Page 49 - Mouse Molecular Genetics

Full Abstracts
Program number is above title. Author in bold is the presenter.
49
and complex phenotypes including diseases, as well as comparison with syntenic regions of the human genome. The CNV data
also provide insight into the mutational mechanisms underlying evolution of genome structure. Knowledge of structural variation
and instabilities in the mouse genome is pivotal in both the study of the evolutionary history of the mouse genome and
identification of de novo copy number variants.
43
PRISM.stanford.edu: 2,500 Human and Mouse transcription factor developmental function predictions.
Aaron Wenger
1
,
Shoa Clarke
2
,
Jenny Chen
3
,
Cory McLean
1
,
Gill Bejerano
1,4
. 1)
Computer Science, Stanford Univ, Stanford, CA; 2) Genetics,
Stanford Univ, Stanford, CA; 3) Biomedical Informatics, Stanford Univ, Stanford, CA; 4) Developmental Biology, Stanford
Univ, Stanford, CA.
Recently we developed GREAT.stanford.edu to derive insights into transcription factor (TF) function from ChIP-seq peaks.
GREAT works remarkably well on a variety of sets. It does so because the majority of TFs bind directly (and often multiple
times) next to a large (10-500) number of contextual target genes. However, the same analysis also shows that when a TF has
multiple functions, as most of them do, only the subset of functions relevant to the experiment at hand are observed. Thus for
example, ChIP-seq of SRF in immune cells reveals its role in regulating the actin cytoskeleton, but not its roles in muscle
development. In GREAT we have collected a vast body of knowledge on all human and mouse genes in many different contexts.
Meanwhile, using ChIP-seq and other technologies we have learned the DNA binding preferences of hundreds of TFs. This led
us to ask the following question: If we take the binding site preferences of a TF, make stringent genome wide binding site
predictions for it, and subject these to GREAT, could we accurately predict additional TF functions not directly observed in
ChIP-seq performed to date? For example, when we predict binding sites for SRF and analyze them with GREAT we rediscover
SRFs role in regulating the actin cytoskeleton. We also learn that SRF regulates muscle, and that its misregulation may result in
dilated heart ventricles, and other true SRF functions not seen (and not relevant) in immune cell ChIP-seq. Starting from a high
quality non-redundant library of 300 transcription factors, and using very stringent cut-offs we obtain 2,500 transcription factor
role predictions (of the form SRF regulates 142 target genes, all annotated for actin cytoskeleton, using these 356 predicted
binding sites, p1E-57), at an average TF function false discovery rate (FDR) of 15% and per binding site FDR of 50%.
Conservative computational analysis immediately confirms over 10% of our function predictions in a variety of contexts,
implicating many novel target genes and binding sites for known TF functions, and predicting exciting novel TF functions.
cDNA co-transfections in primary cells matching our TF function predictions shows that our predictions are highly enriched for
active enhancers that responds to the predicted TF. A portal (http://PRISM.stanford.edu) will provide searchable access to this
unique resource, which we hope will be of great hypothesis generation value to the developmental biology community.
44
Generation of rat-mouse interspecific chimeras for study of organogenesis and elucidation of xenogenic
barrier. Hiromitsu Nakauchi
1,2
. 1)
Division of Stem Cell Therapy, Center for Stem Cell Biology and Regeneration Medicine,
Institute of Medical Science, University of Tokyo, Japan; 2) Japan Science Technology Agency, ERATO, Nakauchi Stem Cell
and Organ Regeneration Project.
Recent development of a culture system using small molecule inhibitors of glycogen synthase kinase 3 (Gsk3) and the Fgf-
MAPK signaling cascade has enabled efficient derivation of pluripotent stem cells (PSCs) from not only all mouse strains but
also a non-mouse species, Rattus norvegicus. Using rat ES cells and iPSCs we established, we recently demonstrated the
generation of mouse-rat interspecific chimeras. These interspecific chimeras were generated in two ways; injecting rat PSCs to
mouse blastocysts and injecting mouse PSCs to rat blastocysts. However, both embryonic survival rate and the degree of
chimerism were lower in interspecific chimeras than in intraspecific chimeras, implying that there is a xenogeneic barrier for the
development of interspecific chimeras. Interestingly, the size of interspecific chimeras grown into adulthood seemed to conform
with that in the species from which the blastocyst originated. The origins of placenta and uterus may have a key role in body size
determination. We also demonstrated in mouse the generation of functionally normal rat pancreas by injecting rat PSCs into
Pdx1-/- (pancreatogenesis-disabled) mouse embryos, providing proof of principle for organogenesis from xenogenic PSCs in an
embryo unable to form a specific organ. Although the pancreas thus generated were composed of almost entirely by rat PSC-
derived cells, the size was that of mouse indicating that the size of an organ is determined not by cell intrinsic manner but by
environmental factors. The results indicate that interspecific chimeras and generation of organs in vivo using PSC-complemented
blastocysts provides a new strategy for elucidation of xenogenic barriers and understanding organogenesis.