Summary
1 Introduction
1.3 The central circadian clock of the fruit fly Drosophila melanogaster
1.3.5 The function of the neuropeptide PDF
14
(Helfrich-Förster et al. 2001; Yang and Sehgal 2001; Shafer et al. 2002; Peng et al. 2003; Veleri et al.
2003; Lin et al. 2004; Sheeba et al. 2008b) and there is evidence, that the l-LNvs function in another circuit, which is driven by the DN2s, and do not contribute to the control of rhythmic locomotor activity (Stoleru et al. 2005). However, studies on "disconnected" mutant flies (disco), which lack most of the clock neurons, revealed that flies without LNvs were arhythmic but a single LNv with projections to the superior protocerebrum was sufficient to drive rhythmicity, even if it was an l-LNv with misexpressed projections (Dushay et al. 1989; Helfrich-Förster 1998). Thus, under certain circumstances the l-LNvs appear to have the ability to substitute for s-LNvs.
1.3.4.4 Rhythmicity is driven by a flexible network of different clock neurons
Other neurons than the LNvs were also shown to contribute to free-running rhythmicity (Helfrich-Förster 1998). The DNs for example contribute to rhythmicity in LD (Veleri et al. 2003; Klarsfeld et al.
2004), functional LNds are sufficient for rhythmicity in LL (Picot et al. 2007), and the LPNs play a major role in temperature entrainment (Yoshii et al. 2005; Miyasako et al. 2007). In addition to the clock neurons, glial cells have been shown to play a role in modulating circadian behavior (Suh and Jackson 2007; Jackson 2011; Ng et al. 2011).
Several conflicting publications implicated distinct clock neurons in the control of M and E activity (Grima et al. 2004; Stoleru et al. 2004; Stoleru et al. 2005; Rieger et al. 2006; Murad et al. 2007; Picot et al. 2007; Stoleru et al. 2007). To date the PDF positive s-LNvs and the CRY1-positive DN1ps are believed to control the M activity (M cells) and the 5th s-LNv, 3-4 LNds, and possibly CRY1-negative DN1ps are thought to control the E peak (E cells, Review: Yoshii et al. 2012). The M cells can drive rhythmicity in DD but not LL and their molecular clock is accelerated by light, while the E cells can drive rhythmicity in LL but not DD and their clockwork is decelerated by light. However, E cells can also contribute to M activity and M cells to E activity. Apparently, the initial model of M and E cells was too simplified. The clock neurons constitute a flexible network of interacting oscillators, which drive rhythmicity. The properties of different cells in this network are affected in different ways by environmental conditions and the hierarchical order may change (Yoshii et al. 2012; Muraro et al.
2013b).
15
The pdf gene was isolated and characterized in 1998 (Park and Hall 1998). Surprisingly, no cycling of pdf RNA could be detected although the peptide exhibited cycling in the terminals of the s-LNvs with a peak around lights-on (ZT 1) and a trough around lights-off (ZT 13, Park and Hall 1998; Park et al.
2000). Expression of pdf was shown to be regulated by different clock proteins: CLK and CYC appear to positively regulate transcription in an indirect, E-box-independent manner, while VRI acts negatively and CLK-independently on a post-transcriptional level (Blau and Young 1999; Park et al.
2000; Allada et al. 2003). Since null mutations of the genes per and tim did not affect RNA levels but abolished the cycling of the peptide, these proteins were suggested to act on a post-translational level (Park et al. 2000). Additional evidence for the requirement of clock genes was provided by overexpression studies: While ectopic expression of pdf in non-clock neurons with projections in the dorsal protocerebrum rendered the flies arhythmic, overexpression in clock neurons neither disrupted PDF cycling nor rhythmicity (Helfrich-Förster et al. 2000).
In the LNvs PDF can be detected in large dense core vesicles (DCVs) in PDF-immunoreactive (PDF-ir) varicosities, indicating a paracrine release mode (Miskiewicz et al. 2004; Yasuyama and Meinertzhagen 2010). Released PDF is metabolically inactivated by neprilysin-like peptidases, which were suggested to play an important role in terminating and thus regulating PDF signaling (Isaac et al. 2007). Next to PDF-ir DCVs, small, clear vesicles that do not contain PDF can be found in output synapses of the s-LNvs. Thus, at least the s-LNvs also express a classical neurotransmitter next to PDF (Yasuyama and Meinertzhagen 2010). Consequently, PDF-independent functions of the LNvs like regulation of cocaine sensitivity (Tsai et al. 2004) or larval light avoidance (Mazzoni et al. 2005) have been detected. In addition, classical neurotransmitter release contributes to the effects of PDF autoreceptor activation (see 1.3.5.4, Choi et al. 2012). However, for the control of circadian rhythmicity PDF release appears to be the main output signal of the LNvs and the classical neurotransmitter appears to play only a minor role (Renn et al. 1999; Blanchardon et al. 2001; Wu et al. 2008a; Umezaki et al. 2011).
