Evolutionary Links Between Circadian Clocks and Photoperiodic Diapause in Insects

Abstract

In this article, we explore links between circadian clocks and the clock involved in photoperiodic regulation of diapause in insects. Classical resonance (Nanda–Hamner) and night interruption (Bünsow) experiments suggest a circadian basis for the diapause response in nearly all insects that have been studied. Neuroanatomical studies reveal physical connections between circadian clock cells and centers controlling the photoperiodic diapause response, and both mutations and knockdown of clock genes with RNA interference (RNAi) point to a connection between the clock genes and photoperiodic induction of diapause. We discuss the challenges of determining whether the clock, as a functioning module, or individual clock genes acting pleiotropically are responsible for the photoperiodic regulation of diapause, and how a stable, central circadian clock could be linked to plastic photoperiodic responses without compromising the clock’s essential functions. Although we still lack an understanding of the exact mechanisms whereby insects measure day/night length, continued classical and neuroanatomical approaches, as well as forward and reverse genetic experiments, are highly complementary and should enable us to decipher the diverse ways in which circadian clocks have been involved in the evolution of photoperiodic induction of diapause in insects. The components of circadian clocks vary among insect species, and diapause appears to have evolved independently numerous times, thus, we anticipate that not all photoperiodic clocks of insects will interact with circadian clocks in the same fashion.

Introduction

The daily rotation of the earth on its axis and the annual revolution of the earth around the sun generate patterns that have profound impacts on daily and seasonal activity patterns of organisms. Feeding, mating, sleeping, and indeed most behavioral and physiological activities show a distinct rhythm with a peak of activity occurring at a certain time of the daily light:dark cycle. A circadian rhythm is defined as one that persists with a periodicity of approximately 24 h, even after the organism is shifted from a light:dark cycle to constant darkness. This feature suggests that the rhythm is endogenous rather than an immediate response to the external light:dark cycle. The other major rhythm is based on the seasonal calendar known as photoperiodism. In this case, the organism responds either to annual changes in day length, or directly to the length of day/night, and uses this information to determine the timing of key events such as migration or the entry into a dormant state. Both circadian and photoperiodic events rely on an ability to precisely measure time. In this article, we discuss evolutionary links between these two timekeeping mechanisms with a special focus on photoperiodic induction of diapause in insects.

Circadian clocks

Circadian clocks correctly time a range of important behavioral and physiological processes in animals including sleep–wake cycles (Czeisler et al. 1986), locomotion patterns (Silver et al. 1996), feeding (Nishio et al. 1979), mating (Sakai and Ishida 2001), and cell division (Matsuo et al. 2003). Given the adaptive importance of circadian clocks (Sharma 2003), it is not surprising such clocks have been documented frequently in organisms ranging from photosensitive cyanobacteria to humans. The key genes involved in animal circadian clocks are highly conserved. For example, there is high homology among core insect and human circadian clock genes, and recent work on circadian clocks in the cnidarian, Nematostella vectensis (Reitzel et al. 2010), reveals a high similarity of the cnidarian clock to both insect and mammalian clocks, thus implying an early evolution of clocks within animals.

Konopka and Benzer (1971) first identified the period gene in Drosophila melanogaster after screening for mutants that were arrhythmic and had altered circadian periods. Since then, additional components of circadian clocks in insects have been identified, and we now have a mechanistic understanding of circadian function in insects. In the circadian clock model for Drosophila (Fig. 1A), the proteins CLOCK (CLK) and CYCLE (CYC) act as positive transcriptional regulators by binding to an E-box promoter on the genes period (per) and timeless (tim) and recruit RNA polymerase, thereby increasing the abundance of per and tim mRNAs. Once in the cytosol, per and tim mRNAs are translated into PER and TIM proteins, which then form a heterodimer that translocates back into the nucleus where the PER component of the heterodimer acts to inhibit action of CLK and CYC, thereby suppressing transcription of itself and tim. Circadian clocks are informed by light through the action of CRYPTOCHROME1 (CRY1) which binds to TIM in the presence of light and degrades both TIM and itself.

Fig. 1

The circadian clock model in Drosophila and other insects. (A) In the nucleus of D. melanogaster, circadian cells CLOCK (CLK) and CYCLE (CYC) proteins form a heterodimer that acts as a transcriptional activator by binding to the E-box promoter region of the period (per) and timeless (tim) genes. per and tim mRNA are translated in the cytoplasm of the cell. PER and TIM proteins then form a heterodimer and translocate back into the nucleus where PER inhibits the action of CLK:CYC, thereby suppressing the transcription of per and tim. The CRYPTOCHROME1 (CRY1) protein degrades TIM and itself in the presence of light. This results in increasing levels of per and tim mRNA throughout the day when CLK:CYC activity is uninhibited, and decreasing levels of per and tim mRNA during the night. (B) The circadian clock in most other insects, such as monarch butterflies, and mosquitoes, differ from the Drosophila clock in that they possess a light-insensitive CRY2 protein that acts as the major negative transcriptional regulator of the core circadian clock. In this case, PER appears to assist CRY2 in nuclear translocation, whereas TIM helps to stabilize PER and CRY2.

