The Satiety Signaling Neuropeptide Perisulfakinin Inhibits the Activity of Central Neurons Promoting General Activity

The metabolic state is one of the determinants of the general activity level. Satiety is related to resting or sleep whereas hunger correlates to wakefulness and activity. The counterpart to the mammalian satiety signal cholecystokinin (CCK) in insects are the sulfakinins. The aim of this study was to resolve the mechanism by which the antifeedant activity of perisulfakinin (PSK) in Periplaneta americana is mediated. We identified the sources of PSK which is used both as hormone and as paracrine messenger. PSK is found in the neurohemal organ of the brain and in nerve endings throughout the central nervous system. To correlate the distributions of PSK and its receptor (PSKR), we cloned the gene coding for PSKR and provide evidence for its expression within the nervous system. It occurs only in a few neurons, among them are the dorsal unpaired median (DUM) neurons which release octopamine thereby regulating the general level of activity. Application of PSK to DUM neurons attenuated the spiking frequency (EC50=11pM) due to reduction of a pacemaker Ca2+ current through cAMP-inhibited pTRPγ channels. PSK increased the intracellular cAMP level while decreasing the intracellular Ca2+ concentration in DUM neurons. Thus, the satiety signal conferred by PSK acts antagonistically to the hunger signal, provided by the adipokinetic hormone (AKH): PSK depresses the electrical activity of DUM neurons by inhibiting the pTRPγ channel that is activated by AKH under conditions of food shortage.


INTRODUCTION
The metabolic state is related to general activity. While energy deficiency is related to wakefulness and activity thereby supporting behavioral tasks to overcome this potentially dangerous state, energy excess is related to resting or sleep. People say you cannot study on a full stomach. Although brain areas and messengers involved in this kind of homeostatic regulation have been identified (Castaneda et al., 2005;VanItallie, 2006), there is less information on underlying mechanisms at the cellular and molecular level. For the study of such mechanisms insects are particularly well-suited models because of the design of their nervous systems. Distinct functions are conferred by identified neurons which is in contrast to the mainly parallel processing within the vertebrate nervous system.
One of the messengers involved in regulating metabolic homeostasis in mammals is cholecystokinin (CCK). This peptide occurs in gut and brain and affects various tasks such as feeding, general activity, memory, and emotional states (Noble and Roques, 2002). When activating the CCK1 receptor, CCK acts as satiety signal and stops food-intake (Strader and Woods, 2005). CCK also reduces general activity thereby promoting sleep (Kapas et al., 1988).
Sulfakinins are insect peptides being structurally and functionally related to the mammalian CCK (Dockray, 2004). They belong to the large family of RF-amides (Nichols, 2003). These peptides bear a common RFamide C-terminus but diverse N-terminal extensions. They are involved in the regulation of various physiological processes in invertebrates and vertebrates. There are two sulfakinins in the cockroach Periplaneta americana, perisulfakinin I (PSK I, EQFDDY-(SO 3 )-GHMRF-amide), which is characterized by a sulfated tyrosine, and the unsulfated PSK II (pQSDDYGHMRF-amide (Veenstra, 1989). PSK I (further referred to as PSK) elicits hindgut contractions while PSK II does not. Also in other insects, such as in Drosophila (Kubiak et al., 2002), unsulfated sulfakinins show no or very weak biological activity. A large variety of physiological responses have been reported for sulfakinins, for example, myotropic effects (Nichols et al., 2002). Antifeedant activity as seen for CCK was also shown for sulfakinins in the locust (Wei et al., 2000) and in the German cockroach (Maestro et al., 2001).