PDF cycling in the s-LNv terminals was suggested to indicate rhythmic PDF release from these neurons, acting as a crucial output signal (Park et al. 2000). Indeed, it was shown that PDF accumulation affects the M activity bout, providing evidence that PDF secretion sets the activity phase (Wu et al. 2008b). In addition, there are studies derogating the importance of rhythmic PDF release: Expression of a fusion protein of atrial natriuretic factor (ANF) and green fluorescent protein (GFP) in the LNvs abolished PDF cycling in the s-LNv terminals while molecular oscillations of the clock proteins were not affected and the flies were fully rhythmic, suggesting only a minor role for PDF cycling (Kula et al. 2006). Similarly, phase-independent, constant PDF receptor (PDFR) activation (see 1.3.5.4) was sufficient to induce strong but abnormal rhythmicity (Choi et al. 2009).
Apparently, PDF accumulation and release is dependent on the electrical activity of the s-LNvs. In the morning, when PDF release is believed to be highest, the s-LNv's resting membrane potential and electrical activity in their dorsal terminals are highest, too (Cao and Nitabach 2008; Cao et al. 2013).
Therefore, it is not surprising, that impairment of the electrical activity abolished PDF cycling:
Silenced neurons displayed constitutive low levels of PDF in the terminals but high levels in the somata (Nitabach et al. 2005; Depetris-Chauvin et al. 2011), while hyperexcited neurons displayed constitutive high PDF levels in the terminals (Nitabach et al. 2006), each correlated with severe impairments of circadian behavior (see 1.3.4).
16
1.3.5.2 About 60 % of the clock neurons express the PDF receptor
In 2005 three groups isolated and characterized the PDFR (CG13758, also termed HAN or groom of PDF, GOP), which is a class II peptide G protein-coupled receptor (GPCR) belonging to the secretin receptor-like B1 subfamily, typically signaling via Gαs and Ca2+ (Hyun et al. 2005; Lear et al. 2005b;
Mertens et al. 2005). No cycling of pdfr RNA or protein could be detected. Instead it was shown to be expressed at a steady-state level, which is per-dependent (Lear et al. 2005b; Mertens et al. 2005).
While immunocytochemistry and in-situ hybridizations did not provide reliable results, the expression pattern could be characterized via PDF-dependent cAMP elevations as well as pdfr promoter driven expression of a reporter gene (Shafer et al. 2008; Lear et al. 2009; Im and Taghert 2010). The PDFR is expressed in about 60 % of all clock neurons (Review: Taghert and Nitabach 2012), such as all s-LNvs, at least two l-LNvs, most LNds, most DN1s, both DN2s, and most DN3s.
Additionally, it is expressed in many non-clock cells of the brain, such as glial cells of the visual system, cells between the tritocerebrum and the subesophageal ganglion (Im and Taghert 2010), cells of the antennal mechanosensory and motor center (AMMC, Vecsey et al. 2013), and cells of the ellipsoid body (EB) of the central complex (Parisky et al. 2008; Pírez et al. 2013). Outside the brain PDFR expressing cells were also detected, for example in ureter muscles and in the hindgut and midgut muscles (Talsma et al. 2012), in pheromone producing peripheral clock cells, the oenocytes (see 1.3.5.5, 1.5.2, Krupp et al. 2013), in the thoracic ganglia (Vecsey et al. 2013), and in the crop, suggesting PDF release into the hemolymph (Veenstra et al. 2008).
