ELUCIDATION OF IMPRINTING MECHANISMS AND PHENOTYPES IN
PRADER-WILLI SYNDROME MICE
STORMY JO CHAMBERLAIN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003
The author thanks the past and present members of the Brannan lab--Winthropian Peery, Karen Johnstone, Chris Futtner, Jessica Wairath, Susan Blaydes, Tom Simon, Mike Elmore, Todd Adamson, and Missy Shelley--for making lab such a fun place to work, for tolerating her daily, and for sticking together through some tough times. She extends special gratitude to Karen Johnstone for her gracious help with this document.
She also thanks Dan Driscoll for being the butt of her jokes and for not letting her forget that she, too, can be a butt, and Jim Resnick for accepting responsibility for her and for keeping all of us together.
The author extends her deepest gratitude to Cami Brannan for allowing her to
work on this project, for her ideas, for giving her someone to look up to, and for years too few of friendship.
Finally, the author thanks her parents for all of their support and for making her come home when she needs to.
TABLE OF CONTENTS
ACKNOW LEDGM ENTS ............................................................................................... ii
LIST OF ABBREVIATIONS ...................................................................... vi
ABSTRACT ........................................................................................ viii
1. INTRODUCTION ......................................................................................................... I
Genomic Imprinting ............................................................................................ I
The Prader W illi and Angelman Syndromes ........................................................ 4
2. IM PRINTED REGULATION OF UBE3A .................................................................... 9
Introduction ............................................................................................................. 9
M aterials and M ethods .......................................................................................... 11
Strains and M ratings ........................................................................................... 11
Identification of Polymorphisms ........................................................................ 12
RT-PCR ............................................................................................................. 12
Results ................................................................................................................... 13
Discussion ............................................................................................................. 15
3. IM PRINTED TRANSGENES ..................................................................................... 23
Introduction ........................................................................................................... 23
Use of Transgenic M ice to Study Gene Regulation ............................................ 23
M aterials and M ethods ......................................................................................... 26
Screening the BAC Libraries ............................................................................. 26
Probe Preparation .............................................................................................. 28
Verification of BAC Clones ............................................................................... 29
End Sequencing BAC Clones ............................................................................ 29
Pulse-Field Gel Electrophoresis ......................................................................... 30
Southern Blot ..................................................................................................... 30
Transgenic M ice Production .............................................................................. 31
M ouse Husbandry .............................................................................................. 32
M ouse Strains .................................................................................................... 32
Results ................................................................................................................... 32
YAC Studies ...................................................................................................... 32
129/Sv BAC Library Screen .............................................................................. 33
RecA M ediated Homologous Recombination .................................................... 34
RecET M ediated Hom ologous Recombination ................................................... 36
C57BL/6J BAC Library Screen ......................................................................... 36
Lambda Red M ediated Hom ologous Recombination .......................................... 38
Production Of Transgenic M ice ......................................................................... 39
Characterization Of Transgenic Lines ................................................................ 40
The M arked Endogenous Snrpn Locus ............................................................... 41
Im printed Expression From Transgenes ............................................................. 42
Discussion ............................................................................................................. 46
4. STRAIN-DEPENDENT DIFFERENCES IN PHENOTYPE ..................................... 64
Introduction ........................................................................................................... 64
M aterials and M ethods .......................................................................................... 66
Strains and M ratings .......................................................................................... 66
Culling and Fostering ........................................................................................ 67
Identification of Polym orphism s ........................................................................ 67
RT-PCR ............................................................................................................. 68
Northern Blot Analysis ...................................................................................... 69
Results ................................................................................................................... 70
Discussion ............................................................................................................. 72
5. TRANSGENIC RESCUE OF THE PWS-IC DELETION MOUSE ........................... 80
Introduction ........................................................................................................... 80
M aterials and M ethods .......................................................................................... 85
Screening the BAC Libraries ............................................................................. 85
Probe Preparation .............................................................................................. 86
Verification of BAC Clones ............................................................................... 86
End Sequencing BAC Clones ............................................................................ 87
Pulse-Field Gel Electrophoresis ......................................................................... 87
Southern Blot ..................................................................................................... 88
Transgenic M ice Production .............................................................................. 88
M ouse Husbandry .............................................................................................. 89
M ouse Strains .................................................................................................... 89
Results ................................................................................................................... 90
Screening the BAC Library for Clones ............................................................... 90
Production and Characterization of Transgenic M ice ......................................... 91
Transgenic Rescue Experim ents: 454 Transgenic Lines .................................... 94
Transgenic Rescue Experiments: 380 and 1707 Transgenic Lines ..................... 95
Discussion ............................................................................................................. 96
BAC Isolation .................................................................................................... 96
Rearranged Transgenes ...................................................................................... 97
Transgenic Rescue ............................................................................................. 99
6. CONCLUSIONS AND FUTURE DIRECTIONS .................................................... 110
REFERENCES ........................................................................................................... 114
BIOGRAPHICAL SKETCH ....................................................................................... 121
LIST OF ABBREVIATIONS AS Angelman syndrome
AS-IC Angelman syndrome imprinting center
AS-SRO Angelman syndrome smallest region of deletion overlap
BAC bacterial artificial chromosome
cDNA complementary deoxyribonucleic acid
DNA deoxyribonucleic acid
DNaseI deoxyribonuclease I
IAP interstitial A particle
IC imprinting center
ICR imprinting control region
Igf2 insulin-like growth factor-2 gene
Igf2r insulin-like growth factor-2 receptor gene
Ipw imprinted in Prader-Willi gene
Magel2 mage-like-2 gene
Mkrn3 makorin-ring-3 gene
mRNA messenger ribonucleic acid
NCBI National Center for Biotechnology Information
NCI National Cancer Institute
Ndn Necdin gene
PCR polymerase chain reaction
PWS Prader-Willi syndrome
PWS-IC Prader-Willi syndrome imprinting center
PWS-SRO Prader-Willi syndrome smallest region of deletion overlap
RNA ribonucleic acid
RNaseI ribonuclease I
RPCI Roswell Park Cancer Institute
RT reverse transcription
RT-PCR reverse transcription polymerase chain reaction
SmN small nuclear ribonucleoprotein particle N protein
snoRNA small nucleolar ribonucleic acid
Snrpn small nuclear ribonucleoprotein particle N gene
Snurf Snrpn upstream reading frame
Ube3A ubiquitin 3A ligase
UPD uniparental disomy
YAC yeast artificial chromosome
ZNF127 human zinc finger protein 127
Zfpl127 murine zinc finger protein 127
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
ELUCIDATION OF IMPRINTING MECHANISMS AND PHENOTYPES IN PRADER-WILLI SYNDROME MICE By
Stormy Jo Chamberlain
Chair: James Resnick
Major Department: Molecular Genetics and Microbiology
Genomic imprinting is an epigenetic phenomenon in which genes are expressed exclusively from one parental allele or the other. Prader-Willi syndrome (PWS) and Angelman syndrome (AS) are human genetic disorders resulting from the absence of genes that are subject to genomic imprinting. PWS is a contiguous gene syndrome that results from the absence of two or more paternally expressed genes, while AS is a disorder that can result from the absence of a single maternally expressed gene, UBE3A, which lies in the same region.
This study uses a mouse model for PWS in which paternal transmission of a PWS imprinting center (PWS-IC) mutation results in complete loss of local paternal gene expression. This targeted mutation accurately models both the molecular and phenotypic aspects of PWS, and it has been applied to both of these aspects in this study.
It has first been shown that imprinted expression of Ube3a, most likely results from the regulation of a paternally expressed transcript that is antisense to the Ube3a gene, suggesting an interesting model for the regulation of gene expression in this imprinted region. The minimum sequence that is sufficient to confer proper imprinted expression of the paternally expressed genes to transgenes in vivo was then identified. This provides a stringent assay that can be used to test the proposed model of imprinted gene expression.
Transgenic rescue of the PWS mouse was also used in our mouse model to
correlate specific attributes of the PWS phenotype to specific gene products. A transgene that shows expression of a single small nucleolar RNA (snoRINA) cluster rescues the small stature phenotype of PWS mice, while three other genes were found to have no obvious effect on the PWS phenotype in mice. We have also found that the PWS phenotype is less severe on certain strain backgrounds, suggesting the presence of modifier genes that may ameliorate the PWS phenotype. These modifier genes may help to identify gene pathways involved in PWS as well as future targets for gene therapy that may ultimately help the PWS patient.
Prader-Willi and Angelman syndromes (PWS and AS, respectively) are two
human genetic disorders that result from deficiencies in genes that are subject to genomic imprinting. The causative genetic regions for both of these syndromes have been mapped to approximately 3 Mb in the 15ql 1-q13 region. They provide a clinically and scientifically interesting situation in which to study the mechanisms and phenotypes involved with one region of genomic imprinting.
Genomic imprinting is a phenomenon in which genes are expressed in a parentof-origin specific manner. Both parental genomes, in fact, are absolutely required for normal development of a mammal (mouse). Work from the laboratories of Davor Solter and Azim Surani showed that embryos with two genetic contributions from the same parent would not develop normally. 2,3 Embryos with two male pronuclei (androgenetic) developed into tissues that were mostly extraembryonic in nature, while those with two female pronuclei (parthenogenetic) developed into tissues that represented the embryo proper. 2 Cattanach showed that uniparental disomy, or the inheritance of specific chromosome segments from a single parent, also led to developmental abnormalities, further suggesting the importance of both parental genomes. This also led to the idea that it may be specific genes or regions that were differentially regulated between maternal and paternal alleles, and thus responsible for the failure of parthenogenetic or
androgenetic embryos to develop correctly. These experiments suggested that certain genes in the mammalian genome were subjected to genomic imprinting and are expressed exclusively from the paternally inherited allele and silenced from the maternally inherited allele, while others are expressed from the maternally inherited allele and repressed from the paternally inherited allele. '
The parent-of-origin specific expression is maintained from generation to
generation, and so with each passage through the germline, the past parental "label" of an imprinted gene is erased, and the gene is "re-labeled", based on the parent that it is now inherited from. 6.7 The ability of the two parental alleles to always be distinguished from one another is fascinating, especially since these alleles can be identical. The existence of identical alleles that are somehow regulated differentially indicates that genomic imprinting involves an epigenetic phenomenon. Epigenetic, meaning 'outside conventional genetics', 9 refers to the distinguishing mark being something other than changes in sequence. There is now considerable evidence to suggest that chemical modifications, such as methylation or acetylation, to DNA and proteins associated with DNA are responsible, in part, for the epigenetic changes that distinguish one parental allele from another. '0 DNA methylation has been known to be associated with silenced alleles, although some preferentially expressed alleles are known to be associated with DNA methylation. 6.11-13 The histone code has also been shown to be associated with either the maintenance or establishment of imprinted domains. The current understanding of how DNA methylation collaborates with proteins that modify nucleosomes to establish a stable chromatin state and epigenetically regulate gene
expression is nicely reviewed by Rudolf Jaenish and Adrian Bird in the March 2003 volume of Nature Genetics. '
Is extremely important to understand the cis acting regulatory sequences that harbor the epigenetic changes and translate them to changes in gene expression, since gene expression from one parental allele or the other is the ultimate goal of genomic imprinting. While many imprinted regions, such as H19-1g12 and Igf2r have been studied in great depth, it has become apparent that different imprinted regions are not necessarily subjected to the same regulation paradigms.
The PWS/AS region of genomic imprinting is an attractive imprinted region to
study for several reasons. First, the imprinted region spans a distance of approximately 3 Mb. The mechanism of a control element that influences gene regulation at such great distances is particularly intriguing. Secondly, this region seems relatively simple. At the outset of this study, there were four known paternally expressed genes and one known maternally expressed gene in the region. "-9 It was also known that this region was regulated by a bipartite imprinting center, with one portion regulating the maternally expressed gene, and the other regulating the paternally expressed genes. 20 Thirdly, at least two mouse models with disrupted imprinting in the PWS/AS region suggested that the imprinting mechanism was conserved between mouse and human, providing a genetic system in which to study imprinting. "' Most importantly, the PWS/AS region has clinical significance, as it was originally identified as a genetic region that was responsible for human disease.
The Prader Willi and Angelman Syndromes PWS is characterized by neonatal hypotonia, failure to thrive, hypogonadism, cryptorchidism, and poor suckle. 23 By 18-36 months, the failure to thrive and poor suckle symptoms subside, and PWS children begin to develop obesity that is complicated by hyperphagia, physical and mental developmental delay that leads to moderate mental retardation and small stature, and behavioral problems reminiscent of obsessivecompulsive disorder.13 AS is characterized by severe mental retardation, ataxic gait, inappropriate laughter, happy affect, and almost absent speech.'4 Each of these disorders occur at a frequency of approximately 1 in 15,000 live births. 24-26 While PWS and AS are relatively rare, they provide a closed genetic system in which to investigate human disorders, such as obesity, failure to thrive, ataxia, and mental retardation, that are much more common.
Ledbetter et al. first noticed that patients with PWS often had cytologically visible deletions of human 15q II-q 13. 14 Butler and Palmer later reported that these same deletions were occurring on the paternally inherited chromosome. 27 Finally, Nicholls et al. determined that maternal uniparental disomy was also a mechanism by which PWS could occur. 2" This led to the suspicion that PWS occurs as from the physical or functional deletion of the genes located on the paternally inherited chromosome. AS, on the other hand was known to be associated with similar deletions that were occurring on the maternally inherited chromosome. ",' Paternal uniparental disomy was also shown to result in AS. 332AS was found to be the result of the physical or functional deletion of maternal 15qI-q 13.
Patients with PWS lack expression of all of the paternally expressed genes in the region and can be divided into 3 genetic classes--maternal uniparental disomy patients, 28 deletion patients, and imprinting defect patients. 6.2 Maternal uniparental disomy patients have two chromosomes 15 that were inherited from their mother, presumably as a result of nondisjunction, followed by reduction in the zygote. Absence of a paternal allele of 15qI 1 -q 13 causes the absence of expression of the paternally expressed genes. The deletion patients have a 3-4 Mb deletion of the paternally inherited 1 5q 11 -q 13, which is facilitated by repeated regions that flank the entire imprinted domain. These repeats cause this type of deletion to occur at a comparatively high frequency. Patients with this deletion are physically missing the paternally inherited allele from which genes are expressed. Imprinting mutations were identified in patients that had inherited both a maternal and paternal chromosome 1 5q 1 -q 13, but the paternal chromosome had a maternal epigenotype as assayed by methylation at the SNRPN gene. Some of these imprinting mutations were found to be due to deletions of the PWS-IC and result in the silencing of the paternally expressed genes. 6,11-1320 Other imprinting mutations show no apparent deletion or mutation in the PWS-IC, and are thought to be epimutations of unknown origin. "
AS patients can be divided into five genetic classes--paternal uniparental disomy
patients, 32 deletion patients, 2"0 imprinting defect patients, 6.20 UBE3A mutation patients, 34,35 and biparental patients with an unidentified genetic lesion. Paternal uniparental disomy patients have two chromosomes 15 from their father. Again, nondisjunction is thought to be responsible for this, but it occurs less frequently in males than females, accounting for the reduced frequency of paternal uniparental disomy
patients. AS deletion patients are physically missing the maternal allele of 15q 11-q 13, and thus the two maternally expressed genes in the region. Imprinting mutations that result from microdeletion of the AS-IC disrupt the maternal pattern of gene expression. These imprinting mutations were shown to be small deletions upstream of the PWS-IC deletion mutations and never overlap them. Imprinting mutations in AS patients have been identified by mutation of the putative AS-IC and not by epimutation, since no known methylation mark has been associated with AS (A. Lossie and D.J. Driscoll, personal communication). Finally, the two biparental classes of AS patients indicate that AS can be caused by disruption of the UBE3A gene and that other genes are likely to be involved in the UBE3A pathway. "" The contribution of the other maternally expressed gene, ATPJOC to the AS phenotype is not well understood. 36
The imprinting center mutations from several PWS patients were mapped to identify the smallest region of deletion overlap for PWS (PWS-SRO). 1'3 The PWSSRO is located 5' of the SNRPN gene, including the first exon of the gene. The PWSSRO is often referred to as the PWS imprinting center or PWS-IC. Although the PWSSRO has been narrowed to approximately 4.3 kb, the actual deletions that have been shown to cause PWS are larger. 37,38 A smallest region of deletion overlap has also been identified for AS, and is referred to as the AS-SRO. The AS-SRO lies around 35 kb upstream of the PWS-SRO, and has been narrowed to 0.8 kb. 37.38 AS imprinting mutations never appear to overlap the PWS-SRO, while PWS imprinting mutations can overlap the AS-SRO. This also indicates a bipartite structure to the imprinting center, in which the PWS and AS components work together to control imprinting in the 15qI 1q 13 region.
