Biology Reference
In-Depth Information
these cells, researchers were at last able to ask systems-
level questions, and specific network features could be
identified. Neurons of the same functional class tended to
make gap junctions with other members of the same class,
but chemical synapses tend to be between classes. Patterns
of triangular connectivity between neurons are highly over-
represented, the circuitry of the nerve ring is highly
directional (unlike the general organization in vertebrates
for example), and processing depth is shallow
proteins or instead include the ever-increasing number of
non-coding transcribed regions of the genome, predicted
gene numbers can vary widely. But compared to gene
number, it is once one enters the realm of gene function that
things begin to become complicated.
A question such as 'What are the functions of all of the
genes encoded in the C. elegans genome?' appears super-
ficially simple. We immediately reach conceptually (for
example) to analogies of cars, with parts lists and amateur
mechanic exploded diagrams of gear boxes and carbure-
tors, each engine 'module' made up of small number of
intimately connected components and the modules linked
together into units of increasing complexity until we can
view the engine as a single unit of function. Unfortunately,
gene function is seldom viewed in such conceptually
simple terms, and it is very rare that two researchers agree
on what a gene 'does'. For example, the function of the key
cellular oncogene c-myc could alternately (and equally
accurately) be described as a transcription factor of the
basic helix
the
number of connections between stimulus and output is very
small. Ultimately, this systems-level view of the nervous
system provides both a map to guide hypothesis-driven
experiments (e.g., in experiments monitoring the real-time
visualization of neuron firing following stimulation) and
a rational framework for interpreting the outcomes of such
studies. By layering newer experimental data onto this
complete network of the neuromuscular system,
researchers now understand phenomena such as the short-
and long-term modulation of responses to odors
[24]
, and
the physical basis for pheromone attraction and social
behavior in worms
[25]
.
The 'first components, then connections, then system-
level properties' approach has thus been central to C. ele-
gans research from its inception. The lineage, the nervous
system structure, the genome sequence are all complete,
permanent, and accurate. For the remainder of this chapter
we will examine how genetic approaches to understanding
how genotype dictates phenotype have been combined with
gene expression analyses and physical interaction mapping
to ultimately generate integrated models for gene function
in vivo. Finally, we end with a brief section regarding future
directions in the field.
e
helix leucine zipper class; as an onco-
gene whose over-expression leads to cancer; as a dimer-
ization partner of the transcription factor MAX; as an
activator of apoptosis and cell proliferation; or as a gene
that is essential for murine embryonic development
(reviewed in
[27]
). Clearly, attempts have been made to
rationalize this kind of mish-mash of molecular, cellular,
biochemical, and organismal function: the Gene Ontology
project
[28]
is the most widely known and used, and
organizes gene function into hierarchies within three broad
domains, cellular component, biological process, and
molecular function.
From the point of view of genetics, however, the
simplest functional question to ask of a gene is 'What are
the phenotypic consequences of inherited variation in the
sequence of that gene?', and this is where we will focus in
this section. We first lay out briefly the progress made
through classical genetic screens in which mutagens have
been used to generate genetic variation leading to mutant
phenotypes, then examine how knowledge of the genome
sequence, along with the recent advances in sequencing
power, has greatly accelerated these approaches, opening
the door to more sensitive and higher-coverage classical
genetic screens.
Classical ('forward') genetics experiments all investi-
gate the connection between genotype and phenotype in
one direction only: mutants are first identified on the basis
of phenotypic differences from the wild-type state, and the
causative sequence variations are subsequently identified
either through classical genetics mapping strategies or
(more recently) through brute force sequencing. In essence,
for any single mutant the question being asked is 'This
worm looks mutant: what is the sequence change respon-
sible for this?'. Identifying multiple independent mutants
that have similar mutant phenotypes, widens the question to
e
loop
e
FORWARD AND REVERSE GENETICS IN
THE WORM: HOW 97MB SAYS 'MAKE
A WORM'
From 'How Does This Work?' to 'What Does
This Do?'
The world beyond the concrete truths of genome sequence
and cell lineage is comparatively confusing and ill-defined.
The genome sequence of the N2 Bristol isolate is indeed
a platform for investigating biology
[26]
; however, whereas
the question 'What is the complete genome sequence of the
N2 Bristol isolate?' has a unique and well-defined answer,
the questions that follow become increasingly complex and
poorly constructed. 'What are the genes encoded in the
C. elegans genome?' sounds beguilingly simple, yet leads
rapidly into the territory usually occupied by first-year
genetics essays on definitions of what makes up a gene.
Depending on the gene prediction method, and on whether
one defines genic units as being only those encoding
