The concept of insect photoperiodism based on a circadian clock has been supported by many studies demonstrating that the behavioural circadian rhythm and the photoperiodic response are driven by the same circadian clock genes. However, the neuronal mechanism of the circadian clock underlying photoperiodism is poorly understood. To examine whether circadian rhythm and photoperiodism share a neuronal mechanism, we focused on the neurons that express neuropeptide pigment-dispersing factor (PDF) in the bean bug, Riptortus pedestris. PDF has been identified as an important regulator of the insect circadian rhythm and is expressed in circadian clock neurons of various insect species. In R. pedestris, PDF immunoreactivity was detected in some clusters of cells and their fibres in the optic lobe and the protocerebrum. cDNA encoding a PDF precursor protein was highly conserved between R. pedestris and many other insects. Differences between day and night were not observed in the immunolabelling intensity in cell bodies of PDF-immunoreactive neurons and pdf mRNA expression levels in the head. Surgical removal of the region containing PDF-immunoreactive cell bodies at the medulla disrupted the photoperiodic regulation of diapause. However, gene suppression of pdf by RNA interference did not affect the photoperiodic response. These results suggest that the region containing PDF-immunoreactive somata is important for the photoperiodic response in R. pedestris, but pdf mRNA expression is probably not required for the response.
Photoperiodism is a seasonal timing system that enables organisms to coordinate their development and physiology to annual changes in the environment using day length (photoperiod) as an upcoming seasonal cue (Nelson et al., 2010). Many insects adopt photoperiodism to overcome harsh seasons unfavourable for continuous growth and reproduction by entering diapause, a hormonally regulated arrest of development (Saunders, 2002). To achieve the photoperiodic response, measuring the length of day (or night) is a crucial process. Although the mechanism of time measurement is still elusive in insects, it has been widely accepted that photoperiodic timing is based on an endogenous circadian clock (Saunders, 2002; Saunders, 2011; Goto, 2013). The circadian clock is a daily timing system used to generate a variety of circadian rhythms in behaviour, physiology and metabolism (Saunders, 2002). The insect circadian clock is driven by a number of circadian clock genes, which form interlocked negative feedback loops that generate rhythmicity at transcriptional and translational levels, as well as post-translational levels, such as phosphorylation (Hardin, 2005). Circadian clock genes and their protein products are expressed in a variety of tissues, but only a small number of neurons in the brain are responsible for the circadian clock regulating locomotor activity (Helfrich-Förster, 2003; Tomioka et al., 2012). While the involvement of the circadian clock genes in the photoperiodic response has been demonstrated in several insect species (Pavelka et al., 2003; Sakamoto et al., 2009; Ikeno et al., 2010; Ikeno et al., 2011b; Ikeno et al., 2011c; Ikeno et al., 2013; Bajgar et al., 2013a; Bajgar et al., 2013b), less is known about the neuronal mechanism of the circadian clock underlying photoperiodism (Shiga, 2012).
In Drosophila melanogaster, the neuropeptide pigment-dispersing factor (PDF) is co-expressed with the circadian clock genes in two groups of the circadian clock neurons, the large and small ventral lateral neurons (l-LNvs and s-LNvs). Co-expression of PDF and circadian clock genes has been reported in many other insect species, including a heteropteran insect, Rhodonius prolixus (Vafopoulou et al., 2010). In R. prolixus, PDF is expressed in lateral neurons (LNs) where expressions of circadian clock proteins exhibit circadian rhythms, indicating that PDF-immunoreactive (PDF-ir) LNs are circadian clock neurons also in Heteroptera (Vafopoulou et al., 2010). In D. melanogaster, PDF functions as an output signal from the circadian clock to regulate locomotor activity, because pdf-null mutant flies showed a disturbed locomotor activity rhythm under constant darkness (Renn et al., 1999). In addition, in other insects, PDF appears to be an important regulator of the circadian rhythm. In the cockroach Leucophaea maderae, lesion and transplantation experiments revealed that the circadian clock regulating the locomotor rhythm is located in the accessory medulla (aMe) containing neurons immunoreactive for the pigment-dispersing hormone (PDH), which is the crustacean homolog of PDF (Stengl and Homberg, 1994; Reischig and Stengl, 2003). The PDH-immunoreactive (PDH-ir) neurons are also called PDFMe neurons and are homologous to the LNv in D. melanogaster. Injection of PDH shifted the onset of the locomotor rhythm in L. maderae (Petri and Stengl, 1997). In the cricket Gryllus bimaculatus, PDF injection caused phase shifts of the locomotor rhythm in a phase-dependent manner (Singaravel et al., 2003). Using RNA interference (RNAi) techniques, it was shown that knockdown of pdf gene expression disrupted the circadian locomotor rhythm of the German cockroach, Blattella germanica (Lee et al., 2009). pdf RNAi was also applied to G. bimaculatus and resulted in suppression of nocturnal activity, shortening of the free-running
large ventral lateral neurons
small ventral lateral neurons
period in constant darkness, and enhancement of the photic entrainability to shifted light–dark cycles (Hassaneen et al., 2011).