The imprinting mutation class of patients for PWS and AS also suggests that the PWS-IC is required for the paternal pattern of gene expression across this approximately
3 Mb region, while the AS-IC is required for the maternal pattern of gene expression. Human chromosome 15ql 1-q13 contains at least six paternally expressed genes or transcripts including MKRN3, MAGEL2, NDN, SNRPN, HBII-13, HBII-436, HBII-437, HBII-438A, HBII-438B, HBII-85, HBII-52, and the UBE3A-antisense transcript 188.8.131.52-44 and two maternally expressed genes, UBE3A and ATPIOC. 19,36
The PWS/AS region lies in a region of synteny with a region of central mouse chromosome 7, 17 and the gene order and structure is highly conserved. In addition, two mouse models-one a model for uniparental disomy, 21 and the other a targeted mutation in the PWS-IC 22 -suggest that the imprinted regulation of the genes in this region is also conserved. Mice are also capable of showing physical phenotypes such as obesity, failure to thrive, and small stature as well as behavioral phenotypes such as obsessivecompulsive disorder. Making the murine system a robust one for the study of the molecular and phenotypic aspects of PWS and AS.
This dissertation focuses on the mechanisms and phenotypes of PWS. The rich genetic conservation between mouse and human has been harnessed to investigate these aspects of the disorder. This dissertation represents the four major accomplishments. First of all, the maternally expressed Ube3a gene was shown to be negatively regulated by an antisense transcript that is, in turn, positively regulated by the PWS-IC. This led to an attractive model for gene regulation in this region. Secondly, transgenes that show correct imprinted expression were identified, thus defining the minimal area sufficient to confer appropriate imprinted expression in vivo. These transgenes also provided
evidence for the functional conservation of the AS-IC in mouse as well as its approximate location in the region. Thirdly, the PWS phenotype in mice is influenced by different genetic backgrounds in mice. This discovery revealed a different phenotype in the PWS mouse model, as well as a method for identifying the particular modifying genes. Finally, transgenic rescue has identified a gene that is responsible for the small stature phenotype in the PWS mouse model and revealed a phenotype associated with snoRNAs. For ease of reading, each chapter contains sufficient background and methodology so that the reader may understand each one independently.
IMPRINTED REGULATION OF UBE3A Introduction
Angelman syndrome (AS) and Prader-Willi syndrome (PWS) are caused by
defects in genes subject to genomic imprinting. AS results from a lack of a maternally inherited pattern of gene expression of human chromosome 15q 1 1-q 13, whereas PWS results from a lack of a paternally inherited pattern of gene expression in the l 5ql 1-q 13 region. The imprinted genes involved in AS and PWS are regulated by a bipartite imprinting center (IC) located upstream of the SNRPN gene. 26The IC is divided into the Angelman syndrome imprinting center (AS-IC) and the Prader-Willi syndrome imprinting center (PWS-IC). Many of the mouse orthologs of these imprinted genes are located in the syntenic region of murine chromosome 7, 17,402,46-48 furthermore, it is known that the PWS-IC is functionally conserved in the mouse. 22
Patients with PWS lack expression of multiple paternally expressed genes.
184.108.40.206-44 PWS patients can be divided into 3 genetic classes: (1) maternal uniparental disomy for chromosome 15; (2) 3-4 Mb deletion of the paternally inherited 15ql 1-q13, or
(3) imprinting mutations that are due to deletions of the PWS-IC and result in the silencing of the paternally expressed genes. 6.49-51 This latter class of patients indicates that the PWS-IC is required for the paternal pattern of gene expression across this approximately 2 Mb region.
AS patients can be divided into five genetic classes: (1) paternal uniparental
disomy for chromosome 15; (2) 3-4 Mb deletion of the paternally inherited 15ql 1-q13;
(3) imprinting mutations that result from microdeletion of the AS-IC and subsequent disruption of the maternal pattern of gene expression; (4) biparental with a mutation in the maternally derived UBE3A gene; or (5) biparental with another as yet unidentified anomaly that cannot be described as an imprinting mutation or deletion. 6,34.50,52 It is the fourth, biparental class that has demonstrated that mutations in the UBE3A are sufficient to cause AS.
Contrary to the known paternally expressed transcripts in the region, UBE3A does not exhibit strict imprinted expression: UBE3A is expressed predominantly from maternal chromosome 15 in the brain but shows biallelic expression in fibroblasts and lymphoblasts.9"" Albrecht et al. demonstrated that while the murine Ube3a gene is expressed exclusively from the maternal allele in hippocampal neurons and Purkinje cells, it is expressed from both alleles in other regions of the mouse brain.48 Another difference between paternally expressed transcripts and UBE3A is that for most of the paternal genes, a region of differential methylation exists between the maternal and paternal alleles.3.4'" This is not the case for UBE3A, as to date, no differentially methylated CpG sites have been shown to exist near UBE3A (A. Lossie and D.J. Driscoll, personal communication). Together, these differences in imprinted expression and parental-specific methylation suggest that the mechanism that regulates the paternal program of gene expression in this region may not be the same mechanism as that which regulates UBE3A imprinted expression.
Rougeulle et al. reported the existence of an antisense transcript to the UBE3A gene. 41 Using RT-PCR of RNA derived from AS and PWS patient brains, they determined that this antisense transcript was imprinted, expressed only from the paternal allele. Rougeulle et al. suggested that the maternal-only expression of UBE3A may result from tissue specific expression of this paternally expressed antisense transcript. Therefore, using a PWS-IC deletion (APWS-IC) mouse model,22,5 the role of the PWS-IC in the regulation of the murine Ube3a antisense transcript and its subsequent effects on the parent-of-origin specific expression of the sense Ube3a transcript were investigated. It was found that the murine Ube3a transcript is expressed exclusively from the paternal allele, the antisense transcript is regulated by the PWS-IC, and the level of paternal vs. maternal Ube3a expression is different between APWS-IC mice and wild-type littermates, suggesting a role for the antisense transcript in the regulation of maternal Ube3 expression
Materials and Methods
Strains and Matings
Mouse strains used were mus musculus 129/Sv (129/Sv or D) and a strain of
C57BL/6J that is congenic for the PWS/AS syntenic region of chromosome 7 from mus castaneus (B6.CAST.c7 or C) 56. For (B6.CAST.c7 X 129/Sv) F, matings, a female B6.CAST.c7 mouse was mated with a male APWS-IC mouse on a 129/Sv strain background. Both APWS-IC mice and wildtype littermates were used. The (129/Sv X B6.CAST.c7) F, mice were generated by mating a female 129/Sv mouse with a male B6.CASTc7 mouse.
Identification of Polymorphisms
Total brain RNA was prepared from both 129/Sv mice and B6.CAST.c7 mice. Each RNA sample was used to create single-stranded cDNA using random primers. These cDNAs were used to program PCR reactions using primers corresponding to Ube3a exons 4 and 7 (5'-CCTGCAGACTTGAAGAAGCAG-3' and 5'GAAAACCTCTGCGAAATGCCTT-3'). The resulting products were cloned, sequenced and compared. A polymorphism found in exon 5 created a Tsp509I restriction site in the 129/Sv clone and a Bst4CI site at the same location in the B6.CAST.c7 clone. RT-PCR
Total RNA was isolated from total brain obtained from neonatal mice. RNA was extracted using RNAzol (Tel-Test, Inc.) according to instructions. This total brain RNA (10tg) was then pretreated with DNAse I (Life Technologies) and half of the reaction was subsequently used to synthesize first strand cDNA with Superscript II reverse transcriptase (RT) (Life Tehcnologies) and either strand-specific primer SF or random primers (Life Technologies). The other half of the reaction was manipulated in parallel in the absence of RT. One-twentieth of the +RT or -RT reactions were used to seed PCR reactions using the following conditions: 10mM Tris-HC1, pH 8.3, 50mM KCI, 0.125mM of the four dNTPs, 1 unit Taq DNA polymerase (Boehringer), and 4tM of each of the appropriate primers. Sequences of the primers were 5F (5'CACATATGATGAAGCTACGA-3'), 5iR (5'-CAGAAAGAGAAGTGAGGTTG-3'), and 6R (5'-CACACTCCCTTCATATTCC-3'). PCR amplification conditions were: 950C for 5 min, followed by 30 cycles of 94*C for 30 s, 60*C for 45 s, and 720C for 45 s. The final cycle was followed by an extension step for 10 min at 72*C. The PCR products
were either left undigested or digested with enzymes and analyzed on either a 2% or
4.8% agarose gel.
The expression of a Ube3a antisense transcript was first verified in mouse brain by performing PCR across intron 5 using randomly primed murine brain cDNA. Two bands were amplified from the cDNA: the smaller correctly spliced Ube3a transcript and the larger unspliced form of the Ube3a gene, which was verified by sequencing (data not shown). The existence of this unspliced form is consistent with a Ube3a antisense transcript that is expressed in mouse brain. The presence of the antisense transcript was then confirmed, and its paternal specific expression was determined using F, progeny derived from reciprocal matings between wild-type 129/Sv mice and B6.CAST.c7 mice which are congenic for the PWS/AS syntenic region from mus musculus castaneus chromosome 7 on a C57BL/6J background. 56 DNAseI treated RNA prepared from brains of F, mice was used to program strand-specific reverse transcription reactions using a forward primer from exon 5 (5F; Fig. 2-1A), thus yielding cDNA made from the antisense transcript. Using this strand specific cDNA as a template, PCR was performed with the forward primer 5F and a reverse primer from intron 5 (5iR; Fig. 2-1A). A polymorphism in Ube3a exon 5 between Mus musculus domesticus and Mus musculus castaneus enabled the determination of the parent of origin by digestion of the RT-PCR products with the enzyme Tsp509I, which digests only the domesticus allele. The RTPCR product derived from the antisense transcript was found to be resistant to Tsp509I digestion when the father was B6.CAST.c7, but was cleaved by this enzyme when the
father was 129/Sv. This demonstrates that the antisense Ube3a transcript is expressed exclusively from the paternally inherited allele (Fig. 2-1B).
As previous studies have shown that all other paternally expressed transcripts in the region are regulated by the PWS-IC in humans and mice, 6.2.".20.22.15 it was necessary to determine whether the paternally expressed Ube3a antisense transcript was also subject to regulation by the PWS-IC. Therefore, the PWS-IC deletion (APWS-IC) mouse strain was used. This mouse strain was created by targeted deletion in ES cells that includes exons 1 through 6 of Snrpn and extends approximately 23 kb 5' of the gene. 225 B6.CAST.c7 females were mated with APWS-IC males and total brain RNA was isolated from the resulting newborn (B6.CAST.c7 X APWS-IC) F, mice. The RNA was DNAseI treated and used to direct strand-specific RT reactions, each containing the sense 5F primer and an antisense f3-actin primer. The resulting cDNA was used as a template for PCR using the primers 5F and 5iR, or two -actin primers. Figure 2-2 shows that the Ube3a antisense-derived 840 bp band present in the wild type cDNA is absent in the APWS-IC littermate cDNA. In contrast, actin is detected in both cDNA preparations. This result demonstrates that the Ube3a antisense transcript is positively regulated by the PWS-IC.
Finally, to determine if the paternal Ube3a antisense transcript regulates the level of paternal Ube3a mRNA, RT-PCR followed by restriction enzyme digestion was performed to examine the levels of Ube3a expression produced from each parental allele in both wild-type and APWS-IC mice. To distinguish between Ube3a expression derived from the two parental alleles, two exon 5 polymorphisms between Mus musculus domesticus (D) and Mus musculus castaneus (C) were used: the enzyme Bst4CI cuts the
castaneus allele once but does not cut the domesticus allele; and the enzyme Tsp509I cuts the domesticus allele twice and the castaneus allele only once (Fig. 2-3A). The experimental cross of B6.CAST.c7 X APWS-IC was established to obtain both wild-type F1 pups (C X D wt) and APWS-IC F, pups (C X D A). The control strain reciprocal cross of 129/Sv X B6.CAST.c7 was also established to obtain wild-type F, pups (D X C wt). After birth, the pups were sacrificed, total brain RNA from all three classes was isolated, and then random primers were used to prepare cDNA. Subsequent PCR amplification with the Ube3a primers 5F and 6R demonstrated that Ube3a was detectable in all three classes of F, mice (Fig. 2-3B, first three lanes). Upon digestion of these RT-PCR products Bst4CI, expression of both parental Ube3a alleles was detected in all three classes. These results are expected, since Ube3a is known be expressed from both alleles in most regions of the brain, exhibiting imprinted expression in only a subset of brain cells. 4 However, in the PWS-AIC F, lane (C X D A) the ratio of paternal (uncut) to maternal (cut) Ube3a expression was found to be elevated compared to that observed in the wildtype F, lane (C X D wt). These results were confirmed by digestion of the RTPCR products with Tsp509I. In this case, the paternal Ube3a allele (138 bp + 87 bp bands) is clearly present at significantly higher levels in the APWS-IC F, lane (C X D A) than the wildtype F, lane (C X D wt). These results demonstrate that the loss of paternal Ube3a antisense expression in APWS-IC mice is accompanied by an increase in paternal sense Ube3a expression.
In this study, the murine Ube3a antisense transcript was shown to be exclusively expressed from the paternal allele and positively regulated by the PWS-IC. Furthermore,
mice that inherit a paternal deletion of the PWS-IC, and thus lack Ube3a antisense expression have an upregulated paternal Ube3a allele relative to the maternal allele, as compared to the wild-type controls. Together, these results strongly suggest that the paternally derived antisense transcript negatively regulates the level of paternally expressed Ube3a mRNA.
The differences between the imprinted expression of paternally expressed genes in the region and UBE3A strongly suggest that the maternal-only expression of the UBE3A gene is regulated by a different mechanism. The hypothesis that the tissue specific imprinting of UBE3A is an indirect consequence of the imprinted antisense transcript has been previously proposed by this lab and others.4"" Specifically, in tissues that express the antisense transcript from the paternal allele, production of UBE3A mRNA from the paternal allele is inhibited (presumably this inhibition occurs at the transcriptional level). However, the maternal UBE3A allele is not prevented from producing mRNA, so in these tissues, maternal specific expression is observed (Fig. 24A). In tissues that do not express the antisense transcript but do express the UBE3A gene, biallelic expression of UBE3A is observed (Fig. 2-4A). The data presented here using the mouse model are consistent with this hypothesis and further suggest that the imprinted expression of UBE3A is not a direct consequence of the UBE3A promoter or the AS-IC. Rather, it appears to be the PWS-IC which negatively regulates the expression of paternal UBE3A via the antisense transcript.
These data support a model in which the primary targets of imprinting in the PWS/AS region are the paternally expressed transcripts, one of which, the UBE3A antisense transcript, results in maternal-specific expression of the UBE3A gene. The
advantage of this model is that it greatly simplifies the seemingly complex pattern of gene regulation in the PWS/AS region. A positive regulatory element, the PWS-IC, is responsible for establishing and maintaining the paternal mode of gene expression in somatic tissues (Fig. 2-413). A negative regulatory element, which is assumed to be the AS-IC, serves to prevent activation of the paternal program on the maternally inherited chromosome, presumably by blocking PWS-IC function in the maternal germline (Fig. 24C). As a result, different patterns of gene expression are observed for the two parental alleles. The paternally expressed genes including the UBE3A antisense transcript are expressed from the paternal chromosome but not from the maternal chromosome. In contrast, UBE3A is "constitutively" expressed from the maternal chromosome in all cell types that exhibit UBE3A expression, whereas the paternal UBE3A allele is only "allowed" to be expressed in cell types that do not express the antisense transcript. Therefore, according to this model, it is only the paternally expressed genes that are subject to regulation by the IC as a whole. This reduces the complex regulation in this region to a simple ON/OFF decision: the PWS-IC operates in the ON mode unless shut OFF by the AS-IC element. Therefore, the key to understanding imprinting in this region will be to determine how the AS-IC inhibits the PWS-IC on the maternal chromosome.