For years, it was assumed that circadian clocks in other insects functioned like clocks in Drosophila, but the discovery that the monarch butterfly, Danaus plexippus, has an additional clock protein CRYPTOCHROME2 (CRY2; Zhu et al. 2005) suggested that this may not be the case. Unlike CRY1, CRY2 is light insensitive and acts as a potent repressor of CLK and CYC, much like the mammalian CRYs. Zhu et al. (2005) also identified CRY2 in other Lepidoptera, as well as in the mosquito Anopheles gambiae, and the red flour beetle Tribolium castaneum. CRY2 has since been found in every non-drosophilid insect examined (Yuan et al. 2007), and hence our understanding of circadian clocks in insects has been greatly enhanced. We now know that CRY2, not PER, is likely the important negative regulator of circadian clocks in most, if not all, non-drosophilid insects (Fig. 1B).

Photoperiodism, the seasonal clock

Just as the earth’s rotation about its axis produces a daily pattern of light and dark, the earth’s 23.5° tilt as it rotates about the sun produces an annual pattern of changing day length, temperature, and precipitation. Consequently, some seasons are favorable for growth, development, and reproduction while others are not. Seasonal timing is just as important as daily timing and is essential for enabling animals to correctly anticipate seasonal environmental changes. In temperate and polar environments, where seasonal differences are pronounced, many animals evolved the capacity to measure and respond to changes in day length, or photoperiod. As the same geographic location will experience the same length of day or night on a specific day each year, photoperiod is a much more reliable seasonal signal than changes in temperature or rainfall, which vary greatly over time. Photoperiod is the primary signal that most temperate-zone animals, including insects, use to anticipate seasonal change.

Among insects, one of the most prevalent photoperiodic responses is the onset of diapause. Diapause is a programmed arrested state of development that allows insects and other arthropods to survive adverse seasonal conditions either by becoming dormant locally, or by first migrating to a more favorable environment. Tropical insects use a variety of signals, such as changes in temperature, rainfall, or population density, to initiate diapause, but the most common signal that insects in temperate environments use is photoperiod (Tauber et al. 1986). Although diverse responses to photoperiod have been documented in temperate zones, most commonly short days (long nights) elicit a high incidence of overwintering diapause, whereas long days (short nights) evoke a low incidence of diapause (Fig. 2). The point at which the incidence of diapause is 50% of its maximal level is known as the critical photoperiod (CPP). The CPP among populations of the same species increases with latitude and elevation (Bradshaw and Lounibos 1977; Bradshaw and Holzapfel 1975), thus enabling populations at higher latitudes and those that reside at higher elevations to adjust to the earlier onset of winter by entering diapause at an earlier date (i.e., increase their CPP). The relationship between CPP and latitude/elevation demonstrates one of many characteristics of photoperiodic diapause that is plastic within a species and is acted upon by natural selection.

Fig. 2

Photoperiodic response curve of diapausing pupae of the flesh fly, Sarcophga bullata, from populations in Illinois and Missouri. Each point represents the mean (±SE) incidence of diapause in progeny of 12–14 females (from Denlinger 1972).

Another defining characteristic of diapause in insects is that each species is typically capable of entering diapause in only a single life stage, although diapause can occur in any of the life stages. For example, in the order Diptera, there are examples of photoperiodically-induced embryonic diapause (the Asian tiger mosquito, Aedes albopictus), larval diapause (the pitcher-plant mosquito, Wyeomyia smithii), pupal diapause (the flesh fly, Sarcophaga bullata), and adult reproductive diapause (the northern house mosquito, Culex pipiens). Such differences in stage-specificity of diapause within a single order, and sometimes even within a single genus, suggest that photoperiodic diapause has evolved multiple times.

The ability to enter diapause has contributed to the evolutionary success of insects (Denlinger 2008). Insects in diapause are resistant to a range of environmental stresses, and these responses are mediated through physiological mechanisms, such as the generation of polyols and heat shock proteins that enhance survival at low temperatures; elevation of cuticular hydrocarbons that protect against desiccation; increased lipid stores and suppressed metabolic rates that enable diapausing insects to survive long periods without food; and boosts in immune responses to combat increased attacks by pathogens (Denlinger 2002, 2008; Hahn and Denlinger 2011). The hormonal regulators responsible for initiating and maintaining diapause have also been well-characterized (Denlinger et al. 2012). A failure of the brain and its associated endocrine organs to produce the hormones that trigger molting (larval, nymphal, and pupal diapauses) or maturation of reproductive organs (adult reproductive diapause) is the dominant endocrine mechanism that halts development. Although we know photoperiod is the token stimulus that most temperate insects use to initiate their overwintering diapause and we know a considerable amount about the ultimate endocrine signals that bring about this halt in development, we know little about the mechanisms used by insects to measure day length and to translate this information into the downstream endocrine signal.