The aim of this study was to unravel the cellular and molecular mechanism of PSK within the nervous system. For this purpose, we have cloned the perisulfakinin receptor (PSKR) and provided evidence that it is expressed within the central nervous system. However, only in a few neurons the receptor is localized in the soma membrane. Among them are dorsal unpaired median (DUM) neurons which belong to the octopaminergic system that-as insect counterpart to the mammalian adrenergic system-is involved in the regulation of general activity (Roeder, 2005) as well as in the modulation of peripheral structures such as visceral, heart, and skeletal muscles (Brä unig and Pflü ger, 2001). The somata of efferent DUM neurons show, in contrast to most other insect neurons, endogenous electrical activity (Grolleau and Lapied, 2000). Investigations on isolated DUM cell somata indicate that they express various ionotropic and metabotropic receptors (e.g., Washio, 1994;Washio and Tanaka, 1992;Wicher et al., 2006b) despite the fact that there are no synaptic structures. By demonstrating that PSK might be released within a cisterneal system of the extracellular matrix from which it diffuses to the soma membrane we propose a mechanism by which ligands may reach their receptors.
We further show by which mechanism PSK attenuates the excitability of DUM cells. Most intriguingly, PSK down-regulates the same ion current that is up-regulated by adipokinetic hormones (AKHs) which are released under energy deficient conditions (Wicher, 2007;Wicher et al., 2006b).
Localization of PSK and the PSKR A nested PCR approach was used to show PSKR expression in neuronal tissue and in single DUM neurons. The following primers were used for single-cell PCR: forward: 5 0 -CGTCGTCATTTTTGTGCTGT-3 0 ; 5 0 -TTCTTA-ATCTGGCGGTGTCC-3 0 ; reverse: 5 0 -AACATGGCGTCCAAGAAAAT-3 0 ; 5 0 -TTTGGATGGCCAATCTTCTC-3 0 . For single-cell RT-PCR analysis the cytosol of single DUM neurons was harvested with patch pipettes and directly transferred to the RT reaction mixture. RNA from P. americana brain and ganglia were isolated using the RNeasy Mini-Kit (Qiagen, Hilden, Germany). RT-reaction was performed with Sensiscript reverse transcriptase (Qiagen). The nested PCR reactions were performed with Advantage Taq 2 polymerase mixture (Clontech, Hamburg, Germany) with initial 60 seconds 948C denaturation followed by 40 cycles 30 seconds 948C, 30 seconds 558C, and 60 seconds 688C. PCR products of the second PCR reaction were visualized on a 2% agarose gel. As a positive control we amplified part of a ubiquitously expressed actin gene from P. americana in a nested PCR approach using a duplex reaction (PSKR and actin-PCR) as first PCR. To exclude false positive results due to genomic DNA contamination PCR-primers were intron spanning according to the Drosophila and Anopheles genomic sequences. All PCR products were directly sequenced to validate specificity for PSKR.
Two different antibodies to PSK were raised, one in rabbits and one in mice, each to a conjugate of the peptide EQFDDY-(SO 3 )-GHMRF-amideglutardialdehyd-thyroglobulin. PSK was synthesized by Bachem Holding AG (Bubendorf, Switzerland). Coupling of peptides to carrier proteins using glutardialdehyde, the immunization protocol for rabbits and mice and the purification of antisera were performed according to (Harlow and Lane, 1988).
The antibody to PSKR used for immunocytochemical demonstration as well as for immunoblots were raised in rabbits against a conjugate of the peptide motif AIKETERHKCR located in the loop between TMS4 and TMS5 coupled to kelim limpet hemocyanin. The peptide was synthesized and coupled to the carrier by Biotrend (Cologne, Germany).
Adult and embryonic P. americana were taken from laboratory colonies at the University of Jena (maintained at 278C under 12/12 hours light/dark regime). Animals were anesthetized by cooling under crushed ice and decapitated. Whole ventral nerve cords, midguts, and the sixth abdominal ganglia were dissected from male animals under freshly prepared fixative: 4% formaldehyde in 0.1 M Millonings buffer (pH 7.35) overnight at room temperature.
Vibratome sections. The preparations were washed in Tris-HCl buffer (in millimoles: 146 NaCl, 50 Tris-OH; pH 7.4) for 12 hours at 48C on a shaker. Thick sections (40 mm) of agar-embedded midguts and sixth abdominal ganglia were cut on a vibratome (Technical Products, St. Louis, USA). The washed free-floating slices were then incubated in rabbit anti-PSKR polyclonal antibody diluted 1:500 and in a mouse anti-PSKpolyclonal antibody diluted 1:1000 in a solution of Tris-HCl buffer containing 2% normal goat serum, 0.25% Triton-X, 3% skim milk powder, and 0.25% bovine serum albumin (BSA; MPB). Subsequently, the tissue was washed in MPB overnight and incubated in the Cy3-tagged secondary goat anti-rabbit (GAR) and in the Cy2-tagged secondary goat anti-mouse (GAM) antibody (Jackson, ImmunoResearch Laboratories, Inc., West Grove, USA) for 3 hours at a dilution of 1:800 in MPB.