Finally, as the antisense transcript appears to be the cause of the tissue specific imprinting of UBE3A, it will be interesting to determine what regulatory elements result in the expression of the antisense in such a restricted cell population in the brain. It is possible that there are positive regulatory elements that drive expression of the antisense transcript in only a few cell types. Alternatively, there may be negative regulatory elements that prevent expression of the antisense in most cell types. At this point, it can
only be said that the PWS-IC exerts a positive effect on the transcription of the antisense. There may be additional levels of control that restrict the antisense to fewer cell types than the other PWS-IC regulated paternally expressed genes in the region.
Exon 5 Exon 6
C (B6.Cast.c7) 21 bp 640 bp 179 bp
D (129/Sv) 21 bpt 87bp 553bp 179bp
B uncut Tsp509I No RT
I -- I I --I r __ lmaternal C D C D C D
Xx x x x x
paternal D C C D D C C D D C
Figure 2-1. The antisense transcript is expressed from the paternally-derived
allele only. A. Map of the exon 5 to exon 6 region of Ube3a and schematic
showing expected PCR products, restriction sites, and fragment sizes. Triangle lollipops represent Tsp509I sites, and horizontal arrows represent primers used
for RT and PCR. B. The antisense transcript was amplified by RT-PCR from
neonatal brains from (B6.CAST.c7 X 129/Sv) Fi and (129/Sv X B6.CAST.c7) F1
mice, and by PCR from genomic DNA from parental B6.CAST.c7, and 129/Sv
mice using primers 5F and SiR. The products were digested with Tsp509I and electrophoresed along with the uncut controls on a 2% agarose gel. C indicates
the B6.CAST.c7 strain, while D indicates the domesticus strain, 129/Sv. Arrows
indicate pertinent fragments and their sizes.
maternal C C
paternal D D
genotype wt A
Figure 2-2. The antisense transcript is not expressed in the APWS-IC mouse. cDNAs were made from a APWS-IC mouse and a wildtype littermate using primers 5F and the 3' fl-actin control. PCR was done using primers 5F and 5iR. The resulting products were run on a 2% agarose gel.
A C (B6.Cast.c7) D (129/Sv)
Exon 5 Exn 6 on 5 1Exon
246 bp 246 bp
108 bp 138 bp Bst 4CI 246 bp
~t Tsp509I t
21 bp 225 bp 21 bp 87 bp 138 bp
uncut Bst4CI Tsp509I No RT
I I I I I I I I
maternal C C D C C D C C D C C D
X XX X x x x x x XXX
paternal D D C D D C D D C D D C genotype wt A wt wt A wt wt A wt wt A wt
246 bp 225 bp
138 bp --- 138 bp
Figure 2-3. Paternal Ube3a expression is increased in APWS-IC mice. A.
B6.CAST.c7 and 129/SvEv alleles are diagrammed showing PCR products,
restriction sites, and fragment sizes. The circle lollipop indicates the location of a Bst4CI site, while the triangle popsicle indicates the Tsp509I sites. Horizontal
arrows show primers used for RT and PCR. B. Sense Ube3a cDNA was made
using RNA derived from (B6.CAST.c7 X 129/Sv) F1, (B6.CAST.c7 X APWSIC) F1, and (129/Sv X B6.CAST.c7) F1 brains using random primers and PCR
amplified using primers 5F and 6R. The spliced, sense band was gel purified
and each sample was divided three ways: one portion was left uncut, the second
was cut with Bst4CI, and the third was cut with Tsp509I. These samples were
loaded on a 4.8% agarose gel. C indicates the B6.CAST.c7 strain, while D
indicates the domesticus strain, 129/SvEv. For the genotype, wt indicates a wildtype mouse; A indicates a APWS-IC mouse. Arrows indicate the location of the
fragments and their sizes.
A Mkrn3,Magel2, Ndn Snrpn Ube3a AtplOC
F* .. .. .. ... 11 ....... D
Figure 2-4. Model of gene regulation in the PWS/AS region. A. In the
brain, two different patterns of UBE3A expression occur. In cell types that express the paternal antisense transcript, UBE3A is imprinted (top box). In
cell types that do not express the antisense transcript, UBE3A is biallelically
expressed (bottom box). B. On the paternally-inherited chromosome, the
PWS-IC establishes the paternal mode of gene expression. C. On the
maternally-inherited chromosome, the AS-IC prevents activation of the
paternal program, presumably by blocking PWS-IC function.
Use of Transgenic Mice to Study Gene Regulation
Transgenes have been used to study regulation of gene expression in many systems. They serve to isolate regulatory elements and relocate them to an ectopic location in the genome where their local activities may be separated from the influence of adjacent cooperating regulatory elements. One of the simplest instances of using transgenes to study regulatory elements involved the tyrosinase gene.589 A 250 kb yeast artificial chromosome (YAC) was found to faithfully express the tyrosinase gene.8 This YAC was subsequently engineered to harbor a deletion of a DNaseI hypersensitive site, demonstrating that the site is necessary for correct expression of the tyrosinase gene.9 The regulation of the 1-globin locus is another example that has benefited from the use of transgenes. In the 13-globin locus, long distance regulatory elements determine the tissue specificity and developmental timing of the expression of various exons in the locus. 60 Similar to the tyrosinase locus, DNaseI hypersensitive sites were deleted from a YAC using homologous recombination, and were determined to be locus control regions in vivo by demonstrating that YACs that were missing or had duplications of these hypersensitive sites no longer showed correct temporal expression of the P3-globin exons. This was accomplished by making transgenic mice that contained single copies of the wild-type or modified YAC.6162
Imprinted regions are also good candidates for study using large transgenes.
Wutz and colleagues used YAC transgenes to examine the role of chromosome location and the role of an intronic CpG island in the imprinted expression of Igf2r, a maternally expressed gene.63 They were able to show that Igf2r imprinting is not dependent on chromosome location, but that it is completely dependent on the intronic CpG island, such that deletion of the CpG island causes derepression of the paternal Igf2r allele. Regulation of another well understood imprinted region, the H19-Igf2 locus, has also been studied using transgenic mice. The Tilghman lab were able to show that multiplecopy insertions of an H19 transgene could exhibit parent-of-origin dependent expression patterns as well as proper DNA methylation patterns, but these patterns were not maintained when the same transgene was present in single-copy. 65 The authors reasoned that the transgene constructs lacked sufficient sequence information to confer proper imprinted expression. Kaffer et al. subsequently identified a 137 kb bacterial artificial chromosome (BAC) that exhibited the proper pattern of imprinted gene expression for the H19-Igf2 locus in single-copy transgenic lines. Hark et al. also used the smaller transgenes that were not imprinted in single copy to test a model of gene regulation for the Igf2 gene.66 The authors used a reporter construct that contained a portion of the imprinted control region (ICR) followed by a polymorphic copy of the H19 gene and its enhancers derived from Mus musculus spretus, another intact ICR, and a luciferase gene. In subsequent transgenic lines a portion of the ICR that acted as a putative barrier element, separating H19 from its enhancers, was deleted. These lines showed higher expression of luciferase, indicating that the barrier element acted as hypothesized."9
Single copy transgenes could be used to elucidate some of the elements necessary to achieve proper imprinted expression in the PWS region. Identification of a transgene that is capable of demonstrating correct imprinted expression in single copy would elucidate the minimal genetic information that is sufficient for imprinting in this region. This transgene could then be used to make various permutations and delineate whether individual components are necessary for either paternal expression or maternal repression. Furthermore, if an efficient method for modifying the transgene were developed, a latent mark could be applied to the transgene that would make it easy to distinguish between transgenic and endogenous expression. The system could then be used to easily introduce new changes into the transgene.
Previously, our lab produced seven lines of transgenic mice carrying either a
human or a mouse P1 clone. The human P1 clone was chosen because it was known to include the human Angelman syndrome smallest region of deletion overlap (AS-SRO) as well as the Prader-Willi syndrome smallest region of deletion overlap (PWS-SRO). Furthermore, it was polymorphic to the homologous mouse locus that is syntenic to 15ql 1-q13.7" The expressed polymorphisms between the human P1 clone and the mouse locus would allow us to easily distinguish expression originating from the transgene from that coming from the endogenous locus. This human P1 clone was never correctly imprinted in 5 transgenic lines produced from it. The human P1 clone was expressed when it was inherited from both father and mother, suggesting that while the human PWS-IC positive element is functional in mouse, the human AS-IC element, which is predicted to be a negative regulatory element, does not behave as one in mouse. The mouse P1 clone, on the other hand, contained the Snrpn gene as well as the functionally
defined PWS-IC, but since the AS-IC has not been functionally or physically identified in mouse, we could not be certain whether or not it contained the AS-IC. Two lines were created from the murine P1 clone. One was multi-copy and showed correct imprinted expression, while the other was single copy and was expressed regardless of the parentof-origin. It was hypothesized that the mouse P1 clone did not contain the entire AS-IC, indicating that more genetic information, specifically 5' sequence, is necessary to imprint a transgene in single copy.
Two murine bacterial artificial chromosome (BAC) clones were identified that were capable of being expressed in the appropriate parent-of-origin dependent manner. These BAC clones specify a minimal region that is sufficient to confer correct imprinted expression at an ectopic location in the genome.
Materials and Methods
Screening the BAC Libraries
The Research Genetics (Huntsville, AL) BAC library was initially screened for BAC clones. This library was made from the 129/Sv strain of mouse. The library consisted of 9 filters, each carrying 27,648 unique clones with an average insert size of 130 kb spotted in duplicate onto 9 different membranes. A PCR-generated probe that included the first exon of Snrpn and a 2.2 kb EcoRI-EcoRV restriction fragment that was
5 kb 3' of Snrpn was used to screen the library.
Prior to hybridization, BAC membranes were washed with 1500 mL of 6X SSC and 0.1% SDS at room temperature for 15 minutes. The membranes were then rinsed twice with 1500 ml of 6X SSC for 15 minutes at room temperature, and hybridized according to the Research Genetics protocol. Briefly, the membranes were prehybridized
in roller bottles with 3 membranes per bottle, each separated by one sheet of Flow Mesh (Diversfied Biotech, Boston, MA), in 120 ml of HyperHyb (Research Genetics, Huntsville, AL) per bottle at 650 C for 20 minutes. At least 10' counts of each probe were boiled for 5 min, snap cooled on ice, and then added to 3 ml] of HyperHyb that was pre-warmed to 65' C. One ml of hybridization solution was added to each of the roller bottles, and the membranes were allowed to hybridize for 2 hours at 650 C.
The membranes were washed in the roller bottles three times for 15 minutes at
650 C with 30 nilIX SSC and 0. 1% SDS. They were then removed from the bottles and washed twice more at 650 C with 1000m1l of IX SSC and 0. 1% SDS for 15 minutes. Finally, the membranes were rinsed with IX SSC at room temperature. The membranes were then wrapped in cellophane and exposed to film (XAR, Kodak) overnight at -~80' C. Positive clones were identified following the manufacturer's instructions, and the following clones were ordered: 181J3, 573138, 518H22, and 397F16.
The Roswell Park Cancer Institute (RPCI) RPCI-23 BAC library was also
screened for potential imprinted transgenes using two probes; the 2.2 kb Eco RI- EcoRV fragment lying 5 kb 3' of Snrpn and a 1 kb SacI-NotI fragment that lies 30 kb 5' of Snrpn. This library is made from the C57BL/6J strain of mouse, with an average insert size of 200kb and is the template library for the NIH mouse genome sequencing efforts. The RPCI-23 library consists of 10 filters, each representing 18,432 independent clones that have been spotted in duplicate. To screen this library, the filters were first rinsed for 10 minutes at room temperature in 6X SSC. The BAC library filters were placed into roller bottles, five membranes per bottle, rolled with intermittently with sheets of Flow Mesh. The filters were pre-hybridized for 2 hours at 650 C with 200 ml of Church buffer
(1 mM EDTA; 0.5M NaHPO4, pH 7.2; 7% SDS; and 1% BSA). Approximately half of the pre-hybridization solution was poured off. The boiled, snap cooled probes (at least 107 counts, each) were then added, and the membranes were hybridized overnight at 650C.
The next morning, the membranes were first washed 5 times for 10 minutes each in the bottle at 650 C with 200 ml per tube of Wash I (1 mM EDTA; 40 mM NaHPO4, pH
7.2; 5% SDS; and 0.5% BSA). The filters were then removed from the bottles and washed 3 times for 15 minutes each at 650 C with Wash II (1 mM EDTA; 40 mM NaHPO4, pH 7.2; and 1% SDS). The membranes were wrapped in cellophane and exposed to film overnight and for 3 hours each at -80 C. Positive clones were identified following the manufacturer's instruction and ordered from RPCI. The following BAC clones were obtained from this screen: 73N19, 78014, 148C8, 160G24, 227M21, 284G21, 359M8, 365C3, 382D19, 438B21, 396A12, 421K3, 9715, 215A9, 264P24, 276M11, 425D18, 380J10, 391B10, and 437G6. Probe Preparation
Probes were labeled using the Prime-It II kit (Stratagene, La Jolla, CA), with
minor modifications. Probe DNA (200 ng) was boiled for 5 minutes in the presence of random hexamers. The DNA-hexamer mix was cooled on ice, and 5X reaction buffer was added along with 25 [.Ci of ap32dCTP and 1U of exo klenow. The labeling reaction was placed at 370 C for 15 minutes. After labeling was complete, the probe was purified using the nucleotide removal kit (QIAGEN) and eluted in 200 Rl of water. Purified probes were boiled for 5 minutes and snap cooled on ice prior to use in hybridization.
Verification of BAC Clones
Single BAC colonies were tested to verify that the clone represented a BAC insert derived from the Snrpn locus. Four single colonies from each BAC were streaked in duplicate onto Luria broth plates that were supplemented with chloramphenicol (LB-CA). The plates were overlayed with a nylon filter and grown overnight at 370 C. The next morning, the bacteria were lysed on the filters as follows: first, the filters were soaked in alkali solution (1.5 M NaCl and 0.5 N NaOH) for 4 minutes, followed by neutralizing solution (1.5M NaCl and 1M Tris, pH 7.4) for 4 minutes, and finally rinsed in 6X SSC for 1 minute. The filters were then pre-hybridized in Church buffer for one hour at 65' C, and then one plate was hybridized with the 5' Snrpn probe, while the other was hybridized with the 3' Snrpn probe. Upon washing and laying the filters to film, the results were used to determine whether or not the clone was derived from the Snrpn locus and whether or not the clone encompassed the 5' or 3' flanking regions of Snrpn. End Sequencing BAC Clones
The BAC clones were end sequenced using the big dye termination reaction to obtain sequences unique to the mouse genomic DNA insert. Reactions were performed on 1 mg of BAC DNA that had been sheared by passing it five times through a 25 gauge needle using a modified version of the manufacturer's instructions, with 12RI of sequencing buffer, 4 [dl of big dye, 16 pmol of primer, and 2.5mM MgC12. The reactions were subjected to an initial 5 minute cycle at 950 C, followed by 30 cycles of 950 C for 30 seconds, 500 C for 10 seconds, and 60' C for 4 minutes. The completed reactions were purified from free nucleotides using Performa DTR Gel Filtration Columns (Edge
Biosystems), dried, and sequenced by the University of Florida Center for Mammalian Genetics Sequencing Core.