Photoperiodic and circadian clock models

Erwin Bünning (1936) hypothesized that plants and animals likely use their circadian clocks to measure day length, and hence initiate photoperiodic responses, because the circadian clock already provides critical information on light/dark cycles. According to Bünning’s hypothesis, also referred to as the external coincidence model, light entrains the circadian clock which sets a light sensitive phase (φ_i) in the late night/early morning, such that if φ_i is illuminated, a long-day response is elicited. This model has been tested extensively in evaluating several photoperiodic responses, including diapause (reviews by Saunders 1981, 2009, 2010b, 2013; Saunders and Bertossa 2011). One protocol for testing this model, Bünsow (1953) or night interruption studies, uses short photoperiods followed by brief pulses of light administered at different times throughout an extended night. If the light falls on φ_i a long-day response is elicited.

Nanda–Hamner resonance experiments have also been used extensively; in this protocol, a short period of light is followed by periods of darkness to create light:dark cycles that range from 24 to 72 h (Nanda and Hamner 1958). In such experiments, animals are expected to show short-day responses when the total period (T) of the light:dark cycle is equal to a multiple of 24 h, and a long-day response when T ≠ n24 h (i.e., T = 36). Over a range of T ’s, insects are expected to show peaks and troughs in the incidence of diapause. Evidence from such experiments supports the hypothesis that photoperiodic responses as diverse as diapause in flesh flies (Saunders 1973), flowering in plants (Somers 2010), and reproduction in Siberian hamsters (Prendergast et al. 2004) have a circadian basis and can be explained using the external coincidence model.

Bradshaw et al. (2003a) demonstrated that there is no relationship between changes in CPP and periodicity of peaks in the incidence of diapause in the pitcher-plant mosquito W. smithii using Nanda–Hamner experiments, but there was a negative correlation between CPP and the amplitude of these peaks. To determine whether this correlation was the result of a causal connection between circadian oscillators and photoperiodic induction of diapause, Bradshaw et al. (2012), using only five rounds of antagonistic selection, were able to reverse their previously observed negative correlation between CPP and amplitude of diapause. They thus present this as evidence that the circadian clock and photoperiodic responses can evolve independently. Although we agree with their ultimate conclusion, their interpretation is predicated on the assumption that amplitude of the diapause response in the Nanda–Hamner experiments has a circadian basis. This may not be the case. Instead, the amplitude of the diapause response may reflect light sensitivity of the circadian oscillator (Tauber 2012), or may be a feature that has no connection to the circadian clock. Although their selection experiments changed both the CPP and amplitude of the diapause response, there was no correlated change in the period between peaks in the incidence of diapause. This indicates that the underlying periodicity of the circadian pacemaker was not affected by their experiments. Therefore, their ultimate conclusion that CPP, a photoperiodic response, can be acted upon by natural selection independently of the clock is supported when one considers the period between peaks, rather than their amplitude, as critical in determining the incidence of diapause. Mechanisms that could achieve this dichotomy are discussed later in this article.

A contrasting mechanism in which circadian clocks could be involved in photoperiodic responses is known as the internal coincidence model. In this model, light entrains separate dawn and dusk oscillators, such that photoperiodic responses result from changing phases between the two oscillators. This model was first proposed by Pittendrigh and Minis (1964) after it became apparent that most multicellular animals have multiple circadian oscillators that may interact with each other in different ways. Saunders (1974) used Nanda–Hamner experiments to determine that the parasitic wasp, Nasonia vitripennis, possesses separate dawn and dusk oscillators, and both oscillators are involved in programming the wasp’s larval diapause.

It is also possible that circadian clocks are not consistently involved in initiating photoperiodic responses. Instead, animals might measure day length or night length through the accumulation of a chemical substance, such that a photoperiodic response is initiated if the substance reaches a critical threshold. Insects and other animals were believed to use such an “hourglass timer” if they failed to show positive responses to Bünsow or Nanda–Hamner experiments. Lees (1966, 1973) suggested that the vetch aphid, Megoura viciae, uses an hourglass-like interval timer to measure night length and initiate diapause, although Vaz Nunes and Hardie (1993) later proposed that M. viciae measured extended periods of darkness repeatedly in a circadian-based manner. Vaz Nunes and Veerman (1982) hypothesized that the spider mite, Tetranychus urticae, which shows strong positive responses to Nanda–Hamner experiments but a negative response to certain skeleton photoperiods and T-experiments, uses an hourglass clock to distinguish long from short nights, but a circadian-based counter to tally the number of long nights. Saunders (2010a) countered that the observations on T. urticae could be explained by the circadian-based external coincidence model by considering the intensity and wavelengths of light used in the T. urticae experiments.

