Insects display an impressive variety of daily rhythms, which are most evident in their behaviour. Circadian timekeeping systems that generate these daily rhythms of physiology and behaviour all involve three interacting elements: the timekeeper itself (i.e. the clock), inputs to the clock through which it entrains and otherwise responds to environmental cues such as light and temperature, and outputs from the clock through which it imposes daily rhythms on various physiological and behavioural parameters. In insects, as in other animals, cellular clocks are embodied in clock neurons capable of sustained autonomous circadian rhythmicity, and those clock neurons are organized into clock circuits. Drosophila flies spend their entire lives in small areas near the ground, and use their circadian brain clock to regulate daily rhythms of rest and activity, so as to organize their behaviour appropriately to the daily rhythms of their local environment. Migratory locusts and butterflies, on the other hand, spend substantial portions of their lives high up in the air migrating long distances (sometimes thousands of miles) and use their circadian brain clocks to provide time-compensation to their sun-compass navigational systems. Interestingly, however, there appear to be substantial similarities in the cellular and network mechanisms that underlie circadian outputs in all insects.
Insects display an impressive variety of daily rhythms, which are most evident in their behaviour. Honey bees visit flowers at certain times of the day, mosquitoes search for blood at dawn and dusk, blood-sucking bugs do so at night, moths fly and mate at night, and at the same time fireflies use bioluminescent signals and male crickets stridulate to attract the opposite sex. These examples show that feeding and mating are the most conspicuous rhythmic activities of insects, but there are other less obvious rhythms. Some rhythms are linked to reproduction, such as pheromone release, spermatophore production, sperm mobility and oviposition; others are associated with metabolism, such as rhythms in respiration, endocrine functions, body colour and cuticle deposition; still others are related to the sensitivity of sensory organs, such as the eyes and antennae.
Circadian timekeeping systems that generate these daily rhythms of physiology and behaviour all involve three interacting elements: the timekeeper itself (i.e. the clock), inputs to the clock through which it entrains and otherwise responds to environmental cues such as light and temperature, and outputs from the clock through which it imposes daily rhythms on various physiological and behavioural parameters. In all organisms that have been studied, the fundamental biological unit of circadian timekeeping is the cell. However, in multicellular organisms, cellular clocks communicate with one another to generate coherent rhythmicity within a multi-oscillator system. Thus the question of the mechanisms of clock inputs, outputs and timekeeping itself can be asked at both the cellular level and the level of multi-oscillator systems.
In insects, as in other animals, cellular clocks are embodied in clock neurons capable of sustained autonomous circadian rhythmicity, and those clock neurons are organized into clock circuits. Mechanisms of intercellular communication of cellular phase within neuronal clock circuits require that the cellular timekeeping mechanism be able to communicate its phase to other clock neurons through modulation of intercellular signals, such as classical neurotransmitters, neuromodulators, and neuropeptides, and also be able to respond through cell-surface receptors and downstream signalling pathways to the signals it receives. In the present review, we discuss the output mechanisms of insect circadian clocks at both the cellular and multicellular network levels, focusing first on a detailed analysis of circadian outputs in the vinegar fly Drosophila melanogaster, and then turning to a broader survey of other insects.
Timekeeping mechanisms in
Input and core oscillator mechanisms
Understanding circadian output in adult Drosophila appears deceptively simple at first glance. Starting with the work of Konopka and Benzer , the core molecular components of the Drosophila circadian clock have been identified [1–6] (reviewed in [7–9]). Further work identified the compelling and clear relationship between the timing of circadian clock cycling and rhythmic free-running behaviour (reviewed in ). From the very beginning, it was clear that null mutations in core clock genes such as per (period) or tim (timeless) lead to arrhythmic free-running behaviour, and other mutations were identified that cause period shortening or lengthening . From the identification of circadian clock components in Drosophila, in parallel with discoveries in elucidating clock components in other organisms, including mammals, a general framework was developed that cell-autonomous TTFLs (transcriptional–translational feedback loops) are the fundamental units of circadian timekeeping. In Drosophila, CLK (CLOCK) and CYC (CYCLE) act heterodimerically as transcriptional activators for per and tim transcripts in the positive arm of the core circadian TTFL, whereas PER and TIM transiently inhibit their own transcription by binding to CLK and CYC (reviewed in ). Phosphorylation signals from protein kinases such as DBT (DOUBLETIME), CK2 (CASEIN KINASE 2) and SGG (SHAGGY, a glycogen synthase kinase 3 homologue) provide additional levels of tuning of TTFL-based clocks by altering the stability and/or subcellular localization of clock proteins (reviewed in ). For example, after PER and TIM have each been phosphorylated by CK2 and SGG respectively, a PER–TIM–DBT heteromultimer appears to enter the nucleus around midnight, thus effecting transcriptional repression of the negative arm of the clock.