In addition to these studies implicating PDF in the circadian rhythm, some studies have shown involvement of PDF-ir neurons in photoperiodism. Surgical removal of the region containing PDF-ir s-LNv neurons, which express the circadian clock protein PERIOD, abolished not only the circadian rhythm in locomotor activity but also the photoperiodic regulation of diapause in the blow fly Protophormia terraenovae (Shiga and Numata, 2009). In this insect, s-LNvs were shown to be synaptically connected to neurons with somata in the pars lateralis, called PL neurons, which are indispensable for diapause induction (Hamanaka et al., 2005). The studies in P. terraenovae suggest that the circadian clock neurons expressing PDF are essential for the neuronal pathway of photoperiodic diapause. In L. maderae, different photoperiods affect the number of PDF-ir neurons anterior to the aMe and their branching patterns. Although L. maderae is a tropical species and there is no immediate need to adjust to changes in photoperiod, the PDF-ir circadian clock neurons might allow for adaptation to photoperiodic cycles during the course of the year (Wei and Stengl, 2011).
To reveal the neuronal mechanism of the photoperiodic timing system, we focused on PDF-ir neurons and the pdf expression in the bean bug, Riptortus pedestris (Fabricius), formerly known as R. clavatus (Thunberg) (Heteroptera: Alyidae), which exhibits a clear photoperiodic response. Riptortus pedestris ovarian development is induced under long-day conditions but is suppressed under short-day conditions, leading to entry into diapause (Numata and Hidaka, 1982). Riportus pedestris is one of the best-studied insects for involvement of the circadian clock in photoperiodism at the molecular level. RNAi targeting for the core circadian clock genes, period, mammalian-type cryptochrome, cycle and Clock, abolished both the circadian rhythm in cuticle deposition and the photoperiodic response, indicating that the same molecular elements are involved in the underlying clock mechanisms of circadian rhythm and photoperiodism (Ikeno et al., 2010; Ikeno et al., 2011a; Ikeno et al., 2011b; Ikeno et al., 2011c; Ikeno et al., 2013). Here, we hypothesized that photoperiodic and daily timing systems share not only molecular but also neuronal mechanisms of the circadian clock; therefore, the ablation of putative circadian clock neurons, PDF-ir neurons, would disrupt the photoperiodic response of R. pedestris. In the present study, we first investigated the characteristics of PDF-ir neurons, the pdf gene and the PDF peptide of R. pedestris. In order to determine whether PDF-ir neurons play a role in photoperiodism, we examined the effect of surgical removal of the region containing PDF-ir neurons on the photoperiodic response. We also performed gene suppression of pdf by RNAi to examine the possibility that the PDF peptide itself functions in the photoperiodic response.
PDF-ir neurons in the brain
Twenty-seven days after adult emergence, brains from intact insects were subjected to immunohistochemistry using the ABC method with the anti-Gryllus PDF antiserum. The antibody clearly labelled some cell bodies and their fibres in the optic lobe and the protocerebrum, and the general labelling pattern was similar overall in the two brain hemispheres (Fig. 1A,B,K,L). At the anterior proximal medulla of the optic lobe, a dense cluster of eight to 10 cells was found (Fig. 1C). These neurons correspond to PDFMe neurons (also called LNs) in other insects (Homberg et al., 1991; Nässel et al., 1991; Sato et al., 2002; Sehadová et al., 2003; Abdelsalam et al., 2008; Vafopoulou et al., 2010). In the cluster, different-sized cells were observed, as reported in other insects (Abdelsalam et al., 2008; Weiss et al., 2009; Sumiyoshi et al., 2011; Vafopoulou and Steel, 2012). The diameter of the largest cell body in the cluster was 14.8±0.65 μm (mean ± s.e.m., N=13), whereas that of the smallest one was 7.7±0.41 μm (N=10), and they were significantly different (Mann–Whitney test, P<0.01). Fine arbours with dense varicosities spread radially from PDFMe somata along the dorso-proximal surface of the medulla (Fig. 1C). The axons from PDFMe neurons formed a bundle and travelled along the dorso-anterior surface of the medulla, producing fine branches into the medulla (Fig. 1D). At the posterior lamina, PDF-ir fibres fanned out and formed a mesh-like structure with dense varicosities along the surface of the lamina (Fig. 1E).
A bundle of axons from PDFMe neurons also projected to the protocerebrum. At the junction between the optic lobe and the lateral protocerebrum, the axons of PDFMe neurons changed direction (Fig. 1F, large arrow). At this point, a branch with varicosities extended from the axonal bundle on the lateral edge of the protocerebrum toward the posterior optic lobe (Fig. 1G, large arrow). In the dorsal protocerebrum, individual axons deposited branches with large numbers of varicosities (Fig. 1H). The axons traversed contralaterally, and bilateral PDF-ir neurons seemed to connect to each other (Fig. 1I).