1.3.5.3 The PDF receptor couples to G
αsand probably also to G
αqTypically for GPCRs of the B1 subfamily the PDFR has been shown to couple to Gαs and to mediate PDF-dependent increases in the cAMP-level when expressed heterologously in cell cultures (Hyun et al. 2005; Mertens et al. 2005), transgenically in larval motor neurons (Vecsey et al. 2013), or in situ (Shafer et al. 2008; Choi et al. 2012; Duvall and Taghert 2012; Talsma et al. 2012; Duvall and Taghert 2013; Pírez et al. 2013). The components of the signaling cascade are packed to signalosomes, whose components differ between clock cells: In the PDF-positive s-LNvs (M cells, see 1.3.4) Gαs60A, adenylyl cyclase AC3, and the A-kinase anchoring protein (AKAP) NERVY are required. In the CRY1-positive LNds (E cells, see 1.3.4) AC3 only plays a minor role, and mainly AC78C and two AKAPs (NERVY and AKAP200) are involved (Duvall and Taghert 2012, 2013). Interestingly, the diuretic hormone 31 receptor (DH31R), which is expressed in the s-LNvs, also couples to Gαs60A in these neurons, but requires a different AC isoform, indicating highly specific signalosomes (Choi et al. 2012;
Duvall and Taghert 2012). Next to increases in the cAMP level, PDF-dependent Ca2+ concentration rises were reported for human embryonic kidney (HEK 293) cells heterologously expressing PDFR (Mertens et al. 2005). In this study, coexpression of the neurofibromatosis-1 gene product (NF1) enhanced the sensitivity and the strength of PDF responses and thus, was suggested to be involved in coupling of PDFR to AC. However, there are studies indicating different signaling cascades for PDF: In one study PDF was suggested to couple to the RAS/MAPK pathway, regulated by NF1 (Williams et al.
2001), and more recently first evidence was provided for PDFR coupling to Gαq and thus phospholipase C (PLC) signaling (Agrawal et al. 2013). It was suggested that PDFR may couple to both Gαs and Gαq. In one study, loss of the pdfr (pdfr5304 flies) surprisingly did not mitigate the effects of
17
constant PDFR activation in oenocytes of D. melanogaster, which could indicate unspecific responses or PDF-binding to a second, yet unidentified PDFR (Krupp et al. 2013).
Further components of the PDF-induced signaling pathway have not been reported so far. However, typical cAMP targets, such as cyclic nucleotide gated (CNG) ion channels, hyperpolarization-activated cyclic nucleotide-gated (HCN) cation channel, protein kinase A (PKA), as well as the guanine nucleotide exchange factor (GEF) "exchange protein directly activated by cAMP" (EPAC, Review:
Gloerich and Bos 2010) will likely be involved (Fig. 9).
Fig. 9. PDF signals via Gαs and probably also via Gαq. A. In clock neurons of D. melanogaster the PDF receptor (PDFR) couples to a trimeric Gs protein. PDF binding leads to separation of the βγ-subunit from the Gαs subunit, which activates adenylyl cyclase (AC). Hydrolysis of ATP elevates the cAMP level. cAMP might directly activate ion channels as well as protein kinase A (PKA), which might also activate or inhibit ion channels or translocate to the nucleus and modulate the transcriptional translational feedback loop (TTFL). Additionally, the "exchange protein directly activated by cAMP" (EPAC) might be activated, which activates the monomeric G protein RAP and thereby increases the target range of cAMP. B. In cells of the subesophageal ganglion, the thoracic ganglion, and the antennal mechanosensory and motor complex of D. melanogaster PDF signaling is required for the regulation of flight. In these cells the PDFR is suggested to couple to the Gαq signaling pathway, activating phospholipase Cβ (PLCβ), which catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2) generating inositol 1,4,5-trisphosphate (IP3) and diacylglycerol (DAG). The increase of IP3 activates IP3
receptors, leading to Ca2+ efflux from the endoplasmatic reticulum (ER). In combination with DAG the elevated Ca2+
concentration might activate protein kinase C (PKC), which in turn could activate or inhibit ion channels or modulate the TTFL. Additionally, DAG may directly activate ion channels.
18
Active PKA might phosphorylate target proteins, such as ion channels, in the cytoplasm or translocate into the nucleus. It might activate cAMP response element-binding protein (CREB), which binds to cAMP response elements (CREs) within promoters of different genes, thereby regulating these genes' transcription. CRE domains in insect clock gene promoters, as shown for the mammalian mPer1 gene (Hida et al. 2000), would provide an effective mechanism for PDF-dependent gene regulation.
1.3.5.4 PDF signaling is required for robust molecular cycling and synchrony
Electrogenic responses to PDF have not been characterized in detail to date. Constant autoreceptor activation in the s-LNvs was shown to disrupt VRMP oscillations of the s-LNvs and to depolarize the cells in the presence or absence of TTX, suggested to block VASCs as well as synaptic transmission (Choi et al. 2012). Recently, acute PDF application on larval motoneurons transgenically expressing PDFR was shown to depolarize the membrane and increase the excitability of the cells (Vecsey et al.
2013). Moreover, acute PDF application was shown to depolarize DN1p neurons in an adenylyl cyclase-dependent but PKA-independent manner, probably via activation of CNG channels (Seluzicki et al. 2014). However, a detailed characterization of the involved ion channels has not been reported yet.