Light provides the primary environmental cue for daily calibration and entrainment of the adult Drosophila core oscillator, along with temperature, food and social interactions. Opsin-based photoreceptor input from the eye (the ocelli and the Hofbauer-Büchner eyelet) and the flavin-based blue-light receptor CRY (CRYPTOCHROME) expressed in subsets of the circadian neurons both mediate light entrainment. Although considerable mechanistic progress has been made on CRY-mediated entrainment, we still know very little about how opsin-based classical photoreceptors effect entrainment processes. In summary, at present there is a relatively coherent picture of the molecular events that underlie the near-24 h periodicity of the circadian clock. However, as may be expected, certain details are controversial.
Genomics of circadian outputs
Given that the circadian clock controls a large number of biological processes that operate in parallel, many groups measured the circadian-regulated cyclic expression of genes in the Drosophila genome using new (at the time) oligonucleotide microarray technology; however, there is remarkably poor consensus between these studies (reviewed in ). Approx. 1% of cycling genes were co-identified between the multiple studies and a number of known core clock genes previously shown to cycle were not identified in some of the individual studies. Furthermore, the number of new circadian genes verified by follow-up analysis has been disappointing, considering that hundreds of cycling transcripts were collectively identified in the initial studies. Different statistical tools, tissues and experimental paradigms are the most probable contributors to the heterogeneity of results in the Drosophila circadian genomics studies.
Mapping oscillators and behaviour
In spite of the impressive set of discoveries elucidating the cellular oscillator, we still lack a clear understanding of the association of the pattern of clock cycling in diverse cells of the circadian circuit with behavioural outputs (i.e. a level of understanding that would allow us to say ‘the activity in these cells is driving this particular behaviour now'). Adult Drosophila exhibit a clear biphasic daily locomotor behavioural rhythm with activity peaks in the morning and evening in LD (light–dark) environmental conditions. Starting in the early 1990s, studies began to reveal the circadian neural circuit in Drosophila [10–12]. These anatomical approaches augured a shift to a more nuanced view of how clock-cycling couples to circadian physiological outputs, from a cell-based focus to a circuit-based outlook. These anatomical studies led to the correct prediction that the neuropeptide PDF (pigment-dispersing factor), expressed solely by the LNv (ventrolateral neuron) anatomical subset of clock neurons, is a central circadian network modulator [11,13]. Some of the first immunocytochemical investigations of brain sections using anti-PER antibodies showed a clear temporal pattern of PER cycling in neurons, photoreceptors and glia, which suggested that the phase of cycling between these cell types is similar, with the peak of PER accumulation between late night and early morning and the cycling trough in late daytime. It is worth noting that oscillators distributed throughout the fly circadian circuit oscillate largely in phase and somehow code for biphasic locomotor behaviour with distinct morning and evening peaks in LD conditions.
Cellular models of biphasic locomotor output
Several models of oscillator co-ordination and output of the Drosophila circadian circuit have been recently suggested. In an attempt to explain the distinct biphasic dawn and dusk locomotor activity peaks exhibited by Drosophila (and, by analogy, other crepuscular animals), an M and E (‘morning and evening') oscillator model posits that individual M and E oscillators are anatomically attributable to two distinct groups of circadian neurons referred to as ‘morning cells' and ‘evening cells' [14,15]. However, the empirical evidence for anatomically restricted neural substrates for the oscillators that regulate the M and E anticipatory behaviour in Drosophila is equivocal, and it appears that many cell groups may jointly regulate the activity–rest rhythm whether in LD, LL (constant light) or DD (constant dark) [15a–15c] (see also [7,8]). Thus the clock circuit in flies appears to act as a distributed network rather than one with ‘labelled lines' of anatomically fixed oscillators (reviewed in ).