Other than the PDFMe neurons, several neurons were also immunolabelled by the anti-PDF antiserum. In the pars intercerebralis, a cluster containing 10–12 cells was found (Fig. 1I). These neurons showed cytoplasmic immunoreactivity. Relatively weak immunolabelling was detected in some cells in the anterior dorsal protocerebrum (Fig. 1G, small arrow) and in the mid-dorsal protocerebrum (Fig. 1H, small arrows). In addition to these cells, very weakly labelled cells were found at the posterior protocerebrum (Fig. 1J, small arrows). These cells were only faintly immunolabelled and even not observed in some specimens (for example, the specimen used for the tracing in Fig. 1K). When fluorescent immunohistochemistry was undertaken, although the labelling pattern of PDFMe neurons were the same as that obtained by the ABC method, signals from other neurons were faint and scarcely detectable (see Fig. 6C).
Temporal analysis of PDF immunoreactivity
To examine temporal changes of PDF immunoreactivity in the PDF-ir neurons, we performed fluorescent immunohistochemistry at different time points at 12 h intervals (Fig. 2). The labelling intensities of the PDFMe cell bodies did not show time-related differences under either the light–dark cycles or constant darkness (Kruskal–Wallis test, P>0.05).
Isolation of cDNA encoding the PDF precursor
pdf cDNA (DDBJ/GenBank/EMBL: AB797131) of R. pedestris was obtained using RACE methods. The obtained nucleotide sequence included the 5′-untranslated region (UTR) (98 bp), the open reading frame encoding the PDF precursor (258 bp), and the 3′-UTR (45 bp). Although two potential polyadenylation signal sites were found in the 3′-UTR, the length of the 3′-UTR region was extremely short compared with those of other insects; therefore, we could not determine the complete sequence of the 3′-UTR. The PDF region (18 amino acids) located in the C terminal of the PDF precursor of R. pedestris was found to be highly conserved among insects (Fig. 3A). In addition to a processing cleavage site (KR), the computer-assisted signal peptide prediction method for eukaryotic sequences (Nielsen et al., 1997; Nielsen et al., 1999) found that the R. pedestris PDF precursor possesses a signal region (24 amino acids) and a PDF-associated peptide (PAP) region (39 amino acids), as found in many other insects (Fig. 3B). In contrast to the PDF region, the length and the amino acid sequences of the signal and PAP regions were less conserved among insects (Fig. 3B).
The temporal expression pattern of pdf mRNA
Quantitative real-time PCR analysis revealed that the mRNA abundance of pdf did not exhibit diurnal oscillation under long-day and short-day conditions (Fig. 3C). Two-way ANOVA detected no significant differences among Zeitgeber times (ZTs: times after lights on) and between long-day and short-day conditions (P>0.05).
Effects of removal of PDFMe neurons on photoperiodism
To clarify the role of PDFMe neurons in the photoperiodism of R. pedestris, we bilaterally ablated the brain regions containing PDFMe cell bodies (Site 1) or control regions (Site 2) of insects reared under short-day conditions 7 days after adult emergence (Fig. 4A). Thereafter, insects were kept for 20 days under short-day conditions or transferred to long-day conditions. In this photoperiodic schedule, a clear photoperiodic response was observed in intact insects: they entered diapause under short-day conditions, whereas ovarian development was induced under long-day conditions (Fig. 4B). Moreover, the Site 2-operated control insects showed a photoperiodic response (Fig. 4B). However, in the Site 1-operated group, ovarian development was induced not only in most insects reared under long-day conditions but also in more than half of insects reared under short-day conditions (Fig. 4B).
To confirm whether PDFMe neurons were ablated by the operation, PDF immunohistochemistry was performed after judgment of reproductive status. The total numbers of PDFMe cell bodies were significantly smaller in the Site 1-operated insects than that in intact insects, both under short-day conditions and under long-day conditions (Fig. 4C, Fig. 5D,E). In addition, fibre distribution of PDFMe neurons showed abnormal patterns in 20 out of 21 Site 1-operated insects (Fig. 5). We found that many Site 1-operated insects showed the same pattern of PDFMe fibres, regardless of the number of remaining PDFMe cell bodies. In the optic lobe, the individual axons did not form an axonal bundle but separately reached to the lamina over the medulla (Fig. 5C, left side in F2–F4). In the protocerebrum, although the dorsal fibres traversing the midline were still observed, the abnormal fibres of PDFMe neurons projected to the posterior ventral protocerebrum (Fig. 5F1,F4). In some specimens, these posterior ventral fibres ran contralaterally to the other hemisphere (Fig. 5F2). Further, in the Site 2-operated control group, abnormal patterns of PDFMe fibres were observed in nine out of 22 insects (data not shown). However, the numbers of PDFMe cell bodies were not different between the Site 2-operated control insects and the intact insects under short-day or long-day conditions (Fig. 4C), suggesting that the abnormal fibres were not caused by removal of PDFMe cell bodies. Statistical analysis did not reveal a significant difference between the numbers of PDFMe cell bodies of Site 2-operated control insects and those of Site 1-operated insects in long-day conditions (Steel–Dwass test, P>0.05); this is probably because the number of brains for which PDF cell bodies were counted was small. There were no statistical differences in the number of PDFMe cell bodies between reproductive and diapause insects with operation either in Site 1 or Site 2 under short-day or long-day conditions (Mann–Whitney test, P>0.05 for all cases). In addition, we did not detect any correlation between the abnormal pattern of PDFMe fibres and reproductive status, irrespective of photoperiodic conditions (Fig. 5F).
