Biology Reference
In-Depth Information
performed to determine the functions and effects on gene
regulation of individual TF-bound sites.
The importance of how CREs (either experimentally
determined TF DNA-binding site motifs or computation-
ally inferred regulatory motifs) are organized within more
complex CRMs such as enhancers, and the importance of
maintaining specific TF-binding site sequences, has only
begun to be explored. Such effects will likely vary across
different types of enhancers with some belonging to the
'enhanceosome' type for which a specific arrangement of
TF DNA-binding sites is required, and others having more
flexible organization or composition of DNA-binding sites
as in 'billboard' type enhancers [187] . Improved compu-
tational methods and experimental testing of many CRMs
will be required to decipher the 'grammar' of how the
organization of TF-binding sites within CRMs confers the
appropriate effect on gene expression in response to
particular cellular and environmental contexts. Under-
standing the 'language' of how gene regulation is encoded
in the genome will also permit more accurate prediction
and understanding of variation within TF e DNA-binding
sites, either within or between species. Accurate identifi-
cation of non-coding regulatory variation and prediction of
the associated effects on gene expression remain significant
challenges and are important for a mechanistic under-
standing of the functional consequences of numerous
disease- or other trait-associated non-coding variation
identified in genome-wide association studies (GWAS)
[188] .
A variety of emerging model organisms is proving to be
powerful in studies of the role of regulatory variation and
the evolution of phenotypic variation. For example,
multiple cases of changes in wing pigmentation patterns
across insects (Drosophila, or more broadly, Diptera) have
been found to be due to changes in CREs [189,190] . Partial
or complete loss of pelvic structures in populations of
three-spine stickleback fish (Gasterosteus aculeatus) show
phenotypic variation across populations, with some pop-
ulations exhibiting partial or complete loss of pelvic
structures; it has recently been shown that pelvic loss in
these populations is due to different deletions overlapping
a tissue-specific enhancer of the Pituitary homeobox tran-
scription factor 1 (Pitx1) gene [191] .
GRNs evolve with the duplication and/or divergence of
members of gene families, including TFs. Different DNA-
binding site sequences and associated target genes can be
recognized by different members of TF families
[18,103,192 e 194] . Accurate models of GRNs will require
an improved understanding of the redundant vs. the non-
overlapping roles of paralogous regulatory factors.
Network models that capture the effects of different
genetic, cellular, or environmental conditions will play
important roles, as the complexity of GRNs is too high to
permit reliable prediction from manual interpretation of the
network. To this effect, it is important to move beyond
qualitative network diagrams that are convenient for visual
representation, to the development of quantitative and
dynamic models that can be used to make accurate
predictions of network behavior, in particular upon genetic,
cellular, or environmental perturbation(s). Except for
a handful of small, well-defined model systems, in general
there is a lack of reliable experimental data on nuclear TF
concentrations, TF e DNA-binding affinity constants (K d
values), protein localization data and measurements of
protein activity (vs. just abundance), all of which are
necessary to parameterize detailed computational models
of transcriptional regulatory networks. Such data will be
important for improved models of GRNs and a better
mechanistic understanding of stochasticity ('noise') in
gene expression [195 e 197] . In addition, there is a need for
visual representations of networks that depict the condi-
tion-specific activity of portions of the networks, beyond
simple graphs of edges connecting interacting nodes.
Experimental testing of predictions arising from the infer-
red models will continue to be crucial for validation and
refinement of network models.
This chapter has focused on the roles of TFs and CREs
in GRNs. Numerous other regulatory mechanisms play
important roles in regulatory networks. Some of these,
including nucleosome occupancy and histone modifica-
tions, chromosomal conformations, non-coding regulatory
RNAs such as lincRNAs and microRNAs, DNA methyla-
tion, RNA-binding proteins and untranslated regions
(UTRs), affect transcript levels, whereas others, such as
protein e protein interactions and small molecule ligands,
affect protein levels or activity. Incorporation of these
additional features into integrated models will be important
for a more complete understanding and more accurate
prediction of the effects of perturbations of GRNs.
ACKNOWLEDGEMENTS
The authors thank Trevor Siggers, Luis Barrera, Raluca Gordˆn, and
Stephen Gisselbrecht from the Bulyk lab, and Alex Tamburino,
Ashlyn Ritter and John Reece-Hoyes from the Walhout lab, for
helpful comments in the preparation of this chapter. We are grateful to
Trevor Siggers for creating the quantitative GRN display item shown
in Figure 4.2 D. Research in the Bulyk and Walhout labs is funded by
grants from the National Institutes of Health [10] .
REFERENCES
[1] Crick F. Central dogma of molecular biology. Nature 1970;
227:561 e 3.
[2] Halder G, Callaerts P, Gehring WJ. Induction of ectopic eyes by
targeted expression of the eyeless gene in Drosophila. Science
1995;267:1788 e 92.
[3] Fukushige T, Hawkins MG, McGhee JD. The GATA-factor elt-2
is essential for formation of the Caenorhabditis elegans intestine.
Dev Biol 1998;198:286 e 302.
Search WWH ::




Custom Search