Imperial College of Science Technology and Medicine,
Department of Biology, Sir Alexander Fleming Building, London
SW7 2AZ, UK. (r.foster@ic.ac.uk
r.j.lucas@ic.ac.uk)
The circadian system allows the timing of cellular and behavioural functions
to be optimized to specific phases of the 24-hour day. Unlike 'driven' physiology
(which is directly controlled by exogenous signals), circadian rhythms are
regulated by an endogenous oscillator which itself is set (entrained) by environmental
signals (zeitgebers) such as light and temperature. The circadian system
has been traditionally represented by three distinct 'black boxes', each representing
a clearly defined task1: environmental entrainment, rhythm generation
using a circadian oscillator or clock, and rhythmic control of effector pathways
(Fig. 1a). This representation is recognized
as an oversimplification, but has nonetheless provided a useful conceptual
framework for defining the function of the cellular and sub-cellular components
of circadian organization. On the basis of this model, abnormal circadian
period or arrhythmic phenotypes have been used as the primary criteria for
identifying clock gene mutants, leading to spectacular advances in our understanding
of how circadian rhythms are generated2. So far, circadian oscillators
in a wide variety of species appear to be based upon autoregulatory transcription/translation
feedback loops using a core set of dedicated 'clock' genes. But despite the
description of feedback loops in the genetic model systems of Neurospora
crassa, Drosophila melanogaster and the mouse, the ways in which endogenous
24-hour rhythms are generated in these organisms are undefined. This is because
additional oscillator components remain to be discovered, and because our
understanding of how the identified genes interact to generate circadian oscillation
is incomplete 2.
Figure 1. Circadian organization has long been conceptualized as comprising three
distinct tasks, represented by 'black boxes' corresponding to the transduction
of environmental inputs, a rhythm-generating circadian oscillator and output
pathways to effector systems (a).
Fitting the molecular components of circadian organization into the simple
model, however, may be assisted by recognizing the possibility that components
can fulfil multiple functions (b). Traditionally, genes have been
assigned roles in either the input, oscillator or output compartments if null
mutations abolish entrainment (1), cause complete arrhythmicity (2) or abolish
output rhythms (3), respectively (a). The latest work with frq9
suggests that arrhythmicity may also be observed following the loss of
an oscillating light input component and suggests the use of T-cycle experiments
(4) to search for additional oscillatory systems (b).
Perhaps we need to break away from the compartmentalized view of the circadian
system to further our understanding of how clock genes work. For example,
in some mathematical models, an arrhythmic phenotype can result from under-
or overexpression of a gene that is either a rhythmically expressed component
of the input pathway or a part of the central oscillator3. The
conceptual limitations inherent in the traditional representation of the circadian
system have been recently confronted in the pages of Nature by Martha
Merrow and colleagues, through their study of the frequency gene (
frq) of N. crassa (Fig. 1; ref. 4).
The elegant Neurospora crassa.(Figure kindly provided by
Martha Merrow and Margit Gorl).
Circadian rhythms in N. crassa can be monitored through the rhythmic
pattern of asexual spore production (conidiation) of cultures growing in glass
tubesas originally demonstrated by the pioneering experiments of Jerry
Feldman5. The subsequent development of this species into a
molecular model was accompanied by the demonstration that frq fulfilled
all the original criteria of a clock component: null mutants of frq
show random (arrhythmic) conidiation under constant lighting conditions;
frq is self-regulated by negative feedback; and experimental induction
of frq expression re-sets the circadian phase of conidiation. Thus,
frq has been represented as the dominant component of the circadian oscillator
of N. crassa2.
The fact that circadian conidiation can occur in a frq null mutant
(frq9) under special circumstances6, however, suggests
that frq is not the sole oscillatory component of the circadian system
in N. crassa. Taking note of this observation, Merrow and colleagues4 employed traditional circadian entrainment protocols in an attempt
to clarify the role of frq within the circadian system. These protocols
involve systematic variation in the period of light or temperature cycles
(T-cycle experiments). Classical circadian experiments indicate that these
protocols can distinguish environmental signals that entrain circadian rhythms
from those that merely drive a response7. For example, under
entrainment, conidiation in N. crassa should occur after the
temperature step in T-cycles of periods shorter than that of the endogenous
clock (for instance, a temperature cycle of 9 hours at 22 °C followed
by 9 hours at 27 °C), but before the temperature step in T-cycles
longer than this. Merrow et al. found that rhythms in conidiation of
both frq-sufficient strains and a frq null mutant (frq9)
showed genuine entrainment to temperature T-cycles. The relationship (slope)
between the period of the T-cycle and the phase of conidiation in all strains
tested (including frq9) was the same, providing overwhelming evidence
that a temperature-entrainable oscillator persists, even in the absence of
the frq protein (FRQ). Experiments using light T-cycles also generated
unexpected results. Unlike temperature T-cycles, these were found to drivebut
not entrainconidiation in frq-sufficient strains of N. crassa
; conidiation occurred at a fixed length of time after the light/dark
transition, irrespective of the length of the T-cycle. As conidiation of the
frq9 mutant is arrhythmic under full light/dark cycle, Merrow et al.