Whole mounts. The ventral nerve cords of 22-day-old embryos were washed at 108C for 48 hours and incubated for 48 hours in rabbit anti-PSK polyclonal antibody diluted 1:5000 in MPB. The whole nervous systems were then washed in MPB 48 hours and incubated for 48 hours at 108C in a 1:800 GAR-Cy3 dilution.
The vibratome sections and whole mounts were mounted and embedded in Mowiol (Sigma, Deisenhofen, Germany). For the photographic documentation we used the bright-field optics on a Zeiss Axiophot microscope, a Hamamatsu digital camera C4742-95, and OpenLab software (Improvision Limited, Coventry, UK). We performed the following controls: (a) the primary antibody was omitted or replaced with an equivalent concentration of normal serum (negative controls); (b) the secondary antibody was omitted. Under these conditions, cells and fibers remained totally unstained. The cross-reactivity of the PSK antibody with other neuropeptides was tested by ELISA. In a dilution of 1:20 000 the antibody recognized the PSK-conjugate with maximal intensity (optical density, od ¼ 3.0), yet a cross-reactivity to the insect-neuropeptides allatostatin A1, A3, B, proctolin, locustatachykinin, CCAP, AKH I, and corazonin was not detectable. There was, however, a moderate crossreactivity to FMRF-amide conjugate (od ¼ 1.3) and a very weak immunoreactivity to the leucomyosuppressin conjugate (od ¼ 0.2). Both leucomyosuppressin and FMRF-amide are insect neuropeptides with Cterminal similarity to PSK. As shown in a competitive ELISA, the PSK antibody could solely be inhibited by PSK (total inhibition at 10 pM/well) but hardly by leucomyosuppressin, FMRF-amide, and RF-amide (maximal inhibition by 20% at 1000 pM/well; Veenstra, unpublished result).
Electron microscopy. After preparation of the abdominal tergites and rapid removal of the gut and fat tissue, the sixth abdominal ganglia were prefixed in situ. After 10 minutes, these ganglia were dissected from ventral nerve cord and continually fixed in vitro. As fixative we used 0.5% glutardialdehyde, 2% formaldehyde in 0.1 M HEPES buffer, pH 7.35 for 2 hours at 48C, followed by washing overnight in 0.1 M HEPES buffer, and postfixation with 1% OsO 4 in 0.1 M HEPES buffer for 1 hour. After a brief wash, the tissue was dehydrated in a series of alcohol and embedded in Durcupan ACM. Ultrathin sections were cut with Reichert Ultracut E and mounted on Formvar carbon-coated single-slot nickel grids. Sections were first soaked for 10-20 minutes in drops of PBS containing 0.5% BSA. Incubation with the antibody to PSK (diluted 1:1000) was performed in a moist chamber for at least 24 hours at 48C. After the grids had been washed 7-8 times by self rotation on a magnetic stirrer, incubation followed with gold 12 -labeled GAR IgG (British Biocell, Cardiff, UK) for 12 hours at 48C. After repeated washes with PBS, sections were doublestained with uranyl acetate and lead citrate. Sections were examined with a Zeiss EM 900 electron microscope.
Western blot analysis. Neuronal and intestinal tissue of P. americana was ground under liquid nitrogen, the tissue powder was homogenized in 250 mM sucrose, 10 mM triethanolamin hydrochloride pH 7.5 containing Complete Protease Inhibitor (Roche, Mannheim, Germany) and freed from larger particles by centrifugation at 1000 g for 10 minutes. Finally, the membrane fraction was collected by ultracentrifugation at 200 000 g for 1 hour at 48C. For Western blot analysis 200 mg protein of both membrane fractions were loaded per lane on a 10% SDS-PAGE. After transfer to nitrocellulose membranes a standard immunodetection protocol with our PSKR antisera (1:1000 dilution), anti-rabbit IG-alkaline phosphates conjugate (Vector Lab Inc., Burlingame, CA, USA), and BCIP/NBT color reaction was performed.