How does PDF contribute to control circadian rhythms? The behavioral phenotypes for ablation of the PDF expressing LNvs (Renn et al. 1999; Blanchardon et al. 2001; Shang et al. 2008), electrical silencing of these neurons (Nitabach et al. 2002; Wu et al. 2008a; Depetris-Chauvin et al. 2011), PDF knockdown via RNA interference (RNAi) in the LNvs (Shafer and Taghert 2009), as well as pdfr mutant flies (Hyun et al. 2005; Lear et al. 2005b; Mertens et al. 2005; Lear et al. 2009; Im and Taghert 2010) resemble the phenotype of pdf01 mutant flies (Renn et al. 1999; Peng et al. 2003; Lin et al. 2004): In LD conditions the M anticipation is lacking and the E peak is phase-advanced. In DD the M peak is absent, the flies display a shortened period length of about 23 h, and the majority of flies become arhythmic after 2 - 3 days. However, in contrast to pdf01 mutant flies, pdfr mutant flies do not lack the M peak consistently (Im and Taghert 2010). Molecular oscillations in all clock neurons in pdf01 mutant flies were shown to dampen in DD, which correlated with the dampening of behavioral rhythmicity. This indicates that PDF is required for robust, high-amplitude cycling in DD (Peng et al.
2003; Klarsfeld et al. 2004). Moreover, PER oscillations in the s-LNvs and LNds of pdf01 mutant flies were shown to continue in DD, but the s-LNvs were shown to desynchronize relative to each other and the phase of the LNds to advance starting from the third day in DD. This indicates that PDF is required for coordination of the phase and thus synchronization of different clock neurons (Lin et al.
2004). More detailed analyses revealed that PDF affects clock cell groups differently: It is required for synchronous cycling in the s-LNvs and DN1as and lengthens the period of these cells. It also lengthens the period of the 5th s-LNv and the CRY1-positive LNds, but shortens the period of the CRY1-negative LNds. PDF is crucial for molecular cycling in the CRY1-positive DN1ps, but does not obviously affect the phase or period of the CRY1-negative DN1ps, the DN2s, and DN3s (Wülbeck et al.
2008; Choi et al. 2009; Lear et al. 2009; Yoshii et al. 2009). PDF signaling from positive to PDF-negative cells and thus PDFR activation in PDF-PDF-negative cells is particularly important to ensure robust rhythmicity and to set the period length in DD (Choi et al. 2009; Lear et al. 2009). Constitutive PDF-autoreceptor activation in the LNvs dose-dependently increases M anticipatory activity, shifts the daily activity to the subjective morning ("increases morningness"), and shortens the period
19
(Choi et al. 2012). Therefore, PDF signaling to both PDF-positive and -negative cells appears to be implicated in setting the period length in DD. Consistently, male flies have higher pdf expression levels and show an earlier M peak compared to female flies (Park and Hall 1998; Helfrich-Förster 2000). Recent studies suggested that the effects of PDF on the molecular clock are mediated via PKA-dependent mechanisms, either promoting stabilization of PER (Li et al. 2014a) or TIM (Seluzicki et al.
2014). The source of PDF release is important too: Release from the s-LNvs is crucial for maintenance of rhythmicity in DD, while release from s-LNvs and l-LNvs regulates the period length in DD and the timing of the evening peak in LD (Shafer and Taghert 2009). Interestingly, PDF levels were shown to decline age-dependently, correlated with attenuations of TIM cycling, period lengthening, and reduced rhythmic strength. Since PDF overexpression was shown to suppress these age-dependent changes, it was suggested that age-dependent PDF decline is responsible for rhythm attenuation of older flies (Umezaki et al. 2012). PDF is not only important for the synchronization of clock cells in the brain. It is also required for synchronization of peripheral clocks, as shown for prothoracic glands (PG, Myers et al. 2003) and oenocytes (Krupp et al. 2013).
1.3.5.5 Additional functions of PDF
Next to its function as synchronization signal in the circadian clock, PDF has been shown to be required for a multitude of additional functions: geotaxis (Toma et al. 2002; Mertens et al. 2005), rival-induced prolonged mating (Kim et al. 2013), LNv-circuit development (Gorostiza and Ceriani 2013), development of the flight-circuit and modulation of flight (Agrawal et al. 2013), myotropic effects like ureter contraction (Talsma et al. 2012), as well as photoperiodic responses (Yoshii et al.