Effects of pdf RNAi on photoperiodism
To clarify the role of the PDF peptide in photoperiodism of R. pedestris, we carried out RNAi targeting pdf. Double-stranded RNA (dsRNA) of pdf was injected into insects reared under short-day conditions at the day of adult emergence. Five days after injection, northern blotting confirmed that the expression of pdf mRNA was clearly suppressed in insects injected with pdf dsRNA, compared with that in insects injected with bla (control) dsRNA (Fig. 6A). We also performed PDF immunohistochemistry with the ABC method 20 days after dsRNA injection. Although the ABC method amplified the signal of the biotinylated primary antibody, the immunoreactivity was much weaker in PDFMe neurons, especially in their fibres in the protocerebrum, of the insects injected with pdf dsRNA than that of insects injected with bla dsRNA (Fig. 6B1–B5, large arrows in B4–B6). However, immunoreactivity of cells other than the PDFMe neurons was not different between insects injected with pdf dsRNA and those injected with bla dsRNA (Fig. 6B5–B8, small arrows). To quantify the labelling intensity of the PDFMe neurons, fluorescent immunohistochemistry was performed. Five days after dsRNA injection, the labelling intensity of PDFMe cell bodies was significantly lower in insects injected with pdf dsRNA than in insects injected with bla dsRNA (Mann–Whitney test, P<0.01; Fig. 6C,D). On day 20, the overall labelling intensity of the PDFMe neurons was greatly reduced in insects injected with pdf dsRNA, and immunolabelling was scarcely observed in the fibres and arborisations at the optic lobe and the protocerebrum (Fig. 6C). These results indicate that injection of pdf dsRNA decreased the expression level of pdf mRNA and PDF immunoreactivity in the cell bodies of PDFMe neurons within the first 5 days, and immunoreactivity in the fibres was gradually decreased within 20 days.
Twenty days after dsRNA injection, ovarian development was checked in insects kept under short-day or transferred to long-day conditions. Under short-day conditions, intact insects and insects injected with bla dsRNA did not develop their ovaries and entered diapause (Fig. 7). Similarly, any insects injected with pdf dsRNA did not show ovarian development (Fig. 7). In the group transferred to long-day conditions, most of the intact insects and insects injected with bla dsRNA developed their ovaries (Fig. 7). Insects injected with pdf dsRNA also developed their ovaries after they were transferred to long-day conditions (Fig. 7). Thus, the clear photoperiodic response was observed not only in intact insects and insects injected with bla dsRNA insects, but also in those injected with pdf dsRNA.
The neuroarchitecture of PDF-ir neurons has been described for the brains of various insect species (Homberg et al., 1991; Nässel et al., 1993; Závodská et al., 2003; Abdelsalam et al., 2008; Sumiyoshi et al., 2011; Vafopoulou et al., 2010; Vafopoulou and Steel, 2012). Three major clusters have been identified in the optic lobe: a cluster located at the anterior edge of the medulla (PDFMe), a cluster at the posterior dorsal edge of the lamina (PDFLad) and a cluster at the posterior ventral edge of the lamina (PDFLav) (Meelkop et al., 2011). In the optic lobe of R. pedestris, only the PDFMe cluster was found, as is the case in some insects, such as the larvae of Rhodnius prolixus and the honeybee Apis mellifera (Vafopoulou et al., 2010; Sumiyoshi et al., 2011). PDFMe neurons have been classified into several groups of different-sized neurons with different axonal projections, and their different functions in the circadian clock have been extensively studied in D. melanogaster (Helfrich-Förster, 2003). In the present study, we observed neurons with larger and smaller cell bodies in the PDFMe cluster. Therefore, based on their sizes, it is most likely that in R. pedestris, the PDFMe cluster contains at least two types of neurons.
Fibre distribution patterns and cell body locations of PDFMe neurons in R. pedestris were comparable to those in other insects studied so far: they sent their fibres into both the optic lobe and the protocerebrum with dense varicosities in the accessory medulla, medulla, lamina and mid-dorsal protocerebrum. However, the present study also revealed some features not observed in other insects. In many insects, the axons from PDFMe neurons spread on the surface of the medulla and lamina (Sato et al., 2002; Závodská et al., 2003; Abdelsalam et al., 2008; Sumiyoshi et al., 2011; Vafopoulou et al., 2010; Vafopoulou and Steel, 2012). In R. pedestris, however, the axons ran on the surface of the medulla by forming a thick bundle and fanned out at the lamina. In addition, some branches from the bundle ran into the medulla. These patterns were not observed even in R. prolixus, which belongs to the same suborder as R. pedestris (Vafopoulou et al., 2010; Vafopoulou and Steel, 2012); therefore, they might be unique to R. pedestris.