Electrophysiology
Isolation of cells was performed as described previously . Briefly, the sixth abdominal ganglia of adult cockroaches (P. americana) were excised, desheathed, and incubated for 10 minutes at room temperature in standard bath solution (composition,in millimoles: 190 NaCl,5 KCl,5 CaCl 2 , 2 MgCl 2 , 10 HEPES; pH 7.4) containing 0.5 mg/ml trypsin (type II, Sigma) and 0.5 mg/ml collagenase (type I, Sigma). After thoroughly washing off the enzyme the ganglia were stored in saline for at least 1 hour. In the sixth abdominal ganglion, four octopaminergic DUM neurons form a cluster (Eckert et al., 1992). From this DUM cell cluster, which can be easily identified, the neurons were separated using thin metal needles.
Ionic currents of isolated DUM neurons were measured at room temperature using whole-cell patch-clamp. Pipettes having resistances of 0.5-0.8 MV were pulled from borosilicate capillaries (Hilgenberg, Malsfeld, Germany). Current measurements and data acquisition were performed using an EPC9 patch-clamp amplifier controlled by PULSE software (HEKA Elektronik, Lambrecht, Germany). Data were sampled at 10 kHz and filtered at 2.9 kHz. Capacitive and leakage currents were compensated using an on-line P/n protocol with a holding potential for leakage measurement of À110 mV. The series resistance remaining after compensation did not exceed 1.3 MV. The holding potential (V hold ) was À90 mV.
Spiking of neurons was recorded under current-clamp conditions without current injection. Neurons were bathed in saline (cf. Cells) and patch pipettes (resistance >1.5 MV) were filled with a solution composed of (in millimoles) 190 K-gluconate, 5 NaCl, 1 CaCl 2 , 3 EGTA, 2 Mg-ATP, and 10 HEPES (pH 7.25). Between recordings (duration 1 second) the cells were held under voltage clamp at a holding potential of À70 mV.
Application or washout of blocking agents was performed by transferring the cells (attached to the pipette tip) within a glass tube into the various solutions. A complete and fast solution exchange within a few milliseconds was achieved by sucking a small amount of solution into the tube.
Data are given as means AE standard deviation (n ¼ number of cells) unless otherwise stated. For data analysis including nonlinear fitting procedures Prism 4 software (Graph Pad Software, San Diego, CA, USA) was used. Current-voltage (I À V ) relationships for Ca 2þ peak currents were fitted taking into account current rectification according to the Goldman-Hodgkin-Katz (GHK) equation where G is the total Ca 2þ conductance, V rev the Ca 2þ reversal potential, V 0:5 the potential for half-maximal activation of total Ca 2þ conductance, and S the slope factor of activation (cf. Hille, 1992). Activation curves were fitted with the Boltzmann equation where G max is the maximal conductance, V 0:5 the voltage of half-maximal activation, and S the slope factor.
cAMP assay DUM neurons were isolated as described in the previous section; 2 hours after isolation cultures were treated with 10 nM PSK (10-25 minutes) and/ or antisera anti-PSKR 1:100 (1 hour). Subsequently, neurons were fixed for 1 hour with 4% paraformaldehyde containing 5% (w/v) sucrose in phosphate-buffer saline (PBS, pH 7.4). After fixation, cells were washed three times for 5 minutes each in PBS and 5 minutes in PBS containing 0.2% Triton X-100 (PBS-T). To block non-specific binding of the primary antibody, cell bodies were preincubated with 4% BSA in PBS-T for 1 hour. Primary antisera (rabbit anti-cyclic-AMP polyclonal antibody, Euromedex, Mundolsheim, France), diluted 1:800 in PBS-T, was applied overnight at 48C. After repeated washing in PBS-T, the secondary antibody fluorescein isothiocyanate-labeled GAR IgG (Euromedex) diluted 1:150 in PBS-T containing 1% BSA was applied at 208C for 3 hours in the dark. Isolated DUM neuron cell bodies were then rinsed in 4% BSA in PBS and mounted on glass slides in glycerol/PBS. The fluorescence detection was captured using an Axiocam HRC camera mounted on a Zeiss Axioskop microscope; images were digitized using Axiovision (Carl Zeiss, Jena, Germany). The immunofluorescence quantification used to compare intensity of fluorescence was performed with ImageJ 1.34 s software (Scion corporation, Frederick, USA).