Pulse-Field Gel Electrophoresis
Pulse-field gel electorphoresis was carried out to create a rough restriction map and tiling pathway for the BAC clones. BAC DNAs digested with the appropriate enzymes were electrophoresed on a 1% genetic technology grade agarose gel (Nusieve, FMC) made with 1X TBE buffer. Samples were mixed with lOX ficoll loading dye and ran at 200V with a switch time of 1-12 seconds for 12.5 hours. This provided resolution of bands from 6-200 kb. The gel was stained with a dilute ethidium bromide solution, photographed, and blotted as described below. Southern Blot
Agarose gels carrying 5 tg of digested genomic DNA or I ptg of digested BAC DNA were photographed and UV nicked for 5 minutes. The gel was then soaked in alkali solution for 45 minutes, followed by soaking in neutralizing solution for 90 minutes. The gel was then transferred to Hybond nylon membrane (Amersham) in lOX SSC overnight. The membrane was then rinsed in 2X SSC and baked at 800 C for 2 hours. Hybridization was carried out according to the method of Church and Gilbert. Membranes were prehybridized in 20 ml of Church buffer for 1 hour at 65 C. The prehybridization solution was poured off and replaced with 5 ml of fresh solution and the boiled and cooled probe. Hybridization continued overnight at 650 C. Washing was carried out three times in 2XSSC and 0.1% SDS at 65' C for 15 minutes each.
Transgenic Mice Production
For production of transgenic mice, BAC DNA was prepared by large scale
plasmid preparation, using the alkaline lysis method. First, log phase cultures of BAC culture grown in LB-CA were centrifuged for 5 minutes at 6,000 rpm to pellet the cells. The supernatant was poured off and the pellet was resuspended in GTE solution (50 mM glucose; 25mM Tris, pH 8.0; and 10 mM EDTA) plus lysozyme. The suspension was then lysed in 0.2 N NaOH and 1% SDS. After addition of the lysis solution, the lysate was neutralized in 0.5 M potassium acetate, and poured through cotton gauze into Oak Ridge tubes. BAC DNA was precipitated with ethanol, and the pellet was resuspended in
4 ml TE with 10 pg/ml RNAse A, and allowed to incubate at 370 C for 1 hour. Cesium chloride and ethidium bromide were added to the DNA solution, and the solution was loaded into a heat sealable tube. The tube was sealed and centrifuged at 58,000 rpm for 18 hours. BAC supercoils were viewed with a UV light, and both nicked and supercoiled bands were pulled. The DNA was extracted with tert-butanol until the ethidium bromide was completely depleted from the BAC DNA. The DNA was then dialyzed twice against
1 L TE buffer, using dialysis tubing for 6 hours each. The concentration of the BAC DNA was measured by spectrophotometer, and then diluted to 2.5 ng/4dl for microinjection.
Transgenic mice were made by pronuclear injection into oocytes that were
obtained from superovulated, fertilized, FVB/NJ female mice. The fertilized, injected oocytes were then implanted into (B6D2) F, females that were made pseudopregnant by mating them with vasectomized (B6D2) F, males.
At 3 weeks of age approximately 2mm of tail was clipped from each mouse, and the ear was punched for identification. Genomic DNA was prepared from the tail piece by incubation in tail lysis buffer (100 mM Tris, pH 8.5; 5 mM EDTA; 2% SDS; and 200mM NaCI) with 100 ng/gl proteinase K overnight at 550 C. The tail lysate was extracted with phenol: chloroform: isoamyl alcohol (25:24:1) and then precipitated with ethanol. The tail DNA was screened for the presence of the transgene by PCR using both end sequences from the BAC.
APWS-IC mice maintained on the C57BL/6J background were developed by
Yang et al. and constituted a 35 kb deletion encompassing 16 kb of sequence 5' of Snrpn and 19 kb 3' of Snrpn 22.55. Lines 215A, 215B, 215C, 425A, and 425B were all transgenic lines produced as described above and maintained on the FVB/NJ strain background. Transgenic line 1707 was previously described and was originally maintained on the DBA/2J strain background 7o. For the purposes of this study, this line was backcrossed for 10 generations onto the FVB/NJ strain background.
A YAC that contained 320 kb of genetic information spanning the Snrpn area (Fig 3-1) was previously identified. Pulse-field gel electrophoresis was used in conjunction with rare cutting enzyme digests to confirm that the YAC was indeed intact and contained the genetic information that we anticipated. Several unsuccessful attempts were then made to subclone the regions that comprised the junction between the YAC
vector arms and the mouse DNA insert. Despite the failure to find these junction fragments, an attempt was made to make transgenic mice using the YAC, but it was not successful. Ultimately, the attention was focused on BACs, since candidate clones for imprinted transgenes had been identified with our screen. It was surmised that the YAC was not ideal for use, since a YAC of that size is difficult to characterize and manipulate. 129/Sv BAC Library Screen
BAC clones were the best suited for transgenic studies, since they are usually larger than the P1 clones that were previously shown not to contain sufficient sequence information to be imprinted in single copy. While they are larger than P1 clones, BAC clones are smaller than YACs and are maintained as supercoils, making them more resistant to shearing. They also have the advantage over YACs of being propagated in E. coli rather than S. cerevisiae, which makes their isolation and handling much easier. Most importantly, BACs are propagated in RecA cells, and are less likely to suffer rearrangement during manipulation and propagation than YACs. Traditionally, the advantage of using YACs as opposed to BACs was the ability to make modifications by exploiting the efficient homologous recombination in S. cerevisiae. However, early in this project a preliminary method of BAC modification using homologous recombination was published, making it possible to make similar changes to BAC clones ".
The first BAC library that we screened was derived from the 129/SvEv strain of mouse, and had an average insert size of 130 kb. Using a mixture of the Snrpn exon 1 probe and a 2.2 kb EcoRI-EcoRV that lies 5 kb 3' of Snrpn as probes in this screen, four BACs were isolated: 181J3, 573B8, 518H22, and 397F16. These BAC clones were each streaked to LB-CA plates to obtain single colonies. Four single colonies representing
each BAC clone were grown in 3 ml overnight cultures and DNA was isolated from each culture. The BAC DNA was cut with EcoRI, and a Southern blot was performed. The blot was first probed with a 1 kb SacI-NotI fragment that yielded a large fragment about 20 kb 5' of Snrpn, and later stripped and re-probed with the 3' Snrpn probe. Two BACs, 397F16 and 518H22 spanned 5' Snrpn, but terminated short of the 3' Snrpn probe, while the other two BACs, 573B8 and 181J3, conversely hybridized with the 3' probe, but appeared to lack 5' sequences (Fig. 3-1). All four BACs were end sequenced. Next, the BAC clones were each cut with Sal, NotI, SacI, or a combination of the enzymes and subjected to pulse-field gel electorphoresis and Southern blot. Using the 5' and 3' probes, probes intergenic to the Snrpn gene, and probes generated using the end sequences from each BAC, the approximate boundaries as well as a rough restriction map for each BAC clone was determined. Following this analysis, only two BACs, 397F16 and 573B8 remained interesting as potential imprinted transgene candidates. The 397F16 clone contained the most 5' sequence of the BAC clones, but it terminated before the polyadenylation signal of the Snrpn gene. This was a problem because it could cause an expressed Snrpn transcript to be targeted for degradation. The other BAC of interest, 573B8 contained the entire Snrpn gene, but did not contain enough 5' sequence. Since neither BAC was ideal for use as an imprinted transgene, a plan was made to join these two BACs using the recently developed BAC modification scheme to produce the additional 5' sequence as well as the complete Snrpn transcript. RecA Mediated Homologous Recombination
The first method attempted was analogous to the yeast two-step homologous recombination procedure, as described by the Nat Heintz group.71 It revolved around
supplying the bacterial RecA gene, which was missing in the BAC host strain, on a selectable episome.7" The bacterial RecA gene was provided on a plasmid with a temperature sensitive origin of replication. The plasmid could be positively selected for using tetracycline resistance. The plasmid could also be subjected to negative selection by plating on plates containing fusaric acid, since fusaric acid is toxic to cells that produce the tetracycline resistance gene. The idea behind this recombination scheme was that one could subclone the desired modification or sequence change flanked by 5' and 3' homology arms of at least 500 bp, into this temperature sensitive recombination vector. When transformed into the BAC strain, positive selection for the modification construct would identify colonies in which insertion of the construct had occurred. Then, negative selection with fusaric acid would cause resolution of the integrant into one of two possible conformations. One possible conformation is that in which the desired change has been made, and the other conformation recovers the unmodified BAC.
The scheme to join the two pertinent BACs involved using homologous recombination to insert a neomycin resistance cassette flanked by loxP sites approximately 30 kb upstream of the Snrpn gene on the 397F16 BAC. A 35 kb Sall fragment from the modified 397F16 BAC could then be excised and recombined into the 573B8 BAC using one arm of homology anchored in the BAC vector and the other end anchored near the first exon of Snrpn. The neomycin resistance gene could then be deleted using cre recombinase, leaving a single loxP site. The resulting BAC would have 35 kb sequence 5' of Snrpn and 55 kb 3' of Snrpn, and would be a good candidate for an imprinted transgene (Fig 3-2). While this method has been successful in other labs, it was not so in this case due to the difficulty of subcloning the initial recombination
construct into a low-copy number plasmid that is temperature-sensitive and already quite large (11 kb).
RecET Mediated Homologous Recombination
The second method, pioneered by Francis Stewart's group, was based on the RecET system in lambda phage. 72 This system is reputedly much more efficient than that of RecA and requires smaller homology arms. The RecET mediated recombination can be carried out using only large oligos and the RecET plasmid electroporated into the BAC containing bacterial strain. The RecE and RecT gene are expressed from the pBAD promoter, making its expression arabinose inducible. This method never worked as the wrong plasmids were sent. Following this attempt, BAC modification was postponed to begin the search for a more suitable BAC transgene. C57BL/6J BAC Library Screen
The RPCI-23 BAC library was screened since our attempts at BAC modification were not working well. The RPCI library had three major advantages to the Research Genetics library. First, the average insert size for this library is 200 kb, increasing the chances of obtaining a BAC that is suitable for use as an imprinted transgene without additional modification. Secondly, this library is the template for the mouse genome sequencing effort, and so complete BAC sequences could possibly be posted as the project continued. Thirdly, if a complete BAC sequence was desired for a given BAC from this library, it could be queued for sequencing. Similarly to the strategy with the 129/Sv library, the filters were probed using both the 5' and the 3' Snrpn probes. Several Snrpn transgenes were identified from this library. They are listed as follows:
73N19, 78014, 148C8, 160G24, 227M21, 284G21, 359M8, 365C3, 382D19, 438B21, 396A12, 421K3, 9715, 215A9, 264P24, 276M11, 425D18, 380J10, 391B10, and 437G6.
After streaking for single colonies and verifying that the clones represented the
Snrpn region, intact BAC molecules were subjected to restriction digestion with a variety of rare cutting enzymes, including Sall, SacII, and NotI and resolved the resulting fragments by pulse field electrophoresis. The gels were transferred to nylon membranes and probed with various probes in the Snrpn area to produce a rough restriction map of each BAC and to determine whether it contained enough genetic information to serve as a putative imprinted transgene. BAC end sequences for many of these clones were readily available on the National Center for Biotechnology Informatics (NCBI) website, and so primers could be ordered and used to create probes for the end of each BAC. The following probes were used in determining BAC tiling pathways and restriction map: a 1 kb SacI-NotI fragment 5' of Snrpn, a 2.2kb EcoRI-EcoRV fragment that is 5 kb 3' of Snrpn, an oligo that localized to an interstitial A particle (IAP) that is located in the 5' end of the Ipw gene and more than 100 kb 3' of Snrpn, and an oligo that is complimentary to a small nucleolar RNA (snoRNA) that was known to exist approximately 65 kb 3' of Snrpn. Using these probes, BACs 215A9, 425D18, and 380J10 were determined to be suitable for further analysis. BAC 215A9 was shown to extend 120 kb 5' of Snrpn and 20 kb 3' of Snrpn, BAC 425D18 spanned a region that was first predicted to cover sequence 40 kb 5' of Snrpn through 65 kb 3' of Snrpn, but was later discovered to extend 90 kb 5' of Snrpn. Finally, BAC 380J10 was shown to contain only 8 kb 5' of Snrpn, but extended nearly 140 kb 3' of Snrpn, into the IAP element and the start of the Ipw gene (Fig 3-3).
Lambda Red Mediated Homologous Recombination
Although intact BAC molecules that seemed well suited to be imprinted without further modification were available, it was desirable to create a subtle mutation in the BAC that would allow transgenic Snrpn expression to be distinguished from the endogenous message. A third method of BAC modification that was developed by Daiguan Yu and Don Court at the National Cancer Institute (NCI) was pursued. This method employed a defective lambda prophage in which the genes allowing entry into the lytic phase of the phage lifecycle were removed, leaving only the minimal elements required to express the red exo, beta, and gam genes under a temperature-sensitive promoter.73 The first version of this system was a modified strain, in which the defective lambda prophage was integrated into the bacterial genome. Unfortunately, this version had some remaining phage genes present that would cause lysis of the bacterial cell when induced, since these genes were also initiated from the temperature-sensitive promoter. The other major problem was that this phage system was developed in an E. coli host strain that had the opposite dam methylation status to the BAC host strain. A BAC clone isolated from a dam methylation-positive strain was therefore methylated and could not transform into the strain bearing the lambda prophage. As a result, none of the aforementioned BAC clones were successfully transformed into this strain. Two improved versions of the same system were later produced that were based on the BAC host strain of E.coli, DH1OB. One version featured the defective lambda prophage as an integrant into the DH1OB genome, while the other harbored a defective prophage that was maintained as a stable episome in the bacterial cell. While the transgenic mice in
this study were derived unsing unmodified BAC clones, the BACs have since been modified using the lambda Red system and will be the focus of future work. Production Of Transgenic Mice
While continuing with efforts to modify the BACs, it seemed prudent to begin producing transgenic mice carrying unmodified Snrpn BACs. BAC DNA was prepared by large-scale plasmid preparation, followed by cesium chloride banding of the resulting DNA. Both nicked and supercoiled bands were pulled and used for production of transgenic mice. It was important to have extremely fresh BAC, as the larger RPCI BACs degraded rapidly after isolation, as evidenced by low yield of supercoils and a diffuse appearance on pulse-field gels. The BAC DNA was dialyzed against TE buffer and diluted for injection with water to a concentration of 2.5 ng/gl. Transgenic mice were made by pronuclear injection into oocytes derived from superovulated, fertilized, FVB/NJ female mice. The injected oocytes were then implanted into pseudopregnant (B6D2) F, mice. When the resulting pups reached 3 weeks of age, they were tail clipped to obtain DNA and earpunched for identification. Genomic DNA was prepared from each tail piece and was screened for the presence of the transgene by PCR using both end sequences from the corresponding BAC. Three transgenic founder animals were produced with the 215A9 transgene, two lines were made with the 425D 18 transgene, and five lines carried the 380J10 transgene.
The transgenic founders were mated with wild-type FVB/NJ females and the resulting litters were genotyped at 3 weeks of age to make sure that the transgene had integrated into the germline of the founders (i.e. that the founders were not chimeric), and to establish additional transgenic animals to secure the survival of the transgenic line.
One transgenic mouse derived from the 380J10 BAC did not show germline transmission to subsequent generations, and most likely represented a chimeric mouse in which the transgene had integrated into the genome sometime after the first cleavage following fertilization. Once the remaining lines were established, the individual lines were characterized for intactness of the transgene and copy number (Fig. 3-4). Characterization Of Transgenic Lines
Characterization of the transgenic lines was necessary to determine the number of BAC molecules that were integrated into the transgene locus as well as the integrity of the molecules that had integrated. DNA used to make transgenic mice tends to integrate into the mouse genome in tandem arrays. For the purposes of our study, it was imperative that we obtain some single copy transgenic lines, since we were adhering to the most stringent determination of the minimum amount of sequence information that was sufficient to confer proper imprinted expression. It was also important that the transgene be intact, since proper imprinted expression may be a result of the relative location of the pertinent elements. Furthermore, the BAC molecules had to be linear prior to integration into the genome, and since we injected supercoils and nicked BAC molecules, the location of the site of breakage was random. This increased the chances of deriving a non-intact transgenic line.