PDF as an LNv output signal
Studies in other insects indicated that PDF functions to co-ordinate oscillator phase and entrainment throughout the circadian circuit . This was shown by the behavioural response of cockroaches following brain injections of PDH (pigment-dispersing hormone, a peptide functional homologue of PDF). This general idea has held up remarkably well, as shown by a long series of molecular-genetic experiments using Drosophila [8,16a,16b]. The main function of PDF appears to be the coupling of molecular oscillations in different clock neurons through intercellular communication of phase information, based on studies of genetic manipulation of PDF-expressing cells and studies of flies with the pdf 01 null mutation (reviewed in ). Recent work indicates complex functions of PDF in addition to synchronization. For example, it appears that, independent of a role in the intercellular communication of clock phase, PDF also functions to gate the outputs of non-LNv clock neurons.
Timekeeping mechanisms in other insects
Physiological and behavioural rhythms
In addition to the daily rhythms of rest and activity reviewed above in detail for D. melanogaster, insects use their clocks for seasonal adaptations, time memory and navigation (sun-compass orientation). In this section, we highlight selected examples of clock outputs in adult insects that cannot be studied in D. melanogaster. Some examples are rather spectacular, such as sky-compass navigation of locusts (Schistocerca gregaria) and monarch butterflies (Danaus plexippus). Other examples concern hormonal outputs in which the anatomical clock pathways are revealed in detail. Before we come to specific outputs, we will briefly describe the general molecular and neural basis of insect clocks.
Molecular and neural basis of insect clocks
As for D. melanogaster, endogenous clocks of other insects are located in the brain and/or peripheral organs. Brain clocks control mostly rhythmic behaviour, whereas rhythms in cuticle deposition, body colour, spermatophore production, sperm mobility and ecdysteroid synthesis are preferentially controlled by peripheral clocks . The core clock molecules (PER, TIM, CLK, CYC and CRY) in central and peripheral clocks are basically conserved across all insect species, but significant differences occur with respect to the composition of the clock molecules utilized and their exact roles in the clockwork . Brain clocks use neuropeptides preferentially as output transmitters and, among these, PDF may play a central role in many insects in addition to Drosophila. Immunocytochemical studies revealed distinct PER-expressing and PDF-expressing neurons in the brains of many different insects (Figure 1). Similar to Drosophila, most insects have PER- and PDF-positive neurons in the lateral protocerebrum and PER-positive neurons in the dorsal protocerebrum. However, in contrast with Drosophila, the PDF-positive neurons in the lateral protocerebrum are not identical with the PER-positive neurons, and, unlike Drosophila, PER does not enter the nucleus at any time during the day in a great majority of insect species . The PDF neurons in the lateral brain arborize in a small neuropil at the base of the optic lobe – the aMe (accessory medulla) – and this neuropil has been shown to contain the master clock in cockroaches, beetles and flies . In Lepidoptera (moths and monarch butterflies), the clock neurons are exclusively situated in the dorsal protocerebrum, suggesting that these insects have a main clock in the dorsal brain, a view that is consistent with the legendary transplantation experiments of Truman and Riddiford  in silk moths.
Simplified model of an insect brain illustrating its most important structures (A), and putative location of the circadian clock neurons in the brain of select insects (B)
Putative output pathways of the brain clock
Clock neurons in the lateral brain are in a perfect position to control rhythms in the optic lobes and eyes, and this has been demonstrated for rhythms in the ERG (electroretinogram) in cockroaches, crickets and beetles (reviewed in ). The wide arborizations of the PDF neurons towards the eye appear most suited to control such rhythms, and the involvement of PDF in the sensitivity of the compound eye has been reported for crustaceans and at least for a few insects .
Most importantly, PDF fibres from the lateral clock neurons additionally project into the dorsal protocerebrum, where they putatively make contact with dorsal PER neurons. Furthermore, PDF fibres of all insects studied to date are close to the calyces of the mushroom bodies, the insect's correlates of olfactory memory. The vicinity of PDF fibres and mushroom bodies is especially striking in the honey bee Apis mellifera, which is famous for its marvelous time memory . Flowers open at different times of the day, and von Frisch  found that bees can remember up to nine such times for their visits. Although time input from the clock into the mushroom body is plausible, a functional connection between PDF neurons or other clock neurons and mushroom bodies has not yet been demonstrated.