Measurement of intracellular calcium concentration
Cells were loaded with Fura-2 (Molecular Probes, Eugene, OR, USA) either by incubation in standard bath solution (cf. above) containing 2 mM Fura-2/acetomethylester (AM) for 20 minutes or by intracellular perfusion with 10 mM K 5 -Fura-2 added to the pipette solution used for spike recording (for calibration). In order to preserve cellular integrity most experiments were performed with Fura-2/AM loading.

Satiety signaling via sulfakinin
Free intracellular Ca 2þ concentration [Ca 2þ ] i was determined with the fluorescence ratio method. Light exciting Fura-2 at 340 and 380 nm, provided by Polychrome II (T. I. L. L. Photonics, Grä felfing, Germany), was coupled via an epifluorescence condenser into an Axioskop FS (Carl Zeiss) equipped with a water immersion objective (LUMPFL 40Â W/IR/0.8; Olympus, Hamburg, Germany). Emitted light was separated by a 400 nm dichroic mirror and filtered with a 420 nm long-pass filter. ½Ca 2þ i was calculated according to the following equation: K eff , R min , and R max were determined using DUM neurons permeabilized with 2 mM ionomycin and three solutions with different concentrations of free Ca 2þ (Ca 2þ free; 5 mM Ca 2þ ; 500 nM Ca 2þ ; the composition of the 500 nM solution was calculated with WEBMAXC v.2.20, Patton et al., 2004). The values of K eff , R min , and R max were 3.72 mM, 0.38, and 5.9, respectively. Fluorescence images were acquired using a cooled CCD camera controlled by TILLVision 4.0 software (T. I. L. L. Photonics). The resolution was 640 Â 480 pixel in a frame of 175 Â 130 mm 2 (40Â/IR/0.8 objective). Image pairs were obtained by excitation for 100 ms at 340 and 380 nm, for recording of image series intervening intervals of 10 seconds were chosen. Background fluorescence was subtracted.

RESULTS
Localization of PSK within the nervous system Sulfakinin-immunoreactive (ir) cells have been described in different insect species (e.g., Agricola and Brä unig, 1995;East et al., 1997). In the American cockroach, a specific antibody against PSK recognizes only a few neurons (Agricola and Brä unig, 1995;East et al., 1997). There are 24 PSK-ir cells in the pars intercerebralis that project into the corpora cardiaca ( Figure 1A). Using the immunogold technique we stained PSK-ir axon terminals in the storage lobe of the corpus cardiacum indicating that PSK acts as neurohormone ( Figures 1B and 1C). Two types of axon profiles differing in the appearance of granula were strongly labeled ( Figures 1B and 1C). A third type showed weak PSK-ir which may result from a RF-amide cross-reactivity (not shown).
Furthermore, there is one pair of PSK-ir neurons in the brain, so-called protocerebral descending sulfakinin-ir (PDS) neurons (Agricola and Brä unig, 1995;Figures 1D and 1F). The processes of PDS neurons run contralaterally and unidirectionally through the ventral nerve cord. While they do not form arborizations within all ganglia at early embryonic development ( Figures 1D and 1F), three bilateral main branches sprout per ganglion at later stages. Extensive varicosities appear within the branched neurites of PDS neurons during further development as shown for the sixth abdominal ganglion ( Figure 1F).