Copy number was determined by preparing genomic DNA from mouse tails
harboring the transgene, performing Southern blot analysis on DNA that was cut with a specific restriction enzyme, and comparing bands from two different probes that would yield bands of similar size upon the same restriction digestion. One probe was located on the transgene, while the other was not. The intensities of the bands were compared and
the copy number was estimated as either single copy or multi-copy. One of the 215A9 lines, one of the 425D18 lines, and one of the 380J10 lines contained multiple copies of the transgene, leaving us with two 215A9 transgenic lines, one 425D18 transgenic line, and three 380J 10 transgenic lines that represented single copy transgene integrations (Fig. 3-4).
Similarly, rearrangements were determined by creating a panel of DNAs from the different transgenic lines digested by a variety of restriction enzymes. These panels were used to produce Southern blots, which were probed with several probes from the region (Fig 3-5). If a probe illuminated a band that was of a different size than that of a wildtype, non-transgenic mouse, then that transgene was considered rearranged. One 215A9 line and one 380J10 line appeared to be rearranged by this analysis. However, due to the relatively few non-repetitive probes in the sequence surrounding the Snrpn locus, we may not have detected other rearrangements.
The Marked Endogenous Snrpn Locus
The inability to mark the BACs prior to their injection into mouse oocytes forced the use of a previously engineered Snrpn deletion mouse strain (ASmN) 22. This mouse strain carries a deletion of exons 4-7 of the Snrpn gene, that results in almost complete absence of the Snrpn gene, and the low-level production of a larger fusion transcript that includes the first exons of Snrpn and the neomycin resistance gene. Using this allele, the endogenous Snrpn locus could be marked, alleviating the absolute requirement for a marked transgene (Fig 3-6).
The APWS-IC mouse strain was also used.22 This strain has a neomycin
resistance cassette inserted in place of exons 1-5 of Snrpn. When paternally inherited,
this deletion prevents expression from the Snrpn locus, allowing us to measure Snrpn expression that originates from the transgene (Fig 3-6). Imprinted Expression From Transgenes
Since the transgenic copy of Snrpn was not subtly marked to be easily
distinguishable from the endogenous allele, the ability of the transgene to show proper imprinted expression had to be tested by a complex breeding scheme using a marked endogenous allele. This breeding scheme is shown in Figure 3-7. First, the expression upon maternal inheritance was examined. Females who harbored a transgene were mated with a male mouse that carried the SmN deletion allele. The resulting pups were genotyped for the presence of the transgene, as well as for the presence of the deletion allele. Pups with both the transgene and deletion allele were sacrificed and their brains were used to make RNA. The RNA was analyzed by northern analysis for the expression of Snrpn. This is possible because the transgenic female mice transmit a transgene, which may or may not be expressed, depending on its epigenetic state, as well as a wildtype copy of the endogenous Snrpn locus that is not expressed since it is inherited from a female. The male mouse, on the other hand transmits the ASmN allele only, which is expressed as a larger, less abundant message. Any wild-type Snrpn message, would then come only from the transgene. Only the 380 transgenes were able to express Snrpn upon maternal inheritance, indicating that this transgene did not contain the sequence information that was required for silencing of the Snrpn gene upon maternal inheritance (Fig 3-8).
Next, expression upon paternal inheritance was tested. In this case, male mice from the above described mating who had inherited both the transgene as well as the
ASmN allele were mated to females who were homozygous for the ASmN allele. The resulting pups were sacrificed and genotyped for the presence of both the transgene and the ASmN allele. The rationale behind this scheme is that the female in this case can only send a ASmN allele, which is silenced. In the presence of a known maternal ASmN allele, a paternally inherited wild-type allele can be distinguished from a paternally inherited ASmN allele. The male mouse transmits the transgene, which may or may not be expressed depending on its epigenetic state, and either a wild-type or ASmN allele. If the male transmits the wild-type allele, the resulting pups cannot be used for analysis, since the paternally inherited Snrpn locus is indistinguishable from the transgenic Snrpn locus. Alternatively, if the male transmits the ASmN allele, the endogenous allele that is expressed produces the longer, less abundant ASmN message (Fig 3-6).
Originally, the idea was to identify Snrpn mice that were transgenic and
homozygous for the ASmN allele, and examine expression in these pups. However, the presence of a large transgene that produced a wild-type Snrpn allele by Southern blot hindered identification of ASmN homozygotes. Instead, the mice were genotyped for the transgene, and RNA was made from the brains of the resulting transgenic mice. The RNA was then used to program RT-PCR reactions. PCR was performed on the RT product, using primers for both the ASmN and wild-type alleles. Pups that express both the ASmN and wild-type alleles inherit the ASmN allele and not the endogenous wildtype allele from their father (Snrpn and the ASmN alleles are only paternally expressed). Thus the wild-type allele must be expressed from the transgene. If the pup did not show expression from the ASmN allele, then it was believed to have inherited the wild-type endogenous allele from its father, and was uninformative as far as transgene expression is
concerned. From this assay, lines from each the two transgenes, 215A9 and 425D 18, were shown to have expression upon paternal inheritance. Thus both 215A9 and 425D18 are capable of showing correct imprinted expression in single copy transgenes (Fig 3-9).
These experiments were repeated using the APWS-IC deletion mice in place of
the ASmN deletion mice. To ascertain transgene expression upon maternal inheritance, a transgenic female mouse is mated with a APWS-IC male mouse (Fig 3-10A). The resulting pups were sacrificed and genotyped. Pups that inherited both the APWS-IC allele and the transgene were used to produce brain RNA, which was subjected to northern blot analysis and probed for Snrpn expression. Again, the females transmit the transgene as well as an epigenetically silenced maternal Snrpn allele, while APWS-IC males transmit a APWS-IC allele that completely lacks Snrpn expression. Any Snrpn message would have to originate from the transgenic copy of Snrpn. All five transgenic lines were subjected to expression analysis upon maternal inheritance, and again, only the 380J10 transgenic lines, with the exception of line 380C, showed Snrpn expression upon maternal inheritance, verifying that they did not contain sufficient sequence to confer epigenetic silencing to the maternally inherited transgene (Fig 3-10B).
To ascertain expression upon paternal inheritance, APWS-IC females were mated with transgenic males, and male pups with both the APWS-IC allele and the transgene were identified by genotyping. These pups survive since the APWS-IC allele is maternally inherited and therefore exists on the chromosome that is already epigenetically silenced. Male pups that have inherited both the APWS-IC allele as well as the transgenic allele were mated with wild-type females from the C57BL/6J strain to produce pups that harbor both the APWS-IC and the transgene (Fig. 3-1 IA). These pups
do not express endogenous Snrpn since their paternally inherited allele is the APWS-IC allele, which is silenced because the major promoter for the Snrpn gene is missing and because it is epigenetically silenced as a result of the imprinting center mutation. The only Snrpn expression would come from the transgenic Snrpn allele, and would occur only if the transgene were expressed upon paternal transmission. In this assay, the pups were sacrificed at birth and genotyped for the APWS-IC allele and the transgene, since pups that inherit the APWS-IC allele from their father suffer from the Prader-Willi phenotype and die shortly after birth. Pups that inherited the transgene as well as the APWS-IC allele were used to make brain RNA. This RNA was subjected to northern analysis, using a probe for the Snrpn message (Fig 3-11IB). Two transgenic lines-215B and 425A-were identified that expressed transgenic Snrpn upon paternal transmission. Since neither of these lines were expressed upon maternal transmission, they were determined to demonstrate correct imprinted expression.
Line 425A contains a single copy transgene that is intact. The transgene in this line is estimated to include the 22 kb Sail fragment that hosts the entire Snrpn gene, as well as 90 kb 5' of Snrpn and 65 kb 3' of Snrpn. This establishes the minimal genomic sequence that is sufficient to allow for correct imprinted expression in this region to be approximately 170 kb. Line 215B, on the other hand, is a single copy transgene that has undergone both 5' and 3' truncation upon integration into the mouse genome. The 5' truncation occurs at least 38 kb 5' of the Snrpn gene, while the 3' truncation occurs approximately 5 kb 3' of Snrpn. While this data suggests that the minimal sufficient region for correct imprinting is approximately 65 kb, the specific rearrangements that led to the truncation of this transgene are not known. The transgene in this particular line
may have also sustained additional mutations that influence its ability to be correctly imprinted.
The 85 kb PI clone that shows correct imprinted expression when present in
multiple copies, but not when present in single copy was later found to only be 65 kb in length. The discrepancy in length occurred because the P1 was isolated from the 129/01a mouse strain, which was found to have an insertion of a 20 kb VL-30 element into the region 5' of Snrpn. The parent-of-origin independent expression of this P1 clone, now known to be 65 kb in length, indicates that either the PI clone is missing important regulatory sequences that lie 5' or 3' of the Snrpn gene, or the P1I clone does not contain enough buffer sequence surrounding the necessary regulatory regions to establish a domain of imprinted expression. The P1 clone is expressed upon both paternal and maternal inheritance when present in single copy, indicating that the defect is most likely in the ability to be silenced in the maternal germline. Since the murine AS-IC has not been functionally or physically identified in mouse and is postulated to be the negative regulatory element required for the silencing of the positively acting PWS-IC (Fig 2-4), the most likely explanation for the absence of maternal silencing of the P1 transgene is that it does not contain the AS-IC regulatory sequences. Absense of the AS-IC would be predicted to prevent silencing of the positive element located on the transgene upon maternal inheritance. Since the P1 clone without the VL-30 insertion extends 23 kb 5' of Snrpn, we believe that the AS-IC is located further than 23 kb 5' of Snrpn. This data spurred the search for transgenes with additional sequence 5' to Snrpn, rather than seek additional buffer sequence. Whether it is indeed the absence of regulatory elements or
buffer sequence is a germane argument in this system, since we were seeking minimal sufficient sequence to achieve appropriate imprinting, rather than determining which specific elements are necessary. The data from the P1 clone suggested that the 65 kb sequence was not sufficient to confer appropriate imprinted expression, and a single-copy imprinted transgene is necessary to establish the minimal sequence that can accomplish correct parent-of-origin dependent expression. Once the minimal required sequence has been determined, modifications can be made to that sequence to delineate specific elements that are necessary for the establishment and maintenance of the imprint.
While initial attempts to make transgenic mice with the 320 kb YAC clone were unsuccessful, the YAC clone was used to establish the approximate distance between the 3' most exon of Snrpn and exons B and C of Ipw to be 100kb. However, the difficulties surrounding purification of a YAC of that size for transgenic injection precluded its convenient use as an imprinted transgene. Previously, the largest YACs that were successfully used in gene regulation studies involving transgenic mice were only 250 kb in length.
Screening the 129/Sv BAC library proved fruitful for mining additional sequence 5' of Snrpn, but did not provide an ideal BAC clone to use as an imprinted transgene. Two clones were obtained, 397F16 and 573H8 that together contained additional sequence 5' of Snrpn as well as a complete Snrpn gene. Unfortunately, the commercially available BAC libraries were of the 129/Sv or C57BL/6J strains, which made finding naturally expressed polymorphisms unlikely since both strains are from the domesticus subspecies of Mus musculus. The possibility of using a more polymorphic human Snrpn
transgene was explored and rejected earlier, since the human transgene is not silenced when present as a murine transgene regardless of copy number.70
Since the 129/Sv BACs were not ideal, the first strategy was to join the two BACs to create a larger BAC that was more suitable for transgene imprinting. The strategy was to place a neomycin resistance cassette into a 35kb Sall fragment derived from the 397F16 BAC that lies 5' of Snrpn, and then excise this fragment and ligate it into the 573H8 BAC near the site of exon 1 of Snrpn. At the time, only the RecA mediated homologous recombination method had been described. This method involved a twostep gene replacement strategy in which the RecA protein, absent in the BAC host strain, was provided from a pSV1-derived plasmid with a temperature-sensitive origin of replication ". Attempts to subclone homology arms into the RecA-containing plasmid failed, probably due to the large size of the plasmid. RecET mediated homologous recombination was also attempted. This method involved inducing RecE and RecT from an arabinose-inducible promoter. 72 This method was never fully explored because the wrong plasmids were sent.
In the absence of a reliable homologous recombination method for BACs, the
C57BLI6J BAC library that was available from the RPCI was screened. The BACs from this library were on average larger than those from the 129/Sv library. Additionally, the NCBI mouse genome sequencing effort was proceeding from this library, so the specific sequence in the region would eventually become available, as well as the addresses of additional clones in the region. Screening efforts for this library were very successful, as 20 clones were isolated with the Snrpn probes. From the 20 clones, two were chosen that contained the most detectable sequence upstream of Snrpn while still carrying the entire
gene. While neither of the two BACs were in the sequencing program, both BACs were present as T7 and Sp6 fingerprints (end sequences) in the NCBI database.
Pulse-field gel analysis initially indicated that BAC 425D 18 contained 40 kb of sequence 5' of Snrpn, while 215A12 contained nearly 120 kb 5' of Snrpn. These estimates are rough, since they are based on large pulse-field fragments. Subsequent analysis suggested that 425D 18 actually extended 80 kb 5' of Snrpn, and that the 40 kb Sally fragment that was originally thought to be the junction fragment between the mouse DNA insert and the BAC plasmid backbone was a doublet of two adjacent fragments.
Three lines of transgenic mice harboring BAC 215A9 and two lines harboring 425D18 were obtained from microinjection of mouse oocytes. These lines were named 215A through C and 425A and B. Among these five lines, lines 215B, 215C, and 425A were all suspected to be present in single copy. Lines 215A and 425B were estimated to be present in 2-4 copy arrays. The method of analysis of transgene copy number that was used is probably accurate in delineating single-copy transgenes, but not in determining the copy number in multi-copy lines. Additionally, all transgenic lines except 215B appeared to be intact.
Transgenes were first tested for correct imprinted expression using the ASmN strain of mouse. Expression upon maternal inheritance was readily assayed using the breeding scheme in Figure 3-7. Lines 215 A-C and lines 425 A and B were not expressed upon maternal transmission, suggesting that they could be silenced in the maternal germline (Fig 3-8). Lines 380A,B, and D, however were not correctly silenced upon maternal inheritance (Fig. 3-8) and were thought to be missing the AS-IC. This would result in the failure to silence the positively acting PWS-IC and lead to expression upon
maternal transmission. Paternal transmission was not conclusively tested using the ASmN breeding scheme (Fig 3-7). Mice homozygous for the ASmN allele and heterozygous for the transgene could not be distinguished from the ASmN heterozygotes by southern blot. Instead, RNA was made from the brains of transgenic mice and subjected to RT-PCR. If the ASmN allele was present, it was assumed that the paternally inherited allele was the ASmN allele and not the wild-type allele, leaving any wild-type Snrpn expression to be from the paternally inherited transgene. According to this assay, lines 215C, 425A, and 425B were all expressed upon paternal inheritance (Fig 3-9). This breeding scheme was not further pursued, since it required the production of RNA and subsequent RT-PCR from every transgenic mouse. It was also not ideal that the genotype of the mice was never certain with respect to the Snrpn versus the ASmN allele. Furthermore, the potential for leaky expression from the maternally inherited allele complicated an assay that required RT-PCR.
The imprinting assays were again tested using the APWS-IC strain and the
breeding schemes shown in Figures 3-10A and 3-1 IA. Snrpn expression upon maternal inheritance was tested in a similar manner and produced the same results as previously (Fig. 3-10B). Snrpn expression upon paternal inheritance of transgenes proved to be much simpler using the APWS-IC strain. Mice inheriting the APWS-IC allele were easily genotyped by PCR that was unimpeded by the transgene. These mice could also be identified shortly after birth by their failure to thrive phenotype. Expression from the transgene in pups with a paternally inherited transgene and a paternally inherited APWSIC allele could then be assayed for Snrpn expression by northern blot analysis, which does not detect the leaky maternal expression. By northern analysis, any Snrpn
expression originated from the transgene. Using the APWS-IC strain, lines 425A and 215B exhibited Snrpn expression upon paternal transmission of the transgene (Fig. 311B). Since both of these lines are also silenced upon maternal transmission, they are both correctly imprinted.