Finally, Emerson et al. (2008) explored whether the photoperiodic counter involved in terminating diapause in the pitcher-plant mosquito, W. smithii, is linked to the circadian clock. The authors did this by exposing diapausing larvae from different populations to 18 h photophases followed by 8, 30, or 54 h of darkness. They reasoned that if an internal coincidence model was responsible for breaking diapause, fewer 8:54 cycles would be required to break diapause than if the mosquitoes were using an hourglass or external coincidence counter. If, however, the mosquitoes required the same number of light:dark cycles to break diapause, this would imply that W. smithii uses an hourglass or external coincidence counter that measures total photoperiod. Among all the populations they examined, the same number of light:dark cycles were required to break diapause, and hence the termination of diapause was considered to be a function of a photoperiodic counter that measured day length. Although the results could not distinguish between an external coincidence model and an hourglass counter, they assumed that W. smithii counted exogenous periods of environmental light, rather than internal circadian periods. Saunders (2013) argues that a circadian basis for counting photoperiods could also be inferred from their experiments.

To date, nearly all classic experiments in plants, vertebrates, and insects suggest a circadian basis for photoperiodic responses. The hourglass model, which assumes that a compound accumulates throughout the day or night independently of circadian clocks, is appealing because of its inherent simplicity, but this model has not been widely supported in studies on the photoperiodic induction of diapause. However, it remains quite possible that a photoperiodic counter, acting either with or downstream of circadian clocks, could work by accumulating a substance in the insect brain that, after reaching its threshold, would initiate diapause. Insects are, of course, incredibly diverse and there is no reason to assume that all species use a single model to measure photoperiodic time.

Neuroanatomical studies connecting the circadian clock to photoperiodic induction of diapause

Photoperiodic time measurement consists of four components: (1) light receptors, (2) a photoperiodic clock that distinguishes long nights/short days from short nights/long days, (3) a photoperiodic counter that accumulates information on successive long nights/short days, and (4) output pathways that generate various photoperiodic phenotypes (Saunders 2002; Koštál 2011). During the induction of photoperiodic diapause in insects, the output pathway of the photoperiodic clock signals the decision to arrest development to neurosecretory cells in the brain and other endocrine organs. Although much is known about the light receptors and the hormonal signals that underlie the photoperiodic diapause phenotype, very little is known about either the clock that measures night length, or the counter mechanism. Given that circadian clock cells and the signaling events that initiate diapause are located within the brain, neuroanatomical studies are a logical place to search for physical connections between circadian clock neurons and the photoperiodic clock underlying diapause induction.

Photoperiodic light receptors have been studied in at least 19 insect species from six different orders, and from these studies, it is apparent that reception of light can occur either retinally and/or through extraretinal receptors located in the brain or stemmata (Goto et al. 2010). The type of photoreceptors that insects use to initiate photoperiodic diapause does not correlate with phylogeny (Numata et al. 1997). For example, both Calliphora vicina and Protophormia terraenovae are flies in the family Calliphoridae, but the former uses the brain as its site of photoreception (Saunders and Cymborowski 1996), whereas the latter uses its compound eyes (Numata et al. 1997; Shiga and Numata 1997). Possibly C. vicina also uses its compound eyes to measure day length and initiate diapause, but this was not tested by Saunders and Cymborowski. Convergence and/or redundancy in the type of photoreceptors used to initiate diapause is a likely consequence of the multiple times that photoperiodic diapause has evolved (Saunders 2012).

Although circadian clocks in insects are present in multiple tissues, the brain-based clock is likely most critical for circadian responses, but interestingly the precise location of clock cells within the brain does not seem to be well-conserved. In D. melanogaster, staining for circadian clock neurons is localized to three groups of lateral neurons and three groups of dorsal neurons in the pars lateralis (PL) (Helfrich-Förster 1995). The circadian clock in P. terraenovae shows a similar arrangement to that of Drosophila (Shiga and Numata 2009), but circadian clock neurons in cockroaches and crickets are localized in the accessory medulla (Petri et al. 1995; Tomioka and Matsumoto 2010). Circadian clock gene expression in the silk moth, Antheraea pernyi, is confined to eight neurosecretory cells in the pars intercerebralis (PI) and PL, as probes for PER protein and per mRNA colocalized in these cells (Sauman and Reppert 1996). Similarly, the central circadian clock in the monarch butterfly, D. plexippus, is localized to four cells in the dorsolateral protocerebrum (PL), while neurosecretory cells in the PI also express PER and CRY1 (Sauman et al. 2005).

