Biology Reference
In-Depth Information
great majority of genes are required for wild-type fitness
and thus have detectable RNAi phenotypes (unlike had
been surmised from previous screens) and, furthermore,
that there is a clear, albeit weak, correlation between the
level of requirement of a gene for wild-type fitness and the
level of negative selection acting on that gene as measured
by ka/ks. The systematic and unbiased datasets deriving
from RNAi screens in the worm thus allow one to confirm
at last some of the basic predictions of molecular evolution
and natural selection.
Finally, one can use systematic RNAi data to examine
large-scale functional organization of chromosomes. At
one extreme, one might imagine the organization of genes
on chromosomes to be entirely random and genes with
similar phenotypes to therefore be scattered evenly across
the entire genome. The other extreme alternative would be
a situation of high compartmentalization of the genome
in some of the dark spaces that are hard to illuminate with
classical screens
combinatorial genetics, early roles for
maternally expressed genes, and so on. Likewise, classical
genetics provides tools in the form of genetic mutants, but
also can attack areas that RNAi will always miss
e
e
hypermorphic alleles, for example, have been critical in
dissecting signaling pathways (e.g.,
[35,102]
), but RNAi
cannot generate these, nor can it trivially generate loss-of-
function phenotypes in screens for genes whose products
have long half-lives. RNAi and classical genetics are
likely to coexist for a long time to come, each providing
different views of gene function and tools for further
experiments.
The loci defined by classical genetics screens, and their
ordering into genetic pathways by epistasis, provide an
entry point for functional studies. However, without
knowing what the identified genes encode, where and when
they are expressed, and how their protein products interact,
a genetic pathway remains an abstract informational entity.
Going from the relationships between genotype and
phenotype revealed by genetics (whether forward or
reverse) to molecular models of the machineries under-
pinning biology requires many other data sources. In the
following sections we describe systematic approaches to
examine gene expression patterns, gene regulation, and
physical interaction networks in the worm, and then the
computational approaches developed to integrate genetic,
gene expression, physical interaction, and other datasets
into single networks that can in turn provide insight into the
functional modules that coordinate and carry out worm
development.
e
all genes affecting neuromuscular development might
reside in clusters, all genes required for normal vulval
development in other clusters. Some clustering is imme-
diately evident: of the genes with detectable phenotypes in
a series of genome-scale RNAi screens, ~70% of those on
autosomes have non-viable (lethal or sterile) phenotypes,
whereas only ~30% of genes on the X chromosome have
non-viable phenotypes
[66]
. Looking more closely, one can
identify clear clustering of genes with specific phenotypes
in broad chromosomal domains: the autosomal central
clusters are highly enriched for non-viable genes and the
autosomal arms are typically under-enriched for non-viable
genes, whereas the X chromosome is enriched for genes
with more subtle, post-embryonic phenotypes
[108]
. None
of the above questions could have been examined without
the systematic, unbiased datasets provided by RNAi
screens.
Genome-scale RNAi screens have therefore been key
for systems biology in the worm. They have fleshed out
many of the pathways initially identified by classical
genetics screens
EXPRESSING AND REGULATING
AN ANIMAL GENOME
One central challenge in systems biology is to understand
e
the first genome-scale RNAi screens to
be published multiplied the number of genes with identi-
fied in vivo loss-of-function phenotypes by a factor of 3
over everything identified in 30 years of forward genetics,
as an example
[66]
. They have identified the great
majority of genes that orchestrate the events of early
embryogenesis, pinpointing exactly which specific process
they control, and such high-resolution high-coverage
screens have allowed both the prediction of molecular
functions of previously unknown genes and the systems-
level analysis of embryogenesis. Specifically, the unbiased
nature of the screens has also allowed researchers to ask
basic questions about genetics
e
and to be able to predict
how genes are expressed in
a reproducible spatial and temporal pattern. Such a grand
challenge can be tackled at multiple levels, for example by
detailed analysis of the cis and trans regulatory control of
individual genes, by focusing on the regulatory interactions
involved in specifying particular cell lineages or tissue
types, and through the global mapping of gene expression
patterns and regulatory interactions. Moreover, whereas
many studies have focused primarily on understanding
transcriptional control, post-transcriptional processes such
as the regulation of mRNA stability, splicing and translation
must also be considered. Ultimately, a predictive model of
gene regulation during development will also need to
incorporate the signaling interactions between cells and how
these connect to gene expression, as well as biophysical
influences on development, pattern formation and gene
expression.
e
how the molecular
function of a gene relates to its organismal role, how new
functions arise in biology, and how natural selection acts
through perturbed phenotypes. RNAi emerged at a perfect
time in the history of worm research, but it also helps fill
e