Clock neurons (from the lateral or dorsal protocerebrum) may furthermore establish contact to the insect endocrine system in the pars intercerebralis and lateralis and thus can time hormonal rhythms of the adult insects. This is especially probable in moths, where the main clock is located in the dorsal brain and where the females of many species release pheromones rhythmically during the night. In turnip moths (Agrotis segetum), the pheromone glands of females contain the highest amount of sex pheromone during the mid-night and this corresponds well with the peak of male pheromone response . Pheromone synthesis is stimulated by the PBAN (pheromone biosynthesis-activating neuropeptide), which is produced in neurosecretory cells of the pars intercerebralis and the suboesophageal ganglion, transported into the corpora cardiaca and then released rhythmically into the haemolymph . Although no double-staining with antibodies against PER and PBAN has been performed, it is likely that fibres of the clock neurons overlap those of the PBAN-positive cells in the dorsal protocerebrum.
Seasonal timing: the photoperiodic clock of
Approximately two-thirds of the world's land mass lies within temperate and polar regions of the earth. Within these regions, the majority of insects escape the exigencies of winter through dormancy or migration, which is initiated by a physiological response to day length called photoperiodism. Many insects have evolved a strategy called diapause – a stage of developmental arrest – that enables them to survive harsh seasonal conditions. Diapause can occur at different stages of the life cycle: in the embryo, the larvae, the pupa (in holometabolic insects) or the adult (which equates to reproductive dormancy).
The photoperiodic response encompasses a photoreceptor for measuring light input, a timer as a reliable reference to measure day (or night) length, and an output signal to the hormonal control of diapause. The synchronized circadian clock could serve as a perfect time reference, but it is not yet clear whether it does so in all species . Similarly, the molecular mechanisms of photoperiodism and diapause programming are still under debate. One timely question is whether the genes that make up the circadian clock are also involved in measuring day length and programming insects for successful overwintering. There exists evidence both for and against a close relationship between circadian and seasonal clock systems .
The blowfly P.terraenovae belongs to the few species in which the clock neurons in the lateral brain seem to be clearly necessary for the photoperiodic response and where the pathway from the clock neurons to pars lateralis neurons have been largely revealed . As do many Diptera, including D. melanogaster, Culex pipiens and Aedes aegypti, female P. terraenovae flies undergo reproductive dormancy (ovarian diapause), which is associated with an increased lifespan and a cessation of vitellogenesis (uptake of yolk protein by the oocyte). Reproductive dormancy depends on the concerted action of insulin-like peptides produced by neurons in the pars intercerebralis: juvenile hormone from the ring gland (corpus allatum) and ecdysteroids from the ovaries (reviewed in ; for differential gene-expression analysis in summer and migratory monarch butterflies, see ). In P. terraenovae, neurosecretory cells of the pars lateralis innervating the ring gland have been shown to be important for photoperiodic diapause . Most interestingly, the pars lateralis neurons receive inputs from the PER- and PDF-containing clock neurons in the lateral brain via regular synaptic contacts, and surgical ablation of these clock neurons renders the flies unable to distinguish between long and short days . These results suggest that a direct neural connection between lateral clock neurons and pars lateralis neurons is involved in the photoperiodic control of P. terraenovae.
Time-compensated sky-compass orientation in locusts and monarch butterflies
Several insects show remarkable capabilities in spatial orientation and navigation. Seasonally migrating monarch butterflies migrate from summer habitats in the U.S.A. up to 4000 km to overwintering places in central Mexico . Adult desert locusts migrate in swarms in response to high density of individuals. Swarms usually maintain constant compass directions over several hours, despite changing wind conditions. Over shorter ranges of distance, honey bees and desert ants (Cataglyphis bicolor) navigate in straight lines back to their nest after complicated feeding excursions [33,34]. During their trips, all of these insects rely on a vector-based mechanism of orientation. They continually monitor direction of travel using a sun- or a polarized-light compass, called a sky compass (reviewed in ).
Animals using sky-compass orientation face a serious problem. As the solar azimuth changes over the course of the day, a constant migratory direction can only be maintained with permanent adjustment of the angle between the navigational vector and the solar azimuth. Thus the navigation compass must be permanently adjusted by an endogenous clock. The neuronal substrates of sky-compass orientation have been revealed in the desert locust [36–40], whereas the necessity of clock input for sky-compass navigation has been demonstrated in the monarch butterfly [41–44].
