell RA, and Williams T. (1996). Neural tube, skeletal and body wall defects in mice lacking transcription factor AP-2. Nature 38 1,238-24 1.

Zhou QY, Quaife CI, and Palmiter RD. (1995). Targeted disruption of the tyrosine hydroxylase gene reveals that catecholamines are required for mouse fetal development. Nature 374,640-3.

IMAGE EVALUATION TEST TARGET (QA-3)

APPLIEO IMAGE. lnc & 1653 East Main Street ,=-= RochFer. NY 14- USA =-s Phone: 71 6/482-O3ûû --= Fax: 716Rûô-5989

Q 1993. AQptied Image, lm. AJî RwîS Resenred