2009). Lack of PDF disrupts male courtship behavior (Fujii and Amrein 2010) and females show increased PDF expression in response to exposure to the male courtship song (Immonen and Ritchie 2012). Additionally, PDF release of both LNv classes plays a major role in the regulation of sleep and arousal (Parisky et al. 2008; Shang et al. 2008; Sheeba et al. 2008c; Chung et al. 2009; Sheeba et al.
2010; Pírez et al. 2013). While the l-LNvs promote arousal and less sleep at night (Shang et al. 2008;
Sheeba et al. 2008c), the s-LNvs are implicated in sleep onset and probably inhibit arousal (Sheeba et al. 2008c; Sheeba et al. 2010). PDF signaling to the LNvs themselves and to cells of the ellipsoid body of the central complex has been shown to be essential in the sleep-regulation circuit (Parisky et al.
2008; Pírez et al. 2013).
1.3.5.6 Parallels between PDF and VIP signaling
Remarkably, there are striking similarities between PDF signaling in insects and vasoactive intestinal peptide (VIP) signaling in the suprachiasmatic nucleus (SCN), the circadian clock of vertebrates (Reviews: Vosko et al. 2007; Meelkop et al. 2011; Taghert and Nitabach 2012). VIP is a 28 amino acid long peptide, which is expressed in the brain, the peripheral nervous system as well as other body tissues, such as the digestive tract. In the SCN, considered as the "master clock" in vertebrates, VIP is expressed by around 10 % of the cells, mainly located in the ventrolateral core region (Ibata et al.
1989; Abrahamson and Moore 2001; Vosko et al. 2007; Taghert and Nitabach 2012). VIP and pituitary adenylyl cyclase-activating polypeptide (PACAP) both belong to the secretin-superfamily and activate
20
the same receptors: VPAC1R and VPAC2R (Ishihara et al. 1992; Lutz et al. 1993; Usdin et al. 1994;
Vosko et al. 2007). VPAC2R expression is highest in the SCN, where it is expressed amongst others by around 30 % of the VIP-positive cells (Usdin et al. 1994; Kallo et al. 2004b). Thus, VIP also signals via autoreceptors, as shown for PDF (An et al. 2012; Choi et al. 2012; Taghert and Nitabach 2012). In contrast to PDF and PDFR, VIP and VPAC2R are both expressed in a circadian manner (Shinohara et al.
1999; Dardente et al. 2004; Kallo et al. 2004a), and similar to PDF, VIP is released in a circadian manner (Shinohara et al. 2000a; Vosko et al. 2007). While VIP does not show any sequence identity with the shorter PDF, the PDFR as well as VPAC1R and VPAC2R are related members of the secretin receptor-like B1 subfamily of GPCRs (Mertens et al. 2005; Vosko et al. 2007; Couvineau and Laburthe 2012). Consistently, the PDFR was shown to be also activated by PACAP-38, when heterologously expressed in HEK 293 cells (Mertens et al. 2005). As shown for PDFR activation, VPAC2R couples to Gαs and activation leads to increased cAMP levels and activation of PKA (Rea 1990; Vanecek and Watanabe 1998; Itri and Colwell 2003; Meyer-Spasche and Piggins 2004; Kudo et al. 2013).
Additionally, VIP signaling was shown to be dependent on AC signaling as well as PLC signaling and it was suggested that VPAC2R couples to Gαs and Gαq (An et al. 2011), as suggested previously for the PDFR (Agrawal et al. 2013). Most interestingly, the phenotypes of mice mutants lacking either VIP (Colwell et al. 2003) or its VPAC2 receptor (Harmar et al. 2002) are very similar to those of pdf or pdfr null mutant flies (Renn et al. 1999; Peng et al. 2003; Lin et al. 2004; Hyun et al. 2005; Lear et al.
2005b; Mertens et al. 2005; Lear et al. 2009; Im and Taghert 2010): Most of these mice behave arhythmic in DD and those mice that retain weak rhythmicity, display a shortened period length.
Molecular oscillations of several clock genes display low amplitudes, single cells loose rhythmicity, and synchrony between different SCN cells is disrupted (Harmar et al. 2002; Maywood et al. 2006).
Additionally, synchronous firing and characteristic rhythms in the neural activity of SCN neurons are abolished (Cutler et al. 2003; Aton et al. 2005; Brown et al. 2007). Thus, similar to PDF signaling, VIP signaling is required for synchronous output of the SCN. Beyond the circadian master clock there are similarities, too: Both peptides have endocrine, myotropic functions in the digestive tract or urogenital system (Talsma et al. 2012 and references mentioned therein).