Other than PDF-ir cell clusters in the optic lobe, several PDF-ir (or PDH-ir) neurons in the protocerebrum have been identified in some insects, and their species-specific functions have been discussed (Homberg et al., 1991; Nässel et al., 1993; Závodská et al., 2003; Vafopoulou and Steel, 2012). In R. pedestris, several PDF-ir cells in the protocerebrum were also detected. However, these cells did not decrease their immunoreactivity after pdf RNAi, suggesting that their immunoreactivity was likely due to a PDF-like peptide with recognition site for the anti-PDF antibody. In D. melanogaster, it was shown by in situ hybridization that pdf mRNA was expressed only in LNvs, although the anti-PDH antibody immunolabelled some other neurons (Helfrich-Förster and Homberg, 1993; Helfrich-Förster, 1997).
We did not detect any differences in the immunolabelling intensity of cell bodies of PDFMe neurons among the end of the photophase, scotophase, subjective day and subjective night. Moreover, pdf mRNA expression levels showed no diurnal rhythmicity under either long or short days in the head of R. pedestris. Although diurnal and circadian expression of pdf mRNA were reported for A. mellifera (Sumiyoshi et al., 2011), pdf mRNA levels and cytoplasmic PDF immunoreactivity showed no prominent oscillation in many other insects (Park and Hall, 1998; Park et al., 2000; Matsushima et al., 2003; Hassaneen et al., 2011). However, circadian oscillations of immunoreactivity have been reported in the terminals of s-LNvs in D. melanogaster (Park et al., 2000). In R. prolixus, the labelling intensity changed diurnally in the axons and varicosities of the terminals of PDFMe neurons (Vafopoulou et al., 2010). From these observations, it has been suggested that the cyclical release of PDF from circadian clock neurons into the dorsal protocerebrum transfers rhythmic signals to downstream neurons to regulate rhythmic activity. Further examination is necessary to investigate whether PDF release is rhythmically regulated in R. pedestris.
Ablation of the brain region containing PDFMe neurons (Site 1) disrupted the photoperiodic response and induced ovarian development even under short-day conditions. The operation at the control region, where PDF neurons did not reside (Site 2), had no effect on the photoperiodic response, indicating that induction of ovarian development by Site 1 operation was not the result of damage to any brain tissue. Therefore, our result suggests that the region containing PDF neurons is essential for the photoperiodic response, as demonstrated in P. terraenovae (Shiga and Numata, 2009). Removal of this region induced reproduction irrespective of photoperiod, which was expected because disruption of proper functions of the circadian clock under constant light induces ovarian development in R. pedestris (Kobayashi and Numata, 1993). Although the number of PDFMe cell bodies and incidence of developed ovaries in Site 2-operated control insects were not different from those in intact insects, arborisation patterns were quite different, i.e. abnormal fibres were found in control insects as well as in Site 1-operated insets. These results suggest that the cell bodies of PDFMe neurons, not the fibre patterns, are important for the photoperiodic response. However, there was no correlation between the number of PDFMe cell bodies and reproductive status, as was also observed in P. terraenovae (Shiga and Numata, 2009). These results could be interpreted in several ways. The first possibility is that only part of the PDFMe neurons are important for the photoperiodic response and ovarian development induced in short-day conditions was caused by absence of these important neurons, as proposed by Shiga and Numata (Shiga and Numata, 2009). This could be supported by the observation that the PDFMe cluster contains different types of PDF-ir neurons. In the present study, however, it was difficult to distinguish the types of cells that survived because these cells usually suffered extensive damage after surgery. The second possibility is that neurons, which reside close to PDFMe neurons but do not express the PDF peptide, are responsible for the photoperiodic response. These neurons may or may not play a role in the circadian clock. However, in R. pedestris, the circadian clock has been shown to be involved in the photoperiodic response (Ikeno et al., 2010; Ikeno et al., 2011a; Ikeno et al., 2011b; Ikeno et al., 2011c; Ikeno et al., 2013), suggesting that these neurons would also be the circadian clock neurons. Such PDF-immunonegative circadian clock neurons are indeed observed in D. melanogaster (Rieger et al., 2006; Picot et al., 2007; Hermann et al., 2012). To interpret the effects of Site 1 removal at a cellular level, a neuroanatomical study of this region will be necessary. PDFMe neurons and adjacent PDF-immunonegative neurons should be identified by immunohistochemical colocalization studies of neuropeptides, amines and proteins, such as the neuropeptide F, choline acetyltransferase, circadian clock proteins, GABA, allatostatin, or leucokinin, which have been identified in cells at the accessory medulla region (Petri et al., 1995; Peschel and Helfrich-Förster, 2011).