Hormones that control diapause are commonly released from neurohemal organs known as the corpus cardiacum (CC) and corpus allatum (CA). Adult reproductive diapauses are characterized by a failure of the CA to release juvenile hormone (JH), which leads to arrested reproductive development. Under short days, neurosecretory cells in the PL of the brain of the Colorado potato beetle, Leptinotarsa decemlineata, likely produce the hormone allatostatin and block the CA from producing JH, which leads to the diapause phenotype (Khan et al. 1986). In another species which has a photoperiodically-induced adult reproductive diapause, P. terraenovea, nickel back-filling demonstrated that neurosecretory cells in the PI and PL innervate the CA and CC (Shiga and Numata 2000, 2001). When the neurosecretory cells in the PI were removed, females that would normally develop their ovaries under long-day conditions displayed a diapause phenotype. In contrast, when neurosecretory cells in the PL were removed, most females continued to develop their ovaries even when they were exposed to short-day diapause-inducing conditions. Although it is unclear from these experiments whether removal of the PI and the PL affects photoperiodic time measurement or ovarian development directly, the results of this experiment suggest that in P. terraenovea, cells in the PI produce a compound (perhaps allatotropin) that is critical for ovarian development, whereas neurosecretory cells in the PL produce a compound (perhaps allatostatin) that inhibits ovarian development.

More recently, Shiga and Numata (2009) have investigated the site of photoperiodic time measurement in the brain of P. terraenovea. Circadian clock cells that synthesize the protein pigment-dispersing factor (PDF) innervate neurosecretory cells in the PL of P. terraenovea (Shiga and Numata 2009). PDF is also co-expressed in many central circadian clock cells in Drosophila and is an important output of circadian clocks, thus suggesting an exciting model that PDF accumulates under successive short nights, reaches a critical threshold that signals neurosecretory cells in the PL to produce allatostatin, and hence initiates diapause (Fig. 3). Indeed, ablation of PDF-expressing circadian neurons, which are likely involved in photoperiodic time measurement, resulted in a lower incidence of diapause in P. terraenovea.

Fig. 3

(A) Schematic representation of the brain of the blow fly, Protophormia terraenovae (adapted from Shiga and Numata 2001, 2009), showing connections between PDF-positive, small lateral ventral neurons (sLN_v) that innervate neurosecretory cells in the pars lateralis (PL). These neurosecretory cells innervate the corpora cardiacum (CC) and corpora allatum (CA) that produce and/or release key hormones involved in development and diapause. (B) An hypothetical model showing the potential role of PDF in initiating diapause. Under short days, high amounts of PDF are transmitted to neurosecretory cells in the PL, promoting the production of allatostatin, which prevents the CA from producing juvenile hormone (JH), thereby resulting in adult reproductive diapause. Under long-day conditions, PDF levels are low and the neurosecretory cells in the PI produce allatotropin, which stimulates the CA to produce JH, leading to reproductive maturation.

In A. pernyi and D. plexippus, two species of Lepidoptera, the central circadian clock seems to be localized within neurosecretory cells in the PL. In addition, PER-expressing cells were found in the CA of A. pernyi (Sauman and Reppert 1996) and in the CC of D. plexippus (Sauman et al. 2005). A CRY1-positive pathway connects D. plexippus circadian clock cells in the PL to both the light input from the eyes and to neurosecretory cells in the PI. These results demonstrate that clock neurons are connected to regions of the brain that control diapause, and as clock genes are expressed in the neurohemal organs themselves suggest the possibility that the cycling of clock genes in these organs may also play an important role in initiating diapause.

Molecular studies on clock genes and diapause

In D. melanogaster, flies that are null for the per gene, and hence arrhythmic, are still capable of entering diapause (Saunders et al. 1989; Saunders 1990). Although the CPP is slightly shifted in per-null flies, the results suggest that per is not involved in initiating the photoperiodic diapause in this species. These results, however, do not preclude the involvement of other clock genes, such as tim, in the diapause response.

Two naturally varying alleles of tim are present in D. melanogaster (Tauber et al. 2007); one that exclusively produces a short form of the mRNA (s-tim) and one that produces both long and short forms (ls-tim). The ls-tim form correlates with higher levels of diapause in Europe, and hence is adaptive in temperate climates. In a related study, Sandrelli et al. (2007) demonstrate that the S-TIM protein binds more tightly to CRY1 than to LS-TIM, such that the shorter form is more readily degraded. As LS-TIM has the potential to decrease light sensitivity of the circadian clock, Sandrelli et al. (2007) postulated that the ls-tim allele would be adaptive in northern latitudes where day length dramatically increases in the summer. However, the allelic distribution demonstrates that the s-tim allele is more common in northern climates, and therefore allelic variation of tim is not linked to the diapause response (Tauber et al. 2007). However, these results might also be explained by the relatively recent origin of the ls-tim allele, which arose ∼10,000 years ago in southeastern Italy (Sandrelli et al. 2007). This allele has been spreading throughout natural populations, and selection may be acting to increase the allelic frequency of ls-tim in seasonal European habitats (Kyriacou et al. 2008). Flies that are tim-null are still capable of entering diapause, but they lose their photoperiodic response, allowing a portion of the population to enter diapause regardless of day length (Tauber et al. 2007). Therefore, some components of the circadian clock, such as tim but not per, may be involved in measuring day length or modulating other features involved in the induction of diapause in Drosophila.