Line 215C was first shown to exhibit Snrpn expression upon paternal inheritance using the ASmN strain, but not with the APWS-IC strain. The most likely explanation for this is that the transgene is indeed expressed upon paternal inheritance, but at a low-level that is detectable by RT-PCR, but not by northern analysis. It was precisely this possibility that led to the use of the APWS-IC strain to test transgene imprinting. It was desirable to maintain a high level of stringency in considering a transgene correctly imprinted, the possibility that low-level expression could come from a correctly imprinted repressed allele existed. Since the expression of Snrpn from line 215C could possibly represent leaky expression of a repressed allele and not appropriate imprinting, line 215C was not considered to be correctly imprinted.
Lines 425A and 215B demonstrated appropriate imprinted expression, indicating that they contained sufficient sequence to confer proper imprinting. This was an important finding, since the AS-IC has to date not been defined functionally or physically in mouse. According to the current model for imprinting in the PWS/AS region, 7 a correctly imprinted transgene must contain the minimal sequence to harbor the positively acting PWS-IC element as well as the negatively acting AS-IC element. The identification of a correctly imprinted transgene indicates that the AS-IC can be functionally defined in mouse, since in this instance the silencing function occurs on a transgene that is integrated into a random genomic location. The transgene then serves to
define an approximate location of the AS-IC, and combined transgene data suggests that it lies upstream of the PWS-IC, as it does in humans. The 425A line definitively narrows this boundary to a region within 90 kb of the Snrpn gene. The 215B transgenic line has undergone truncations at both the 5' and 3' end of the BAC, but is still imprinted. These truncations narrow the probable location of the AS-IC to a region 38 kb 5' of Snrpn, but since the nature of the transgene rearrangements are not well understood, the use of this data to suggest the location of the AS-IC must be done with caution.
The ability to use the lambda Red system of homologous recombination in BACs will greatly simplify the study of transgene imprinting. A latent mark can now be placed into the Snrpn locus, and its expression upon both maternal and paternal inheritance can be ascertained in a single generation. This same system can be used to make deletions in the most 5' end of BAC 425D18 to better identify the AS-IC. Then, specific elements can be deleted, replaced, or rearranged to test different models of imprinting mechanisms in the PWS/AS region.
genelA 5' probe 3' probe
23 kb 1 30 kb
- 967 PI (murine)
Figure 3-1. Restriction map of YAC clone and 129/Sv BAC library clones. YAC and BACs were subjected to restriction digestion with rare cutting enzymes and analyzed by pulse-field gel electrophoresis and southern analysis. Thin vertical lines represent Sail sites. Black boxes represent the Snrpn gene or Ipw IAP element. The 5' probe is a 1 kb SacI-NotI fragment, while the 3' probe is a 2.2 kb EcoRI-EcoRV fragment.
397F16 V V
Homologous recombination 397-neo
S Ligation into Sall digested BAC
35 kb 55 kb
Figure 3-2. Strategy to combine BACs 396F16 and 573B8. The strategy for joining two BACs to create one of an ideal size is diagrammed. Vertical hash marks indicate Sall sites. Dashed line represents BAC vector.
gene 5' probe 3' probe
I I -I
I I I kb 967P1(muri) 23kb 2 30kb
I I I I 8kb 140 kb
170 kb 20 kb
I I I I I I
Figure 3-3. Restiction map of C57BL/6 BAC clones. BAC clones were subjected to restriction digestion and analyzed by pulse-field electrophoresis and southern analysis. The 5' probe is a I kb SacI-NotI fragment, while the 3' probe is a 2.2 kb EcoRk-EcoRV fragment. Vertical hash marks represent Sall sites. Not all clones are diagrammed.
PI-967 1 1707 1 copy, intact
P1-9671737 2 copies, intact
A 2-4 copies, intact 215 00 B I copy, rearranged
C I copy, intact
425 10 A 1 copy, intact
B 2-4 copies, intact
A 2-4 copies, intact 380 ,B I copy, intact
380 00 C 1 copy, truncated
D 1 copy, intact
Figure 3-4. Transgenic lines investigated for imprinted expression. Copy number and intactness is indicated according to available probes.
A B C D
1 2 3 45 67 8 910
I I IIII II
Figure 3-5. Location of probes around the Snrpn locus. Probe A represents A 0.8 kb PCR fragment 38 kb 5' of Snrpn. Probe B is a 1 kb SacI-NotI fragment that is 30 kb 5' of Snrpn. Probe C is a probe for exon 1 of Snrpn. Probe D is a 2.2 kb EcoRI-EcoRV fragment 5 kb 3' of Snrpn.
1 2 3 45 6 7 8 910
1 2 3 4 8 9 10
Neo I I
Figure 3-6. Snrpn locus, ASmN allele, and APWS-IC allele structures. The ASmN allele represents a deletion of exons 5-7 of the Snrpn gene. The Neomycin resistance cassete (Neo) produces a larger fusion transcript by Northern blot, thus marking that allele. The APWS-IC allele represents a 35 kb deletion that includes 16 kb of sequence 5' of Snrpn as well as exons 1-7 of the Snrpn gene.
Tg X P
Paternal transmission ko X Tg ko
Figure 3-7. Breeding scheme for testing expression upon maternal and paternal transmission of transgenes. Tg represents any transgene, while ko represents the ASmN allele. Maternal and paternal alleles are indicated where relevant. Only offspring of the desired genotype are indicated.
Figure 3-8. Expression of Snrpn from transgenes upon maternal transmission.
The upper band results from a paternally inherited ASmN allele that creates a fusion transcript that is larger than the normal Snrpn message. Snrpn is only expressed from transgenes upon maternal inheritance in lines 380A, 380B, and 380D.
ASmN allele Neo
Endogenous allele/ transgene
425A 425B 215C
1 2 1 2 1
Figure 3-9. Expression of Snrpn from transgenes upon paternal inheritance.
A. PCR primer locations for each allele. B. RT-PCR showing ASmN and
wild-type alleles. The presence of the ASmN allele indicates that the paternally inherited endogenous allele was the ASmN allele, and then
expression of the wild-type allele indicates expression of the transgene. If
the ASmN allele is not present, then the paternally inherited endogenous allele
is assumed to be a wild-type allele and this mouse is uninformative for transgene
Figure 3-10. Transgenic expression of Snrpn upon maternal transmission of transgenes. A. Breeding scheme used to assay transgenic expression using the APWS-IC strain. Tg represents any transgene, while ko represents the APWS-IC allele. Maternal and paternal alleles are indicated where relevant. Note that APWS-IC pups are missing all paternally expressed gene products in this region and succumb to neonatal lethality when the APWS-IC allele is paternally inherited. B. Northern blot showing Snrpn expression in transgenic pups inheriting a paternal APWS-IC allele and a maternal transgenic allele.
k X Tg +
Tg ko m
X Tg m
Tg p p
Figure 3-11. Transgenic expression of Snrpn upon paternal transmission
of transgenes. A. Breeding scheme used to assay transgenic expression using
the APWS-IC strain. Tg represents any transgene, while ko represents the
APWS-IC allele. Maternal and paternal alleles are indicated where relevant.
Note that APWS-IC pups are missing all paternally expressed gene products in this region and succumb to neonatal lethality when the APWS-IC allele is
paternally inherited. B. Northern blot showing that Snrpn is expressed upon
paternal expression in lines 425A and 215B. 380 lines are not shown
STRAIN-DEPENDENT DIFFERENCES IN PHENOTYPE Introduction
Angelman Syndrome (AS) and Prader-Willi Syndrome (PWS) are caused by
deficiencies in genes subject to genomic imprinting. PWS is characterized by infantile hypotonia, gonadal hypoplasia, short stature, a moderate delay in physical and mental development, and obsessive/compulsive behavior, as well as neonatal feeding difficulties followed later by hyperphagia leading to profound obesity.74 AS is characterized by severe mental retardation, absent speech, ataxic gait and a happy demeanor.24 Approximately 70% of PWS patients have a 3-4 megabase deletion of the paternal chromosome 15q1 1-q13.49'75'76 The clinically distinct AS results from the same 15ql 1q 13 deletion in about 70% of patients, however the deletion is always on the maternally inherited chromosome.4577'78 Either syndrome can also result from uniparental disomy (UPD), in which both copies of chromosome 15 are inherited from only one parent. The UPD is always maternal in PWS patients 29.76 and paternal in AS patients.3'79 The identification of both of these classes of patients has led to the conclusion that PWS is caused by a loss of gene expression from the paternally inherited chromosome, whereas AS is caused by a loss of gene expression from the maternally inherited chromosome.
Many AS patients have been shown to contain intragenic mutations in the UBE3A gene, indicating that mutations in UBE3A are sufficient to cause AS. 34,35 In contrast to AS, no single gene has been identified for PWS, strongly suggesting that PWS is a
contiguous gene syndrome, requiring the loss of two or more paternally expressed genes to cause the PWS phenotype. 80 All the PWS candidate loci identified to date in humans and mice are expressed exclusively from the paternally inherited chromosome. These include: MKRN3/Mkrn3 (previously known as ZNF127/Zfp127), 46.81 MAGEL2/Magel2, 42,65NDN/Ndn, 40 SNRPN/Snrpn, 11,82 HBII-486, HBH-13/MbII-13, 41 HBII-487, HBII438A, HBII-85/MbH-85, 41,44 IPW/Ipw, '8'47 HBII-52/MbI-52, 43 HBII-438B, and an antisense transcript to the UBE3A gene. 41 Recently, the latter genes (SNRPN through the antisense UBE3A transcript) were shown to be derived from a single transcriptional unit.83
The imprinted genes involved in AS and PWS are regulated by a bipartite
imprinting center (IC) located upstream of the SNRPN gene.26 The IC is divided into the Angelman Syndrome Imprinting Center (AS-IC) and the Prader-Willi Syndrome Imprinting Center (PWS-IC). Previously, it was demonstrated that the PWS-IC is
functionally conserved in the mouse. A 35 kb deletion mutation (the deletion was originally reported to be 42 kb, but is now known that the original map was based on a strain that contained a VL30 insertion) was created in a male embryonic stem (ES) cell line. This 35 kb deletion included 16 kb of upstream sequence plus exons 1-6 of the Snrpn gene. Breeding male chimeric founders (129/Sv) harboring a maternal deletion mutation to wild-type females (C57BL/6J) resulted in mutant offspring that usually died within 48 hours, but never survived beyond 7 days after birth. These mice exhibited several phenotypes similar to those found in PWS infants including small size, poor feeding, and failure to thrive. In addition, these progeny were shown to lack expression of
the local paternally expressed genes Mkrn3 (aka Zfp 127), Ndn, and Ipw22 as well as Magel2, MblI-13, MbII-85, and MblI-52. 43,44,84
Since germline transmission from chimeric males produced mutant offspring that died prior to weaning, this 35 kb deletion mutation was recreated using J4 female
(XO) ES cells. As expected, transmission of the deletion mutation from female chimeras resulted in normal heterozygous carrier offspring.55 Consistent with previous observations, heterozygous carrier males (129/Sv) bred to C57BL/6J females produced mutant offspring that did not survive until weaning, even if the wild-type siblings were removed to reduce competition. This strain has been used to breed unaffected males APWS-IC carriers with mice of different genetic backgrounds to study phenotypic variation. Here, long-term survival of APWS-IC mice upon breeding to several strain backgrounds is reported.
Materials and Methods
Strains and Matings.
Mouse strains used were APWS-IC deletion mice backcrossed onto a C57BL/6J genetic background for 10 generations, C57BL/6J, FVB/NJ, 129/Sv, C3H/HeJ, DBA/2J, and C57BL/6J congenic for a region of Mus musculus castaneus chromosome 7 (B6.CAST.c7).56 APWS-IC males were mated with females of the strains described above to produce affected PWS progeny pups, which were used to ascertain survival and gene expression. For (B6.CAST.c7 x FVB/NJ) F, matings, B6.CAST.c7 females were mated with FVB/NJ males. Conversely, (FVB/NJ x B6.CAST.c7) F, matings were produced using FVB/NJ females and B6.CAST.c7 males.
Culling and Fostering
Litters resulting from matings with APWS-IC males were selectively culled to aid survival of affected APWS-IC pups. Culling was carried out by sacrificing all of the wild-type pups except one within two days after birth, depending on whether or not wildtype pups could be distinguished from their APWS-IC littermates. One wild-type pup was left to stimulate milk production in the absence of robust pups. Mice were fostered to another mother by removing the APWS-IC pups from one mother (usually from the C57BL/6J strain) and giving them to the other mother (usually from the FVB/NJ strain) whose pups had been removed. Additionally, urine from the foster mother was wiped on the foster pups to aid in her acceptance of the new pups. In some fostering cases, two mothers of different strains were set up to litter together, and the litter from the FVB/NJ mother was then removed and the wild-type littermates from the other mother were culled. This allowed two new mothers to care for a small number of APWS-IC pups. Identification of Polymorphisms
Total brain RNA was prepared from both 129/Sv mice and B6.CAST.c7 mice. Each RNA sample was used to create single-stranded cDNA using random primers in a reverse transcription reaction. These cDNAs were used to program PCR reactions using primers corresponding to Ube3a exons 4 and 7 (5'-CCTGCAGACTTGAAGAAGCAG3' and 5'-GAAAACCTCTGCGAAATGCCTI-3'). The resulting products were cloned, sequenced and compared. A polymorphism found in exon 5 created a Tsp509I restriction site in the 129/Sv clone and a Bst4CI site at the same location in the B6.CAST.c7 clone.
A polymorphic AvaIL site between Mus musculus castaneus and Mus musculus domesticus was identified and kindly provided to us by Marisa Bartolomei (M.
Bartolomei, personal communication). This polymorphism results from a cytosine nucleotide, rather than a thymine nucleotide at position 117 of the Ndn transcript. AvalI cuts the domesticus allele, but not the castaneus allele due to the presence of the cytosine nucleotide. PCR primers flanking the polymorphic restriction site were designed as follows: NdnpolyF 5'-ACAAAGTAAGGACCTGAGCGACC-3' and NdnpolyR 5'CAACATCTTCTATCCGTTCTTCG-3'. The PCR product amplified using NdnpolyF and NdnpolyR was gel purified on a 2% agarose gel, extracted from the gel using the QIAGEN gel extraction kit, and digested with Avail. The digested products were electrophoresed on a 4.8% agarose gel (2:1 low-melt agarose: agarose). RT-PCR
Total RNA was isolated from complete brains obtained from neonatal, 1 week
old, 2 week old, and 3 week old mice. RNA was extracted using RNAzol (Tel-Test, Inc.) according to instructions. The total brain RNA (10tpg) was pretreated with DNaseI (Invitrogen), and half of the reaction was subsequently used to synthesize first-strand cDNA with Superscript II reverse transcriptase (RT) and random primers (Invitrogen). The other half of the reaction was manipulated in parallel in the absence of RT. Onetwentieth of the +RT or -RT reactions was used to seed PCRs using the following conditions: 10mM Tris-HC1, pH 8.3, 50mM KC1, the four dNTPs at 0.125 mM each, 1 unit Taq DNA polymerase (Boehringer), and the appropriate primers at 4 [LM each. Sequences of the primers were as follows: Ndn 9F (5'- GTATCCCAAATCCACA GTGC -3'), Ndnl0R (5'- CTTCCTGTGCCAGTTGAAGT -3'), Ndnpoly F (5'- ACAAA GTAAGGACCTGAGCGACC -3'), Ndnpoly R (5'- CAACATC'TTCTATCCGTTC
TTCG -3'), Ube3a 5F (5'- CACATATGATGAAGCTACGA -3'), Ube3a intron 5 R (5'CAGAAAGAGAAGTGAGGTrG -3'), Snrpn N1.1 (5'- CTGAGGAGTGATTGCA ACGC -3'), Snrpn N2.2 (5'- GTTCTAGGAATTATGAGCCCC -3'), P3-actin F (5'GTGGGCCGCTCTAGGCACCAA 3'), and -actin R (5'- CTCTTTGATGTCA CGCACGATTTC 3'). PCR amplifications conditions were 95* C for 5 min., followed by either 20 or 30 cycles of 940 C for 30 s, 60* C for 45 s, and 72* C for 45 s. The final cycle was followed by an extension step for 10 min at 72* C. Northern Blot Analysis
To detect the presence of processed snoRNAs, total RNA was separated on 8% denaturing polyacrylamide gels (7M urea, IX TBE buffer) and transferred to nylon membranes (Hybond N+) using a semi-dry blotting apparatus (Trans-blot SD, BioRad) as described by Cavaille et al. 4' RNA was fixed to the membrane by baking at 800 C in a vacuum oven. The membranes were prehybridized and hybridized overnight at 580 C, according to the Church and Gilbert method. Oligonucleotides complementary to snoRNA sequences were end-labeled using a32 P-dATP (NEN) and T4 polynucleotide kinase (Invitrogen). Their sequences are as follows: MblI-85 (5' TTCCGATGAGAG TGGCGGTACAGA 3'), MbII-52 (5' CCTCAGCGTAATCCTATTGAGCATGAA 3'), and 5.8S rRNA (5' TCCTGCAATT'CACATT'AATTCTCGCAGCTAGC 3'). Probes were purified from free nucleotides using the nucleotide removal kit (QIAGEN). Membranes used for detecting snoRNA expression were washed twice for 15 min at room temperature using 2X SSC, 0.1% SDS, and then exposed to Kodak XAR film at 800 C for 12-36 hours.