Neuronal substrates of sky-compass orientation in desert locusts
As in other insects, specialized photoreceptor cells in a dorsal-rim area of the compound eye sense polarized light (Figure 2). These photoreceptor cells project to specialized areas in the optic lobe and from there the signal is passed via synapses in defined midbrain neuropils to the central complex. The central complex is a higher-order brain area for spatial orientation that integrates visual input with locomotor output. It consists of four interconnected neuropils: the protocerebral bridge, the upper division of the central body (termed the fan-shaped body in D. melanogaster), the lower division of the central body (termed the ellipsoid body in D. melanogaster) and a pair of ventral noduli (reviewed in ). The central complex is of extraordinary regularity in neuroarchitecture, and is composed of vertical columns and horizontal layers. The lower division of the central body receives visual inputs including that of polarized light from the dorsal-rim area of the compound eye. From there, the polarized-light information is carried to the protocerebral bridge through specific columnar neurons called CL1a neurons . In the columns of the protocerebral bridge, the celestial E-vectors (electric vectors) of polarized light are presented as a topographic map . The protocerebral bridge is also the place where input from the circadian clock may occur. Neurons from the aMe project through the posterior tubercle to the protocerebral bridge . Although the master clock of locusts has not yet been revealed, the aMe of the locust closely resembles that of cockroaches with respect to anatomy, neurochemistry and electrophysiological properties, suggesting that it similarly comprises the circadian timing centre. A recent study shows that the aMe is also connected with the dorsal-rim area of the compound eye and that time-compensated polarized-light signals may reach the protocerebral bridge through the aMe . From the protocerebral bridge, the putatively integrated information about the polarization pattern of the sky and the time of day is passed down (through other specific columnar neurons termed CL1b–d) to the upper division of the central body and to the lateral accessory lobes that are the major targets of central-complex outputs . The lateral accessory lobes communicate with the ventral nerve cord and have a role in directional orientation (for details, see ). Thus the information about the polarization pattern of the sky, the actual position of the insect in space and the information about time of day may be processed in the central complex of the locust, enabling the animal to navigate to its desired destination.
Sensory and neuronal basis of time-compensated sky-compass orientation in the desert locust
The importance of the clock for sky-compass navigation in monarch butterflies
Monarch butterflies have similar polarized-light-sensitive photoreceptors in the dorsal-rim area of the compound eyes to locusts, but their neural connection to the brain seems different : CRY-positive neuronal fibres seem to connect the dorsal-rim area directly to the clock neurons in the dorsal protocerebrum (in the pars lateralis and intercerebralis). There are also CRY-positive fibres in the lower and upper part of the central body , but their relation to the photoreceptors and the clock is unclear. Nevertheless, in the monarch butterfly the importance of the clock for proper sky-compass orientation has been demonstrated. When the circadian clock is disrupted by environmental manipulations, such as constant bright light or phase-shifting the LD cycle, the monarchs become lost . Recently, it has been shown that monarchs possess antennal clocks in addition to the brain clocks, and that the antennal clocks may provide the primary timing mechanism for sun-compass orientation: when the antennae were painted black, the butterflies lost their orientation in space . It is not known whether these antennal clocks provide input into the central complex of monarchs, but these unexpected findings certainly open a new line of investigation into clock–compass connections.
Circadian rhythms play important biological roles for insects as diverse as flies, locusts and butterflies. Drosophila flies spend their entire lives in small areas near the ground, and use their circadian brain clock to regulate daily rhythms of rest and activity, so as to appropriately organize their behaviour to the daily rhythms of their local environment. Migratory locusts and butterflies, on the other hand, spend substantial portions of their lives high up in the air migrating long distances (sometimes thousands of miles), and use their circadian brain clocks to provide time compensation for their sun-compass navigational systems. Interestingly, however, there appear to be substantial similarities in the cellular and network mechanisms that underlie circadian outputs in all insects. Further extensive fascinating work remains to be performed to address the specific cellular and molecular mechanisms of circadian outputs in these diverse insects.
Insects possess internal circadian timing systems composed of a network of clock-gene-expressing neurons in the central nervous system.
Neuropeptide and classical neurotransmitter systems provide important mechanisms for intercellular communication within these neural networks.
Insect circadian neural networks are highly plastic and can alter their functional properties in response to environmental conditions.
Drosophila flies live near the ground and use their circadian timimg system to organize their daily activity around dawn and dusk.
Locusts and butterflies migrate extremely long distances through the atmosphere and use their circadian timing system as a component of sun-compass navigation.
Work in the laboratory of C.H.-F. is funded in part by the European Community (6th European Framework, Project EUCLOCK, no. 018741). Work in the laboratory of T.C.H. is funded in part by the National Institutes of Health, National Institute of Neurological Disorders and Stroke (NINDS) (R01NS046750). Work in the laboratory of M.N.N. is funded in part by NINDS (R01NS056443, R01NS055035 and R21NS058330).