As indicated above, Drosophila has a unique molecular clock that, unlike most other insects, lacks cry2. Moreover, Drosophila has a weak diapause that may be more akin to quiescence (Tatar et al. 2001). Thus, D. melanogaster is not the best organism to use for examining the relationship between clock genes and the photoperiodic diapause response. Also, as photoperiodic diapause has likely evolved numerous times in insects, it is essential to examine the connection between circadian clock genes and diapause in diverse insect species to gain a more comprehensive view of this interaction.

Research on the drosophilid fly Chymomyza costata implicates tim as a potential link between the circadian oscillator and the photoperiodic induction of diapause. Riihimaa and Kimura (1988) identified a non-photoperiodic diapausing mutant strain of C. costata, and Pavelka et al. (2003) mapped the mutation to a deletion in the 5′ leader sequence of the tim gene. The deletion contains sites for transcription activation and regulatory motifs, including the E-box promoter, which inhibit tim transcription, resulting in low TIM protein levels in non-diapausing mutants (Stehlik et al. 2008).

The flesh fly, S. bullata, offers additional insight into a potential role for per in the photoperiodic diapause response. Goto et al. (2006) identified a non-diapausing strain of flesh flies that also display arrhythmic eclosion patterns. The lack of circadian rhythmicity suggested a malfunction of the clock. However, rather than showing suppression of per or tim, both genes were more highly expressed in mutant flies. A comparison of per sequences isolated from wild-type, high and low diapausing strains of S. bullata revealed that incidences of diapause were higher in strains that had shorter PER C-terminal regions, designated the C-terminal photoperiodic (CP) region (Han and Denlinger 2009). Future research is needed to identify proteins that interact with this region of PER as such proteins could be involved in the photoperiodic induction of diapause.

Clock genes have also been implicated in the adult photoperiodic diapause of the bean bug, Riptortus pedestris. When R. pedestris are reared under diapause-inducing short days or diapause-averting long days per is expressed at nearly identical levels (Ikeno et al. 2008). When per or cry2 expression was knocked down using RNA interference (RNAi) both the daily rhythm of cuticle deposition (a circadian response) and ovarian arrest under short-day conditions (a photoperiodic response) were disrupted (Ikeno et al. 2010, 2011a, 2011b). R. pedestris females treated with per or cry2 double-stranded RNA (dsRNA) produced a single dark layer of cuticle as opposed to alternating dark and light layers in control insects. RNAi directed against per and cry2 also caused females reared under short-day conditions, which normally enter a reproductive diapause characterized by undeveloped ovaries, to develop their ovaries. When RNAi was used to knock down cyc or Clk, the opposite effect was observed: the bugs produced a single bright layer of cuticle and failed to develop ovaries even when reared under long-day conditions (Ikeno et al. 2010; Goto 2013). This work was repeated in male bean bugs with identical results: per and cry2 dsRNA caused short-day males to avert reproductive diapause, while cyc dsRNA caused long-day males to enter diapause (Ikeno et al. 2011a). These results, and other targeted studies on clock genes, demonstrate that circadian clock genes are involved in photoperiodic induction of diapause, suggesting that insects may have co-opted their ancestral circadian clocks during the evolution of photoperiodic responses.

Pleiotropy, plasticicty, and programming

The problem of pleiotropy

Diapause initiation is a complex process, beginning with perceiving light and interpreting photoperiod, followed by storing this information and translating it into hormonal signals that then lead to a complex suite of physiological, morphological, and behavioral changes that characterize the diapause phenotype. Emerson et al. (2009) correctly pointed out that the central circadian clock, or its individual genes, may be interacting pleiotropically in any one or all of these steps. This is noteworthy, as circadian clocks regulate numerous behavioral and physiological processes, and there may very well be overlap between circadian and diapause phenotypes. One recent and elegant example of this (Bajgar et al. 2013) demonstrates in the linden bug, Pyrrhocoris apterus, that CLK and CYC interact in the gut with JH to increase expression of the clock gene Pdp1_iso1 and suppress expression of cry2 under long days. In contrast, cry2 expression is elevated and Pdp1_iso1 is suppressed under short days. These clock genes are therefore acting to regulate the diapause status of the P. apterus gut and are doing so independently of their daily oscillations in circadian clocks. It is thus challenging to determine where, and whether, circadian clocks (modular pleiotropy) or a clock gene (gene pleiotropy) is exerting its effects on the initiation of photoperiodic diapause.