In contrast to the fully penetrant neonatal lethality phenotype previously seen on the C57BL/6J strain background, APWS-IC carrier males on the 129/Sv strain background or APWS-IC carrier males made congenic on the C57BL/6J background bred to wild-type female FVB/NJ mice, produced several mutant offspring that survived to two weeks of age. Furthermore, if most of the wild-type competitor sibs are removed, the mutant offspring could survive to adulthood. Although these surviving APWS-IC mice are significantly smaller than wild-type sibs (Fig. 4-1), both males and females are fertile.
To examine whether the increased survival was simply due to superior mothering abilities of the FVB/NJ females as compared to C57BL/6J females, newborn pups were fostered to mothers of the opposite strain and also trio matings of a APWS-IC carrier male with both an FVB/NJ and a C57BL/6J female were established. In both situations, survival of only those F, mice born to the FVB/NJ females was observed, demonstrating that it is the genetic background of either the offspring or the mother that leads to survival rather than the skill of the mother (data not shown).
Next, the expression of several PWS candidate genes in these surviving (FVB/NJ x APWS-IC) F, mice was examined. Surprisingly, low levels of Ndn and antisense Ube3a expression was detected by RT-PCR in newborn mice (data not shown). By Northern blot, expression of several snoRNAs mapping to the region was also detected (data not shown). These results were unexpected as expression of these genes in pups with a paternally inherited PWS-IC deletion had not been previously detected. However, when RNA expression from the 5' half of the Snrpn gene, a region that is physically deleted
from the paternally inherited APWS-IC allele, was examined, expression was detected by RT-PCR (data not shown). Therefore, this strongly suggested that low-level expression detected for Ndn, antisense Ube3a, MbII-85 and Snrpn is derived at least somewhat from leaky expression of the maternal FVB/NJ allele. A time course for the observed lowlevel expression of Ndn and Snrpn at newborn, 1 week, 2 weeks, and 3 weeks of age was followed, and the leaky expression was detectable in all ages of mice (Fig4-2). It was hypothesized that expression of these genes is sufficient to rescue the lethality phenotype associated with the APWS-IC mutation.
Next, the extended survival and leaky expression of these imprinted genes was investigated to determine if it was restricted to the FVB/NJ strain. Females for the C57BL/6J, DBA/2J, BALB/cJ, C3H/HeJ, and 129/SvEv strains were mated to APWS-IC carrier males on the C57BL/6J genetic background. Wild-type pups were removed from the litters within two days after birth and the longest possible survival of APWS-IC pups was determined. Pups derived from C57BL/6J and DBA/2J females had the shortest survival time, with the longest surviving affected individual from either strain living only
7 days. Pups derived from all other strains had at least one affected individual that survived for at least 3 weeks. Leaky expression of Snrpn, Ndn, and MbII-85 was observed at comparable levels in all strains (Fig. 4-3), regardless of their ability to survive.
Finally, the leaky maternal expression of these PWS candidate genes was
investigated in wild-type mice. FVB/NJ females were mated with wild-type males from the B6.CAST.c7 strain, a strain that is congenic for the mus musculus castaneus chromosome 7 on a C57BL/6J background. The reciprocal mating was also performed.
Total brain RNA was isolated from the resulting (FVB/NJ X B6.CAST c.7) F, or (B6.CAST.c7 X FVB/NJ) F, mice at birth. RT-PCR followed by a polymorphic restriction enzyme digest of the RT-PCR product was used to distinguish between expression of Ndn produced from the domesticus and castaneus alleles. Only the paternal castaneus allele was expressed in each case (Fig 4-4). B6.CAST.c7 females were then mated with APWS-IC carrier males and the same RT-PCR and restriction digestion on brain RNA made from (B6.CAST.c7 X APWS-IC) F, pups was performed. While expression from wild-type pups was limited strictly to the domesticus allele, the expression in pups that had inherited the paternal APWS-IC allele was biallelic, representing both domesticus and castaneus alleles. This indicates that leaky maternal expression is only observed or detectable when in combination with a paternal APWS-IC mutation.
This study demonstrated that strain background affects the phenotype of the APWS-IC mice. Heterozygous carrier males (either 129/Sv or C57BL/6J) bred to C57BL/6J or DBA/2J females produced mutant offspring that do not survive until weaning, even if wild-type siblings are removed to reduce competition. However, longterm survival was achieved upon breeding these same carrier males to several other strains, including FVB/NJ, C3H/HeJ, 129/Sv, and BALB/cJ. While surviving APWS-IC mice are smaller than wild-type littermates, they do not become obese on a low-fat diet, both males and females are fertile, and they appear to be normal in all other respects.
Expression analysis of Snrpn, the antisense transcript of Ube3a, and the MbII-85 snoRNA reveals leaky expression of these paternally expressed genes in APWS-IC pups.
This expression occurs at a very low level, and originates, in part, from the maternally derived chromosome. This leaky expression, originally detected in PWS-IC deletion pups with FVB/NJ mothers, is also present at comparable levels in all other strains, including C57BL/6J and DBA/2J. This suggests that the increased survival of APWS-IC pups is not due to the low-level expression of genes in the PWS region. Furthermore, this expression is not sufficient to ameliorate the failure to thrive phenotype and permit survival in APWS-IC pups. Similarly, humans that have PWS due to maternal uniparental disomy or a balanced translocation have also been reported to show weak, but detectable expression of genes in the PWS region.8" No obvious clinical differences were reported from individuals with leaky expression from imprinted genes when compared to affected individuals that did not show leaky gene expression from these genes.","8
Why would APWS-IC pups survive on some strains and not others, when the gene expression in the PWS region appears to be identical between survivors and nonsurvivors? The most likely explanation for this would be the presence of specific modifier alleles in some inbred strains that either permit or prevent survival of affected pups. An alternate explanation may be the general unfitness of a particular strain for survival of neonates. This is unlikely, however, as the APWS-IC pups are genetically F, individuals and should benefit from hybrid vigor.
Heterozygous carrier males on the 129/Sv genetic background bred to C57BL/6J females, produce APWS-IC pups that never survive. Conversely, carrier males on the C57BL/6J genetic background bred with I 29ISv females produce APWS-IC pups that are capable of surviving to adulthood, even though the genetic background of these reciprocal matings are identical. One explanation of this is that the modifier gene or
genes responsible for survival of APWS-IC mice could be required in the oocyte. The oocyte specific factor would not be present in C57BL/6J oocytes that are not permissive for survival, but would be present in 129/Sv oocytes and allow survival of affected pups. An alternate explanation is that the modifier gene or genes are themselves imprinted. If the modifier gene was only expressed from the maternally inherited chromosome, and the allele of this gene that was permissive for survival was the 129/Sv allele and not the C57BL/6 allele, then survival could only be achieved when the maternal allele of this gene is from the 129/Sv strain. Pronuclear transplant, where the pronuclei from a (B6 X APWS-IC) F, oocyte could be moved into "emptied" oocytes from fertilized FVB/NJ or (FVB/NJ X APWS-IC) F females, could be used to distinguish between these two possibilities. If the transplanted pronuclei gave rise to pups that were able to survive, then the modifier gene is most likely oocyte specific, however, if the resultant pups still do not survive, then the modifying gene is most likely imprinted.
It is interesting that the leakiness of the maternal allele is only detectable when in combination with a paternal APWS-IC allele, but not when in combination with a wildtype paternal allele. There are three possible explanations for this. First, low-level expression from the maternal allele would produce a few transcripts from the maternal allele that could simply be swamped by the more abundant transcripts originating from the paternal allele. Perhaps imprinting in mouse is only strong allele bias, and low-level expression could mark the repressed chromosome as a region that is repressed in a facultative manner as opposed to a constitutive manner. Another possibility is that the expression of genes or transcripts downstream of Snrpn could be allowed by the splicing of the upstream exons of Snrpn 17that are left intact on the APWS-IC chromosome into
the downstream exons of Snrpn or the run-on transcripts from which the snoRNAs are spliced."3 These upstream exons have been shown to be functional alternate promoters for Snurf, but their level of expression is much weaker than that of the major promoter. Perhaps this could explain the weak expression of the snoRNAs. Additionally, the physical interaction between the silenced imprinting center and the upstream genes, including Mkrn3, Magel2, and Ndn, could be required for the silencing of these genes that lie more than 1 Mb away from the imprinting center. The missing imprinting center could cause the alternate Snurf promoters to be juxtaposed to the regulatory elements of the upstream genes, and result in leaky expression. While this may explain leaky paternal expression, it doesn't explain residual expression from the maternal chromosome. The third explanation is that communication between one silenced and one expressed imprinting center is required for the proper silencing of the maternal allele and the expression of the paternal allele. Specifically, the unmethylated, unsilenced, paternal PWS-IC may be required in trans to silence the maternal region. While this explanation could account for the bi-allelic: expression in the APWS-IC mice, it doesn't explain why the expression levels of the imprinted genes are so low. Furthermore, trans-activation is a phenomenon not documented in mouse and would be difficult to prove or disprove. Leaky gene expression in the region is not unique to mice, as human patients with maternal uniparental disomy have also been reported to show leaky expression "5,".
--O-wild-type 10 -O---wild-type
---IC-deletion 5. --IC-deletion
3 6 9 12 s15
Figure 4-1. Surviving APWS-IC mice are smaller than their wild-type littermates. APWS-IC mice and their wild-type littermates were weighed at 3, 6, 9, 12, and 15 weeks of age. Their weight in grams is plotted versus their age in weeks. Filled symbols represent APWS-IC mice and open symbols represent the wild-type littermates.
No RT control
Figure 4-2. Leaky expression from paternally expressed genes is not age dependent. APWS-IC mice were sacrificed at birth, one week, two weeks, and three weeks of age. Total brain RNA was made from the sacrificed mice and subjected to either RT-PCR or northern analysis. RT-PCR is shown for Ndn and 5' Snrpn since the expression level is not detectable by northern analysis. 5.8S rRNA and P-actin were used as controls for northern analysis and RT-PCR, respectively. Low-level expression is noted in APWS-IC mice regardless of their age.
B6 FVB/N C3H/HeJ Balb/c 129/Sv DBA/2J
WA Aw A A wA A wA Aw A A wA A
No RT control
Ndn RT-PCR 20 cycles, blot
20 cycles, blot
Figure 4-3. Leaky gene expression does not account for strain dependent
differences in phenotype. APWS-IC mice made congenic on the C57BL/6J strain
background were mated with either C57BL/6J, FVB/NJ, C3H/HeJ, Balb/cJ, DBA/2J, or 129/Sv females. Total RNA was prepared from the brains of two
APWS-IC mice and one wild-type littermate. The RNA was subjected to RT-PCR and northern analysis. Ndn and 5'Snrpn were analyzed by RT-PCR at 40 cycles or at 20 cycles and blotted and probed. 5.8S rRNA and f-actin were used as controls
for northern analysis and RT-PCR, respectively. APWS-IC mice showed leaky
expression in all strains analyzed.
A FVB/NJ x B6. Castc7/B6.Cast.c7 x FVB/NJ
FxC C x F genomic genomic
B B6.Cast.c7 x APWS-IC
Figure 4-4. Leaky gene expression is biallelic in APWS-IC pups. A. Brain RNA from reciprocal crosses between FVB/NJ and B6.CAST.c7 was subjected to RT-PCR with primers flanking a polymorphic Avall site in the Ndn transcript from mus musculus domesticus. The product was digested with AvalI and run on a 4.8% agarose gel. Wild-type Ndn expression in both crosses originates from the paternally inherited chromosome. B. B6.CAST.c7 females were mated with males that were carriers for the APWS-IC allele. APWS-IC and wildtype littermate pups were sacrificed at birth and total RNA was prepared from their brains. RT-PCR and Avall digestion of the Ndn transcript was performed as described above. Wild-type Ndn expression occurs from the paternally inherited allele, while APWS-IC expression occurs from both paternal and maternal alleles.
TRANSGENIC RESCUE OF THE PWS-IC DELETION MOUSE Introduction
Prader-Willi syndrome (PWS) patients first present with severe infantile
hypotonia and failure to thrive that leads to a requirement for gavage feeding in most PWS neonates. The infantile phenotype subsides after several months and gives way to a period of normal feeding behavior, but soon progresses to hyperphagia and severe obesity, if uncontrolled.23 PWS children also possess a distinctive behavioral phenotype that is similar, but not identical to obsessive-compulsive disorder. Outwardly visible signs of the disease in humans may also include almond-shaped eyes, strabismus, small stature, small hands and feet, and hypogonadism.23'" Caused by the physical or functional deletion of paternal chromosome 15q 11-q 13, PWS occurs at a frequency of about 1/15,000 live births.226 Approximately 70% of PWS patients have sustained a large (3-4 Mb) deletion of their paternally inherited chromosome 15. Other PWS patients have either inherited both copies of chromosome 15 from their mother (maternal uniparental disomy), or have an imprinting mutation. Patients with an imprinting mutation have either a deletion in the 5' region of Snrpn, including exon 1, an area
knon a th ceter ofhe6,11-1,2
known as the imprinting center (IC) of the q 1-q1 3 region,' 13.20 or have no detectable DNA sequence mutation, but still possess a paternally inherited chromosome that epigenetically behaves as a maternally inherited chromosome.
Human chromosome 15q 1-q 13 contains at least six paternally expressed genes or transcripts and is syntenic with central chromosome 7 in mouse. The genes and transcripts include: MKRN3, MAGEL2, NDN, SNRPN, HBII-13, HBII-436, HBII-437, HBII-438A, HBII-438B, HBII-85, HBII-52, and the UBE3A-antisense transcript.17"18,21,39-A Expression of each of these genes is completely absent in most PWS patients, although some patients show low-level expression, but still present with classical PWS.85'86 A large region of sequence between SNRPN and NDN still remains unexplored and may contain additional genes. This region encompasses nearly 1.5 Mb, and additional genes discovered in this region might be predicted to be imprinted and paternally expressed.
The distinct paternal inheritance pattern of PWS identifies 15qI 1-q13, as well as the syntenic mouse chromosome 7 as a region of genomic imprinting.7-'2'39 Genomic imprinting involves the expression of only one parental allele in somatic cells. The other parental allele is transcriptionally silenced. The non-transcribed alleles of the PWS genes, maternal in the case of PWS-related genes, are often associated with nearby methylated cytosine residues present in CpG islands, while the transcribed alleles, are associated with the lack of methylation at these same CpG residues.6"''3 This indicates that the maternal allele has a different epigenotype than the paternal allele. It is not known whether the eventual silencing of the maternal allele causes this epigenetic mark, or whether it is a mark to identify the allele to be silenced.