concluded that light must drive conidiation via FRQ.
A number of questions about the organization of the N. crassa circadian
system are underscored by these studies. For example, what is the nature of
the temperature-entrained oscillator, and what is its relationship with the
frq negative feedback loop? We know that whatever constitutes the temperature-entrainable
oscillator, it must be closely associated with the control of oscillation
in frq/FRQ, because FRQ is required to produce both a sustained oscillation
and a rhythm within the circadian range (close to 24 hours). One might also
ask how FRQ mediates the driven effects of light on conidiation, and how temperature
and light interact to generate the circadian rhythm of conidiation in the
natural environment. As temperature acts as a zeitgeber for the entrainment
of circadian rhythms and FRQ mediates the acute effects of light on conidiation,
it seems likely that the temperature-entrainable oscillator gates the effects
of light on conidiation by adjusting the phase of FRQ expression.
So how should the 'frq9' results alter our approach to the assignment
of gene function within the circadian system? The use of T-cycle experiments
by Merrow et al. has shown that these classical circadian protocols
have an important role in the molecular dissection of the circadian clock.
By demonstrating residual rhythmicity, T-cycle approaches have the potential
to take the functional examination of null mutants beyond an observation of
arrhythmicity, and identify the presence of additional oscillatory components.
These approaches have re-defined the role of the frq gene in N.
crassa, and reinforce the view that clock components should be defined
on the basis of a range of criteria, which should routinely include the results
from T-cycle experiments (Fig. 1b).
The work of Merrow et al. is one of several recent reports suggesting
that, while the tasks of environmental entrainment and rhythm generation are
conceptually distinct, single genes may in fact contribute to both. Such dual
roles have been suggested for the mammalian cryptochromes, CRY1 (8) and CRYs (9),
in addition to FRQ. To investigate this possibility, we suggest adopting criteria
based on separating the tasks of the light input pathway into light detection,
signal transduction and oscillator response (Fig. 2).
While Frq meets criteria of both light-input and oscillator components,
it is impossible to confidently assign the mammalian cryptochromes to any
part of this pathway10. Their genetic ablation results in behavioural
arrhythmicity, indicating some involvement within the circadian system, perhaps
as oscillator components (Fig. 1b). While the
cryptochromes are presumed to absorb light, there is as yet no indication
that this alters their activity in vivo or in vitro, and their
absorbance spectra do not match action spectra (Fig. 2).
Figure 2. A simple conceptualization of the circadian light entrainment pathway,
in which the first stage is the absorption of a photon by a light-sensitive
pigment.
The change induced by the photon is then transmitted by a signal transduction
process to induce a phase adjustment of the circadian oscillator. The components
of this system can therefore be defined using the following criteria.
Photopigment: components that absorb light, showing a consequent change
in their activity and have an absorption spectrum matching the action spectrum
for circadian phase shifts; genetic ablation of these reduces circadian photosensitivity,
alters the action spectrum in a predictable manner or abolishes photoentrainment.
Transduction pathway: non-light absorbing components, genetic ablation
of which abolishes photoentrainment or reduces circadian photosensitivity.
Oscillator component: for criteria, see Fig. 1.
Because the original assignment of gene function so profoundly influences
the interpretation of subsequent experimental results, it is critical that
the criteria used to assign a role are both appropriate and sufficient. The
work of Merrow and colleagues enforces two important aspects of this approach:
the power of using strict criteria when applied to a simple model, and the
constant revision and augmentation of both model and criteria on the basis
of unexpected results. By establishing that frq contributes to two
different tasks (light input and oscillator) and that a temperature-entrainable
'sloppy' oscillator exists in the absence of frq, the model of the
N. crassa circadian system has to be revised and additional oscillator
genes sought. The rate at which new 'clock' genes are being found in a whole
variety of organisms leads one to wonder how much of the genome will eventually
turn out to encode components of the circadian system.