To empirically determine whether an individual clock gene or the clock as a functioning module affects diapause, Emerson et al. (2009) encouraged researchers to examine the effects of multiple, single clock gene knockouts. To date, the previously mentioned RNAi experiments in the bean bug are the only experiments that have taken such an approach (Ikeno et al. 2010, 2011a, 2011b; Goto 2013). Knocking down per and cry2, which are both negative regulators of the circadian clock, stopped the cuticle-deposition rhythm in the same phase (only a dark layer was produced), and both caused the same effect: aversion of diapause under short-day conditions (Ikeno et al. 2010, 2011a, 2011b). In contrast, knocking down cyc or Clk, which are positive regulators of the circadian clock, arrested the cuticle-deposition rhythm in a different phase (only bright layers were produced), and both male and female bugs entered diapause under long-day conditions (Ikeno et al. 2010, 2011a, 2011b; Goto 2013). This is strong evidence that the circadian clock, as a functioning unit, is involved in the diapause response in R. pedestris because circadian genes that have the same role in the clock exert the same effect on adult reproductive diapause.

Yet, the question remains: where in the process are these clock genes affecting the diapause response? The experiments by Ikeno et al. (2010) offer further insight: the long-day diapause induced by RNAi against cyc could be terminated by applying a JH analog to the cuticle. This demonstrates that knocking down cyc affects diapause upstream of JH release from the CA, but it is still unclear whether the central R. pedestris circadian clock is involved in photoreception, measurement of day length, or neurosecretory signaling to the CA.

Plasticity in the diapause program

As previously mentioned, diapause appears to have evolved numerous times in insects. This is supported by the number of life stages that enter diapause within given taxa, as well as by the variation in the types of cues, photoperiodic and otherwise, that insects use to enter diapause. The origin of diapause in insects is still unclear, as more primitive animals, such as copepods (Hairston and Munns 1984), also enter diapause, and the capacity for diapause is extremely widespread. The diapause of some temperate insects can be traced to the tropics where diapause is also common (Tauber and Tauber 1981; Denlinger 1986). As several tropical insects, such as flesh flies, colonized temperate environments natural selection altered various aspects of the underlying diapause phenotype, most notably by adding the ability to use photoperiod to predict seasonal changes in the environment. In the tropics, where changes in day length are slight, insects use cues other than photoperiod, such as temperature, rainfall, food quality, and population density, to initiate non-photoperiodic diapause (Denlinger 1986). As insects colonized higher latitudes, mechanisms for measuring day length were acquired. For example, tropical populations of the pink bollworm, Pectinophora gossypiella, have a lower incidence of diapause but are more likely to enter diapause in response to non-photoperiodic changes in food and temperature relative to populations from higher latitudes (Ankersmit and Adkisson 1967). In contrast, as other species moved into subtropical or tropical regions from temperate zones, they lost their ability to enter a photoperiodically-induced diapause (e.g., embryonic diapause in Ae. albopictus; Lounibos et al. 2003).

Additionally, flexibility in features of the diapause program, such as critical day length and duration and depth of diapause, has allowed insects to colonize new areas and has contributed to their evolutionary success. The strong positive correlation between CPP and latitude/elevation in W. smithii nicely demonstrates the importance of variation in this trait (Bradshaw et al. 2003b, 2006). Although the CPP differs among populations of W. smithii, the period of clock gene cycling remains constant although there are some changes in amplitude of cycling in the clock genes (Mathias et al. 2005). The authors argued that it would be maladaptive to couple circadian clocks, which need to maintain a stable period, to a photoperiodic diapause response that needs to remain flexible (Bradshaw et al. 2003b; Bradshaw and Holzapfel 2007).

Reconciling the need for a flexible diapause program with a stable clock

As suggested above, circadian clocks likely evolved early and, due to their adaptive significance (Sharma et al. 2003), the DNA sequences of clock genes have been well-conserved. There are, however, significant changes in the composition of insect clocks that occurred during evolution, such as the loss of cry2 in Drosophila, the loss of cry1 in the flour beetle, T. castaneum, and the loss of both tim and cry1 in the honey bee, Apis mellifera and other hymenopterans (Yuan et al. 2007; Sandrelli et al. 2008). Yet, changes in the gene composition of circadian clocks did not change the inherent circadian periodicity (i.e., their circadian oscillators continue to cycle with endogenous periods that range from ∼22 to 25 h in constant darkness), thus demonstrating the diverse structural possibilities of the clock. The fact that the endogenous period of most circadian oscillators (τ) is not equal to 24 h indicates that insects have to continually reset their circadian pacemakers to match the periods of their environments. Therefore, circadian pacemakers can reset, or entrain to diverse inputs, such as temperature (Zimmerman et al. 1968) and food (Frisch and Aschoff 1987), as well as to light:dark cycles. Other features of circadian clocks in insects, such as their ability to measure night length by resetting to “lights-off cues,” may have predisposed the clock to measure photoperiod (Saunders 2009). Pittendrigh and Daan (1976) used sophisticated models to demonstrate that when τ ≠ 24 h, animals are able to reset their endogenous circadian pacemakers to “lights-on” (when τ > 24 h), or “lights-off” (when τ < 24 h) cues over a broader range of photophases than when τ = 24 h. Therefore, having an endogenous circadian period that differs from 24 h allows temperate insects to maintain circadian oscillations in environments in which day length changes dramatically throughout the year.