Conversely, Angelman syndrome (AS) results from the functional deletion of the maternal complement of the same interval of chromosome 15.24 A unique class of AS patients with specific mutations in the UBE3A gene suggests that loss of UBE3A expression can be solely responsible for the major clinical features of AS.3435 Another
maternally expressed gene, ATPIOC, has also been identified in the PWS region, but its contribution to the phenotype of AS is yet to be determined.36 AS can result from maternal deletion of 15q 11-q 13, paternal uniparental disomy, IC mutations, and from putative single gene mutations, as evidenced by completely normal epigenotype in the region and the UBE3A mutation class which present with classical AS.24 The absence of the class of PWS patients in which a single gene mutation is responsible for the complete phenotype, as well as the presence of multiple paternally expressed genes that are disrupted, indicate that PWS is a contiguous gene syndrome where the clinical features are the cumulative effect of multiple disrupted genes."0'8
PWS and AS patients with imprinting defects have either a relatively small deletion or no noticeable mutation at all, but these mutations alter the entire 3-4 Mb region on the paternal and maternal chromosomes, respectively.20 In PWS-IC mutation patients, the paternally derived genes assume maternal, or silenced, epigenotypes. The deletion breakpoints of several PWS patients with IC mutations were mapped to identify the smallest region of deletion overlap for PWS (PWS-SRO).'3"7 The PWS-SRO is located 5' of the SNRPN gene, includes the first exon of the gene, and is often referred to as the PWS-IC. Although the PWS-SRO has been narrowed down to approximately
4.3kb, the actual deletions that have been shown to cause PWS are much larger.37'38 A smallest region of deletion overlap has also been identified for AS, and is referred to as the AS-SRO. The AS-SRO lies around 35 kb upstream of the PWS-SRO, and has been narrowed to 0.8 kb.37'38 AS-IC mutations never appear to overlap the PWS-SRO, while PWS-IC mutations can overlap the AS-SRO, suggesting that in the absence of the PWSIC, the AS-IC is not necessary. This indicates that the sole purpose of the AS-IC may
be to regulate the PWS-IC in the maternal germlineY This also indicates a bipartite structure to the IC, in which the PWS and AS components work together to control imprinting in the 15ql 1-q13 region. In mouse, the PWS-IC has been functionally defined by deletion, indicating that the bipartite IC is most likely functioning similarly between humans and mice.
Our lab previously created a mouse model of PWS that emulates both the
phenotypic and molecular characteristics of PWS by deletion of the mouse PWS-IC.22 Mice that inherit a PWS-IC deletion (APWS-IC) from their father lose expression of the known paternal-specific genes, including Snrpn, Mkrn3, Ndn, and the Ube3a antisense transcript. The APWS-IC mice also demonstrate poor suckle and failure to thrive, indicative of the infantile characteristics of human PWS. These mice thus provide an excellent mouse model of PWS. Unfortunately, APWS-IC mice on the inbred C57BL/6J background die within seven days of birth, presumably a result of poor suckle and failure to thrive. Progression to obesity and the development of other PWS related phenotypes has yet to be seen in APWS-IC mice due to neonatal lethality. APWS-IC mice mated with mothers of other strain backgrounds survive to adulthood if the wild-type siblings are removed shortly after birth. These surviving APWS-IC mice remain smaller than their wild-type siblings, but are otherwise grossly normal. Furthermore, surviving APWS-IC females and males are both fertile and never become obese on an ordinary rodent diet.
Individual knockouts of Snurf, Snrpn, and Mkrn3 appear completely normal and show no PWS associated phenotype.22899 Conflicting data regarding the phenotype of the Ndn gene knockout have been reported. Tsai and colleagues report that the Ndn
knockout mouse has no discernable phenotype, while Gerard et al. and Muscatelli et al. report partial neonatal lethality that is strain dependent and not consistent from generation to generation.9'-92 On the other hand, mice inheriting a paternal deletion spanning from exon 2 of Snrpn to Ube3a succumb to neonatal lethality in 80% of the affected pups."' Together, this data strengthens the hypothesis that PWS is a contiguous gene syndrome caused by the absence of multiple gene products, and not caused by the absence of any single gene product. Despite the efforts of many labs to create individual gene knockouts, the contribution of each gene in human chromosome 15q II-q 13 and the syntenic mouse chromosome 7 to the PWS phenotype is still a mystery. Figure 5-1 shows a summary of the individual gene deletions that have been reported.
Efforts to model PWS using individual gene knockouts have been largely
unfruitful .22,19,90 -9 However, engineering multiple gene deletions is also limited by the location of the imprinting centers (several desired mutations would remove them, creating an imprinting mutation). Therefore, a different approach was adopted to determine the individual gene contribution to this contiguous gene syndrome. This approach involved transgenic rescue, and the strategy was to produce transgenic mice harboring either individual or multiple PWS genes and mate them to APWS-IC mice. In mice paternally inheriting an imprinting center mutation and maternally inheriting a transgene, the ability of the transgene to rescue the phenotype of the APWS-IC mice could be ascertained. The genes responsible for the individual aspects of the PWS phenotype could then be definitively determined, since one transgene could conceivably rescue the neonatal lethality defect, while another could rescue the small stature defect. In addition, the rescued mice may live longer or recapitulate other human PWS
symptoms and allow the dissection of additional later onset phenotypes in our mouse model.
Materials and Methods
Screening the BAC Libraries
The Research Genetics (Huntsville, AL) BAC library was initially screened for BAC clones. This library was made from the 129/Sv strain of mouse. The library consisted of 9 filters, each carrying 27,648 unique clones, with an average insert size of 130 kb, spotted in duplicate onto 9 different membranes. Probes for Ndn, Mkrn3, Ube3a, and the gene formerly known as Ipw were used to screen the library simultaneously.
Prior to hybridization, BAC membranes were washed with 1500 ml, of 6X SSC and 0. 1% SDS at room temperature for 15 minutes. The membranes were then rinsed twice with 1500 ml of 6X SSC for 15 minutes at room temperature, and hybridized according to the Research Genetics protocol. Briefly, the membranes were prehybridized in roller bottles with 3 membranes per bottle, each separated by one sheet of Flow Mesh (Diversfied Biotech, Boston, MA), in 120 ml of HyperHyb (Research Genetics, Huntsville, AL) per bottle at 650 C for 20 minutes. At least 10' counts of each probe were boiled for 5 min, snap cooled on ice, and then added to 3 ml of HyperHyb that was pre-warmed to 650 C, and 1 ml of hybridization solution was added to each of the roller bottles. The membranes were allowed to hybridize for 2 hours at 650 C.
The membranes were washed in the roller bottles three times for 15 minutes at
65' C with 30 ml IX SSC and 0.1% SDS. They were then removed from the bottles and washed twice more at 650 C with 1OO0ml of IX SSC and 0. 1% SDS for 15 minutes. Finally, the membranes were rinsed with IX SSC at room temperature. The membranes
were then wrapped in cellophane and exposed to film (XAR, Kodak) overnight at -80 C. Positive clones were identified following the manufacturer's instructions, and the following clones were ordered: 454N20, 205L17, 581E10, 173C16, and 143C10. In addition to screening the 129/Sv library, a single RPCI-23 BAC, 452P17 was found by screening the library virtually with a probe to the 3' most intron of the Ube3a gene. Probe Preparation
Probes were labeled using the Prime-It II kit (Stratagene, La Jolla, CA), with
minor modifications. DNA (200 ng)was boiled for 5 minutes in the presence of random hexamers. The DNA-hexamer mix was cooled on ice, and 5X reaction buffer was added along with 25 [tCi of c32P dCTP and 1 U of exo klenow. The labeling reaction was placed at 370 C for 15 minutes. After labeling was complete, the probe was purified using the nucleotide removal kit (QIAGEN) and eluted in 200 p.l of water. Purified probes were boiled for 5 minutes and snap cooled on ice prior to use in hybridization. Verification of BAC Clones
Single BAC colonies were tested to determine which clones were derived from each gene locus. Four single colonies from each BAC were streaked onto LB plates supplemented with the antibiotic, chloramphenicol (LB-CA). The plates were overlayed with a nylon filter and grown overnight at 370 C. The next morning, the bacteria were lysed on the filters as follows: first, the filters were soaked in alkali solution for 4 minutes, followed by neutralizing solution for 4 minutes, and finally rinsed in 6X SSC for 1 minute. The filters were then baked at 800 C for 2 hours, pre-hybridized in Church and Gilbert buffer for one hour at 650 C, and hybridized with the Ndn probe. Upon washing and laying the filters to film, the results were used to determine which clones were
derived from the Ndn locus and whether the culture sent represented a pure colony from a single clone, or a mixed culture with multiple BAC clones. The filters were then stripped in boiling 0.5% SDS and the entire hybridization procedure was performed again with the remaining probes.
End Sequencing BAC Clones
The BAC clones were end sequenced using the big dye termination reaction to obtain sequences unique to the mouse genomic DNA insert. Reactions were performed on 1 mg of BAC DNA that had been sheared by passing it five times through a 25 gauge needle, using a modified version of the manufacturer's instructions, with 12 [d of sequencing buffer, 4 id of big dye, 16pmol of primer, and 2.5mM MgCl2. The reactions were subjected to an initial 5 min. cycle at 950 C, followed by 30 cycles of 950 C for 30 s, 500 C for 10 s, and 600 C for 4 min. The completed reactions were purified from free nucleotides using Performa DTR Gel Filtration Columns (Edge Biosystems), vacuum dried, and sequenced by the University of Florida Center for Mammalian Genetics Sequencing Core.
Pulse-Field Gel Electrophoresis
Pulse-field gel electorphoresis was carried out to create a rough restriction map and tiling pathway for the BAC clones. BAC DNAs digested with the appropriate enzymes were electrophoresed on a 1% genetic technology grade agarose gel (Nusieve, FMC) made with IX TBE buffer. Samples were mixed with loX ficoll loading dye and ran at 200V with a switch time of 1-12 seconds for 12.5 hours. This provided resolution of bands from 6-200 kb. The gel was stained with a dilute ethidium bromide solution, photographed, and blotted as described below.
Agarose gels carrying 5 pg of digested genomic DNA or 1 [.tg of digested BAC DNA were photographed and UV nicked for 5 minutes. The gel was soaked in Alkali solution for 45 minutes, followed by soaking in neutralizing solution for 90 minutes. The gel was transferred to Hybond nylon membrane (Amersham) in loX SSC overnight. The membrane was rinsed in 2X SSC and baked at 80* C for 2 hours. Hybridization was carried out according to the method of Church and Gilbert. Membranes were prehybridized in 20 ml of Church and Gilbert Buffer for 1 hour at 65* C. The prehybridization solution was poured off and replaced with 5 ml of fresh solution and the boiled, cooled probe. Hybridization continued overnight at 650 C. Washing was carried out three times in 2XSSC and 0.1% SDS at 65' C for 15 minutes each. Transgenic Mice Production
For production of transgenic mice, BAC DNA was prepared by large scale
plasmid preparation, using the alkaline lysis method. First, log phase cultures of BAC culture grown in LB-CA were centrifuged for 5 minutes at 6,000 rpm to pellet the cells. The supernatant was poured off and the pellet was resuspended in GTE solution plus lysozyme. The suspension was then lysed with NaOH and SDS. After addition of the lysis solution, the lysate was neutralized in 0.5M potassium acetate, and poured through cotton gauze into Oak Ridge tubes. BAC DNA was precipitated with ethanol, and the pellet was resuspended in 4 ml TE with 10 [tg/ml RNAse A, and allowed to incubate at 370 C for 1 hour. Cesium chloride and ethidium bromide were added to the DNA solution, and the solution was loaded into a heat sealable tube. The tube was sealed and spun at 58,000 rpm for 18 hours. BAC supercoils were viewed with a UV light, and both
nicked and supercoiled bands were pulled. The DNA was extracted with tert-butanol until the ethidium bromide was completely depleted from the BAC DNA. The DNA was then dialyzed twice against I L TE buffer, using dialysis tubing for 6 hours each. The concentration of the BAC DNA was then measured by spectrophotometer, and then diluted to 2.5 ng/Rl for microinjection.
Transgenic mice were made by pronuclear injection into oocytes that were
obtained from superovulated, fertilized, FVB/NJ female mice. The fertilized, injected oocytes were then implanted into (B6D2) F, females that were made pseudopregnant by mating them with vasectomized (B6D2) F, males. Mouse Husbandry
At 3 weeks of age approximately 2mm of tail was clipped from each mouse, and the ear was punched for identification. Genomic DNA was prepared from the tail piece by incubation in tail lysis buffer with proteinase K (100 ng/pl) overnight at 550 C. The tail lysate was extracted with phenol: chloroform: isoamyl alcohol (25:24:1) and then precipitated with ethanol. The tail DNA was screened for the presence of the transgene by PCR using both end sequences from the BAC. Mouse Strains
APWS-IC mice were developed in our lab 22.55 and constituted a 35 kb deletion encompassing the sequence froml6 kb 5' of Snrpn and 19 kb 3' of Snrpn. These mice were backcrossed for 10 generations to C57BL/6J males and were thus made congenic on that strain background.
Transgenic mouse strains were originally maintained on the FVB/NJ genetic background, but selected strains were then backcrossed for 10 generations to the
C57BL/6J strain background. Four strains of 454N20 transgenic mice are referred to as 454A-D, one strain of mice representing the P1 clone 967 is referred to as line 1707, and
5 lines representing transgene 380J10 are referred to as 380A-E.
Screening the BAC Library for Clones
The first step toward transgenic rescue was to screen a commercially available BAC library for clones that would putatively contain the PWS genes. BAC clones were chosen to perform the rescue as they are likely to contain all of the necessary regulatory regions for the individual genes due to their size. They are also relatively easy to handle, as they are propagated in bacteria, rather than yeast. Additionally, they can be modified by homologous recombination.7'3 The Research Genetics BAC library was first screened for clones. This library is made from the 129/SvEv strain of mouse, and the average insert size for BAC clones is 130kb. Some clones by screening the RPCI-23 BAC library were also obtained. This library is made from a female C57BL/6J mouse. The average insert size for this library is 200 kb and is the template for the public mouse genome sequencing effort.
The BAC libraries were screened using probes for Ndn, Mkrn3, and the gene formerly known as Ipw. With the first screen of the Research Genetics library, two BACs were obtained that hybridized with the probe for Mkrn3, one of which also hybridized with the Ndn probe. No clones for Ipw were identified. From subsequent screenings of this library, one BAC containing Ipw, one that hybridized with a probe for the Ube3a gene, and one that hybridized with Ndn, Mkrn3, and Magel2 were isolated. From screening the RPCI library with the Snrpn probes, a single BAC that spanned most
of the distance between Snrpn and Ipw was isolated. An electronic database search also identified a single BAC that contained the 3' most intron to Ube3a.
Addresses for the BAC clones were determined, the BACs were ordered, and then they were streaked out to determine their clonality and to verify the presence of the probes that the library was screened for. The span of the individual BAC clones were determined by end sequencing, making end probes by PCR, and using them to probe southern blots produced from restriction digestion with rare cutting enzymes, separated by pulse-field electrophoresis. From this it was determined that the previously published gene order in the Ndn region was incorrect. The correct order places Magel2 between Mkrn3 and Ndn, and in a transcriptional orientation opposite that of Mkrn3. The BACs that were isolated were 454N20, which contains Mkrn3, Magel2, and Ndn; 205L17, which contains Frat3 and Mkrn3; 173C16, which contains Ube3a; and 380J10, which extends from Snrpn to the MbII-85 snoRNA cluster (Fig 5-2). Production and Characterization of Transgenic Mice
Once the BACs of interest had been identified and mapped, transgenic mice were made. BAC DNA was prepared by large scale plasmid preparation, followed by cesium chloride banding of the resulting DNA. Both nicked and supercoiled bands were pulled and used for production of transgenic mice. It was important to have extremely fresh BAC, as the larger BACs from Roswell Park degraded rapidly after isolation. The DNA was injected at 2.5 ng/ul. Transgenic mice were made by pronuclear injection into superovulated, fertilized, FVB/NJ female mice. The fertilized oocytes were then implanted into pseudopregnant (B6D2) F, mice. When the resulting pups reached 3 weeks of age, they were tail clipped to obtain DNA and earpunched for identification.