In the cockroaches L. maderae and Blattella germanica and in D. melanogaster, PDFMe neurons (or LNs) are circadian clock neurons regulating the locomotor activity rhythm, and they use PDF as an output signal (Stengl and Homberg, 1994; Petri and Stengl, 1997; Renn et al., 1999; Helfrich-Förster, 2003; Reischig and Stengl, 2003; Lee et al., 2009). Therefore, we examined the hypothesis that PDF itself represents the circadian output signal from these neurons and is used for the photoperiodic response in R. pedestris. Injection of pdf dsRNA clearly suppressed pdf mRNA expression in the head and decreased the PDF immunoreactivity of PDFMe neurons. Although the labelling intensity in the cell bodies decreased within 5 days of dsRNA injection, it needed a much longer time to decrease immunoreactivity in the fibres. These results indicate that PDF peptides were not produced after dsRNA injection, but it took time to degrade already-existing PDF in the fibres. In G. bimaculatus, the decrease of PDF immunoreactivity by RNAi was much slower than that of pdf mRNA levels, and complete disappearance was not observed during the 20 days after dsRNA injection, although pdf RNAi decreased PDF immunoreactivity to below the detection limit within 10 days in B. germanica (Lee et al., 2009; Hassaneen et al., 2011). The long lifetime of the PDF peptide would explain the high PDF immunoreactivity in the fibres of PDFMe neurons, even 20 days after the cell bodies were completely removed. Despite these decreases of mRNA and peptide levels, however, no effects of pdf RNAi on the photoperiodic response of R. pedestris were observed. It is still possible that a small amount of PDF peptides remaining in the cell bodies or fibres was sufficient for the photoperiodic response in R. pedestris; however, even a slight decrease of PDF peptides affected the locomotor activity rhythm in G. bimaculatus (Hassaneen et al., 2011). Therefore, it is most likely that the PDF peptide itself or at least pdf mRNA expression is not involved or plays a negligible role in the photoperiodic response in R. pedestris. Although PDF is thought to be the only output factor involved in the control of locomotor activity in cockroaches (Lee et al., 2009), D. melanogaster likely uses molecules other than PDF as output signals under light–dark cycles because pdf-null mutant flies only showed arrhythmic locomotor activity under constant darkness (Renn et al., 1999). Thus, it might be reasonable that PDF is not required for photoperiodism because the photoperiodic response is largely based on light–dark cycles.
In conclusion, our results suggest that the region containing PDFMe somata is involved in the photoperiodic response but pdf transcription is probably not important. Indeed, it was shown in P. terraenovae that PDF immunoreactivity was not detected in the synaptic vesicles at the synaptic connection between PDF-ir neurons and PL neurons, suggesting that circadian clock neurons regulate photoperiodic diapause using substances other than PDF as neurotransmitters (Hamanaka et al., 2005). We have not yet succeeded in revealing whether PDFMe neurons are circadian clock neurons regulating behaviour rhythms in R. pedestris, because individuals of this species do not show a clear circadian rhythm in locomotor activity in laboratory conditions. The oviposition rhythm is the only observable behavioural rhythm under control of a circadian clock in this insect, but oviposition itself could be easily disturbed by manipulation and is not suitable for assaying circadian behavioural rhythms (Numata and Matsui, 1988). Nevertheless, the similarities observed in the morphology of PDFMe neurons and expression patterns of pdf mRNA and PDF peptides between R. pedestris and many other insects support the hypothesis that PDFMe neurons are also circadian clock neurons in R. pedestris. Further studies are necessary to examine whether PDFMe neurons of R. pedestris express circadian clock genes and proteins. It was previously shown in R. pedestris that the circadian clock involved in photoperiodism and regulation of the circadian cuticle deposition rhythm are driven by the same molecular mechanism (Ikeno et al., 2010; Ikeno et al., 2011a; Ikeno et al., 2011b; Ikeno et al., 2011c; Ikeno et al., 2013). Thus, together with previous studies, the current findings for R. pedestris support the idea that the clock system involved in photoperiodism and that governing the circadian rhythm are driven by the same mechanism at both molecular and neuronal levels.
MATERIALS AND METHODS
Adult R. pedestris were collected in Osaka City (34.6°N, 135.5°E) from July to October in 2010 and 2011. Their progeny were reared under short-day conditions (12 h:12 h light:dark at 25±1°C) and fed soybean grains and water containing 0.05% sodium ascorbate and 0.025% l-cysteine (Kamano, 1991). Only female adults were used in the present study.
The insects were decapitated at ZT1–3, and the antennae and rostra were removed. The heads were fixed in 4% paraformaldehyde for 4 h at room temperature or overnight at 4°C. To reveal the PDF-ir neurons, immunohistochemistry was performed using the ABC method (Vectastain ABC standard kit; Vector Laboratories, Burlingame, CA, USA).