Circadian clocks control diverse behaviors, and we are now gaining a better appreciation for the extent of physiological and molecular processes that are also under circadian control. A recent transcriptome analysis of the malaria mosquito, A. gambiae, revealed that at least 15.8% of the genes are under either circadian or diel control (Rund et al. 2011). Genes involved in transcription and translation, metabolism, detoxification, olfaction, vision, and immunity all show daily expression profiles both under light:dark cycles and continuous darkness. Similarly, studies of Drosophila demonstrated that hundreds of transcripts are under the control of circadian clock genes (Claridge-Chang et al. 2001; McDonald and Robash 2001).

In contrast to circadian clocks, the photoperiodic diapause response has evolved numerous times in insects, and many diapause characteristics remain plastic and variable both within and between populations. Yet, the circadian clocks that operate among populations of insects with diverse photoperiodic diapause responses must and do allow insects to maintain circadian oscillations over the 24-h environmental period. Therefore, it is necessary to think of ways in which circadian clocks can be coupled to photoperiodic time measurement such that the stability of circadian clocks and flexibility of the diapause response are maintained. Goto (2013) and Tagaya et al. (2010) summarize several arguments for how this might be accomplished. Under the external coincidence model, insects from different populations can alter timing of their light sensitive stage (φ_i), such that populations from higher latitudes would be expected to have a later φ_i than that of southern populations, thereby accounting for differences in CPP. Different populations may also have different thresholds for a diapause-inducing substance, or different rates of synthesis or degradation of such a compound (Tagaya et al. 2010; Goto 2013). These scenarios for altering CPP for the induction of diapause do not require changes in the central circadian clockwork, and we argue that such possibilities could explain how insects have co-opted their existing circadian clocks to measure photoperiod and properly time the initiation of diapause.

Conclusions and future directions

Although more than 75 years have elapsed since Erwin Bünning proposed that photoperiodic responses are linked to endogenous circadian clocks, we still lack a mechanistic understanding of the central circadian clock’s involvement in photoperiodic diapause. Yet, a wide variety of classical experiments, such as Nanda–Hamner resonance and Bünsow night interruption protocols, point to a circadian basis for photoperiodic induction of diapause in nearly all insect species that have been studied. Additionally, several insect species with mutations in their circadian clock genes show abnormal diapause responses, including either changes in their CPP as in per-null Drosophila, or failure to enter diapause as in the CP-per mutant of S. bullata, and deletion of E-box sequences in the tim promoter in C. costata. Targeted RNAi experiments have demonstrated that the central circadian clock, as a functioning unit, is necessary for photoperiodic diapause initiation in the bean bug, R. pedestris. Similar approaches in additional species are needed to examine whether individually knocking down multiple positive and negative circadian regulators elicits the same effect on the induction of diapause. Experiments of this type will enable us to determine whether the central circadian clock is functioning as a module or whether individual clock genes are acting independently in initiating the diapause program.

Although it remains unclear how insects connect their circadian clocks to photoperiodic time measurement without compromising the integrity or stability of their circadian oscillators, several possibilities have been suggested (Tagaya et al. 2010; Goto 2013). Future research is needed to determine whether and how insects shift their light sensitive stage φ, and also to determine whether a clock-derived compound is involved in photoperiodic diapause initiation. Currently, we do not have solid evidence for such a substance, although PDF is an enticing candidate, especially in light of Shiga and Numata’s (2009) work with pdf-expressing neurons in P. terraenovea. Such neuroanatomical studies in additional insects can be expected to yield further insights into physical connections between the circadian clock and the photoperiodic diapause response.

Bradshaw and Holzapfel (2007) urged researchers to abandon studies that examined the effects of individual clock genes on the diapause response in favor of forward genetic techniques. Although we disagree that targeted research on clock genes has little promise, we do believe that researchers should also employ forward genetic techniques to identify non-circadian genes involved in photoperiodic diapause responses. As photoperiodic diapause has evolved multiple times in insects, and as the core circadian clock also differs among insect taxa, it is quite likely that circadian clocks play different roles in the photoperiodic responses of different species. We urge the use of every experimental technique in our arsenal, from continued classical experiments through RNAi and RNAseq, to determine the exact role of circadian clocks in photoperiodic induction of diapause in a wide range of insect species.

Funding

This work was supported in part by the National Institutes of Health Grant 2R56-AI058279 and the NSF Graduate Research Fellowship Program.

Acknowledgments

We thank the reviewers and Harold Heatwole, editor of this journal, for their thoughtful consideration and helpful critique of this article.

Author notes

From the symposium “Keeping Time During Animal Evolution: Conservation and Innovation of the Circadian Clock” presented at the annual meeting of the Society for Integrative and Comparative Biology, January 3–7, 2013 at San Francisco, California.