The brains were excised and washed overnight in phosphate-buffered saline (PBS) with 0.5% Triton X-100 (PBST), with the PBST changed several times. Brains were incubated in 0.3% H2O2 in PBS for 1 h at room temperature to reduce endogenous peroxidase activity, and in PBS for 4 h at 4°C to expel any remaining air from the tracheas. Brains were then incubated with rabbit polyclonal anti-Gryllus bimaculatus PDF (1:10,000) for 3 days at 4°C. The antiserum was provided by Dr K. Tomioka (Okayama University, Okayama, Japan). This was followed by incubation in anti-rabbit immunoglobulin conjugated with biotin (1:200) for 1 day at 4°C. Both primary and secondary antisera were diluted in PBST containing 10% normal donkey serum. Specimens were incubated in an avidin-biotin complex solution (1:100) for 1 day at 4°C. After washing in PBST, the brains were preincubated in 0.03% diaminobenzidine (Sigma, St Louis, MO, USA) for 1 h at 4°C, and incubated in a mixture of 0.01% H2O2 and 0.03% diaminobenzidine for ~30 s at room temperature. After washing with PBST, whole-mount preparations were dehydrated in an ethanol series and cleared in methyl salicylate for observation. Preparations were observed under a photomicroscope (BX50F; Olympus, Tokyo, Japan) and digitalized with a CCD camera (DS-Ril; Nikon, Tokyo, Japan). PDF-ir neurons were traced with a drawing attachment (U-DA; Olympus).
For the temporal analysis of PDF-ir neurons, insects were reared under short-day conditions and transferred to constant darkness at the end of the photophase (ZT12), 9 days after adult emergence (day 9). The heads were cut off 11, 23, 35 and 47 h after the beginning of the photophase of day 9 and fixed in 4% paraformaldehyde for 12 h at 4°C. The brains were excised and PDF immunohistochemistry was performed at the same time using the fluorescent immunohistochemistry method. Brains were incubated in PBS for 4 h at 4°C and incubated with anti-Gryllus PDF as described above. We used tetramethylrhodamine isomer R (TRITC)-conjugated anti-rabbit immunoglobulin (R0156; Dako, Glostrup, Denmark) as a secondary antibody (1:200) for 1 day at 4°C. After washing in PBST, preparations were dehydrated in an ethanol series and cleared in methyl salicylate for observation. Preparations were observed under an epifluorescent microscope (BX50-34FLA-3; Olympus) using a high-pressure mercury burner equipped with NIBA filter sets. The images were converted to grey scale and grey values of the stained pictures and of the background were scored using ImageJ software (http://rsb.info.nih.gov/ij/), with a scale from 0 (black) to 255 (white). Grey values of the background were subtracted from those of the stained cell bodies.
For the RNAi experiments, insects were decapitated at ZT0.5 on day 5 and day 20 and fixed for 4 h at room temperature. The brains were excised and PDF immunohistochemistry was performed using the ABC method or the fluorescent immunohistochemistry method as mentioned above.
The specificity of the antiserum was tested on whole-mount preparations. The anti-Gryllus PDF antiserum was subjected to liquid phase preabsorption with the antigen overnight at room temperature. Immunohistochemistry was carried out using the ABC method with the pre-absorbed antiserum instead of the primary antiserum. PDF immunolabelling completely disappeared at a dilution of 1:10,000 of the antiserum with 0.4 μmol l−1 Acheta domesticus PDF, NSEIINSLLGLPKVLNDA-NH2 (Peptide Institute, Osaka, Japan).
Total RNAs were isolated from the whole body of adult females reared under long-day conditions (16 h:8 h light:dark at 25±1°C) using Trizol (Life Technologies, Carlsbad, CA, USA) according to the supplier's instructions. cDNA was synthesized from total RNA using a SMARTer RACE cDNA Amplification Kit (Clontech, Palo Alto, CA, USA). The primary 3′ RACE was performed using a pdf3-F primer (5′-AAG CGC AAC TCN GAR MTV ATC AA-3′), and the secondary 3′ RACE was performed with the primary PCR product using a pdf5-F primer (5′-CCT CGG AAT CCC TAA AGT CA-3′). For 5′ RACE, the primary PCR was performed using a pdf6-R primer (5′-CGA CCA GCA TCA TTT ATG ACT TTA GGG-3′). The fragments that were 300–500 bp in length were purified and used as the template for the secondary 5′ RACE. The secondary PCR was performed using a pdf1-R primer (5′-TAN TTN CNC CCT GCA TCA TT-3′). Fragments that were ~200 and 300 bp in length were amplified in 3′ and 5′ RACE reactions, respectively.
PCR products were cloned into plasmids using a pGEM-T Easy Vector System (Promega, Fitchburg, WI, USA). Plasmids were sequenced on an ABI PRISM 3130 Genetic Analyzer (Life Technologies) with a BigDye Terminator v1.1 Cycle Sequence Kit (Life Technologies).
Quantitative real-time PCR
The temporal expression pattern of the pdf gene was investigated. Total RNAs were isolated on day 9 from the heads of five insects reared under long-day or short-day conditions at ZT0, 4, 8, 12, 16 and 20. DNA in the RNA samples was digested with Deoxyribonuclease I, Amplification Grade (Life Technologies). cDNAs were synthesized from each RNA sample using a High Capacity cDNA Reverse Transcription Kit (Life Technologies). For real-time PCR analysis, 0.2% of the cDNA was used in a final concentration of 1×GoTaq qPCR Master Mix (Promega) and 0.05 μmol l−1 of each primer using a 7500 Real-Time PCR System (Life Technologies) according to the supplier's instructions; each reaction was performed in duplicate. The primers were pdf7-F (5′-GGC GTT AGC ATT ATT TTC TTT GC-3′) and pdf7-R (5′-GAG GGA GTA AGC GGG AAC AGA-3′) for pdf, and tub4-F (5′-CTG TCA ACA TGG TCC CAT TCC-3′) and tub4-R (5′-GGA ACA GTG AGG GCC CTG TA-3′) for β-tubulin (tub), which was used as a control gene for normalization. Each primer set was designed to amplify an amplicon of ~100 bp. In all reactions, the generation of only a single expected amplicon was confirmed by melting analysis. Quantification of cDNAs was performed by the standard curve methodology. All real-time PCR experiments were performed on three independent RNA samples.
Surgical removal of PDF-ir neurons
On day 7, insects were mounted in clay with the dorsal heads exposed and then placed on ice for ~5 min. Subsequently, small windows were cut bilaterally in the cuticle between the ocelli and the compound eye using a razor blade to expose the dorsal region of each optic lobe. The portions of the anterior base of the medulla (Site 1 for the test group) or the posterior base of the medulla (Site 2 for the control group) were bilaterally ablated with a sharpened tungsten needle. Following the operation, the cuticles were returned to their original positions. After the operation, insects were reared under short-day conditions or long-day conditions. Twenty days after the operation (day 27), the reproductive status was checked, and then PDF immunohistochemistry was performed to count the PDF cell bodies in both optic lobes. All individuals used for reproductive assessment were subjected to the immunohistochemistry; however, we eliminated a certain number of brain samples from the counting of PDF cell bodies because of damage caused by the operation, which made it impossible to accurately count the PDF cell bodies on both sides.
The abdomens of insects were dissected in saline under a stereoscopic microscope. Females were classified as being reproductive or non-reproductive (diapause) on the basis of ovarian development, i.e. females with light-blue yolk deposition in the oocytes were judged to be reproductive, and those with no deposition were judged to be in diapause (Numata and Hidaka, 1982).
dsRNA of pdf was synthesized from the plasmids containing the pdf gene fragment. For a control, dsRNA from β-lactamase (bla), which provides bacteria with ampicillin resistance, was also synthesized using pGEM-T Easy Vector (Promega). Each plasmid was used as a template for PCR with Pwo Super Yield DNA Polymerase (Roche, Penzberg, Upper Bavaria, Germany) according to the supplier's instructions. The primers used in the reactions were pdf8-F (5′-ACG CAG AGT ACA TGG GGA GT-3′), pdf8T7-R (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC GTA GGG ATT CCG AGG AGT GAG T-3′), pdf8T7-F (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC GTA CGC AGA GTA CAT GGG GAG T-3′) and pdf8-R (5′-AGG GAT TCC GAG GAG TGA GT-3′) for pdf, and pGBetalacm-F1 (5′-TCG CCG CAT ACA CTA TTC TC-3′), pGBetalacmT7-R1 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC GTA CGA TAC GGG AGG GCT TAC-3′), pGBetalacmT7-F1 (5′-TAA TAC GAC TCA CTA TAG GGA GAC CAC GTC GCC GCA TAC ACT ATT CTC-3′) and pGBetalacm-R1 (5′-TAC GAT ACG GGA GGG CTT AC-3′) for bla. dsRNAs were synthesized using a T7 Ribomax Express RNAi System (Promega) according to the supplier's instructions. dsRNAs were dissolved in 0.9% NaCl (saline) and stored at −20°C until use. Formation of dsRNAs was confirmed by native electrophoresis using a 1.5% agarose gel. One microgram of each dsRNA in 1 μl of saline was injected into the head using a glass capillary within 24 h of adult emergence. Twenty days after the dsRNA injection, the reproductive status was checked.
To verify the silencing of pdf transcription after dsRNA injection, northern blotting was performed. Total RNAs were isolated from the heads of four insects reared under short-day conditions at ZT8 on day 5 as described above, and mRNA was purified using a PolyATtract mRNA Isolation Systems Kit (Promega). For the analysis, 420 ng of mRNA was used, and tub was used as a control. DNA probes were generated from the linearized plasmid DNA containing each gene fragment using a PCR DIG Probe Synthesis Kit (Roche). Hybridization was performed at 50°C with DIG Easy Hyb Granules (Roche). Chemiluminescent signals were detected with a Lumino Image Analyzer LAS-1000 (Fujifilm, Tokyo, Japan).
We are grateful to Dr K. Tomioka at Okayama University for kindly providing the antiserum.
This work was supported by a Grant-in-Aid for Scientific Research (B) [70172749 to H.N.] and a Grant-in-Aid for the Japan Society for the Promotion of Science Fellows [09J03885 to T.I.] from the Japan Society for the Promotion of Science.
The authors declare no competing financial interests.