Double stainings using a polyclonal rabbit antibody to PSK ( Figure 2A) and a monoclonal mouse antibody to octopamine ( Figure 2B) visualized the varicosities of PDS neurons as well as the four octopaminergic DUM neurons with their arborizations. Electron microscopical immunocytochemistry revealed two different axon profiles showing PSK-ir. Intense staining was observed in fibers without synaptic contacts ( Figure 2E). The other one is equipped with synaptic vesicles and displayed only Other PSK-ir neurons are described in (Agricola and Bra¨unig, 1995). SEG, subesophageal ganglion; PTG, prothoracic ganglion; MoTG, mesothoracic ganglion; MaTG, metathoracic ganglion; AG, abdominal ganglion; TG, terminal ganglion. Bar, 200 mm. (F), PSK-immunostaining (whole-mount) in the sixth AG of an adult animal. Note the intense staining of the PDS1 neurite arborizations characterized by varicosities. At each side three bilateral main branches sprout from the primary neurite. Bar, 100 mm. moderate staining presumably due to cross-reactivity to RF-amide peptides ( Figure 2F). There is no direct innervation of the somata of DUM neurons by the PDS neurons ( Figures 2C and 2D). However, in all preparations (n ¼ 28) PSK-immunoreactivity (PSK-ir) appears within the extracellular matrix, the interface between the neuropile and the soma cell layer ( Figure 3A). Tiny extensions of this extracellular matrix penetrate the glial layer that surrounds the somata of neurons. This might attribute to the PSK-ir observed near the somata ( Figures 2C and 2D).
The presence of PSK within the extracellular matrix indicates that the peptide may be also secreted extrasynaptically as proposed in the concepts of volume transmission (Zoli and Agnati, 1996) or extrasynaptic transmission (Vizi et al., 2004). Using electron microscopy, we detected PSK-ir material within the extracellular matrix channel system ( Figures 3A and 3C). Source of this immunoreactivity is probably peripheral axon terminals containing PSK-ir ( Figure 3B). Interestingly, there is also immuno-gold staining within the extracellular space surrounding these axons ( Figure 3C). These morphological features support the view that PSK may also act via volume transmission.
Molecular cloning and localization of the PSK receptor Using a combined degenerated PCR and nested RACE strategy we obtained a 1945 bp cDNA of the PSKR (GenBank acc no. AY865608). This cDNA encodes a 1866-bp open-reading frame for a 622 amino acid protein with significant homology to two Drosophila (23.6% identity, 32.3% homology to AAF48875; 29.8% identity, 41.3% homology to AAF48879) and one Anopheles (41.9% identity, 51.4% homology to AAR28375) sulfakinin receptors. In addition, all typical structural features of a G-protein coupled receptor were identified, including seven transmembrane regions, an extracellular N-terminal region and a cytosolic C-terminus. One noticeable feature of all sulfakinin receptors is the large intracellular linker between TMS5 and TMS6.
A nested PCR approach was used to identify the expression of the PSKR in neuronal tissue. In brain and in thoracic and abdominal ganglia a prominent 362-bp PSKR PCR product was obtained ( Figure 4A), verified by sequencing. In addition, in 7 out of 10 isolated actin-positive DUM neurons a PCR signal was obtained ( Figure 4B), indicating a significant expression of PSKR in these cells.
A Western blot analysis showed a prominent band at the expected molecular weight of 67.8 kD (arrow in Figure 4C) in the membrane fraction of intestine tissue. In neuronal tissue no specific signal was found indicating that PSKR expression in neuronal tissue is low. This was confirmed by immunocytochemistry ( Figure 5). Pronounced PSKRimmunostaining was seen both in midgut and hindgut (not shown). Within the nervous system we found very weak PSKR-ir. Only a few neuronal cell bodies such as those of the DUM neurons ( Figure 5B) showed any immunostaining, which was weak compared with that in the gut cells. Within the neuropile, commissural and longitudinal tracts showed PSKRir, presumably associated with axonal membranes ( Figure 5C). Interestingly, we also found immunostaining within the extracellular channel system surrounding the somata (cf. Figure 3A).

PSK affects spiking of DUM neurons by modulating ion currents
The presence of PSK-like ir nerve endings near the surface of the DUM neuron somata indicates a modulatory control of these octopaminergic cells by PSK-neurons. To test this we investigated the effect of PSK on endogenous spiking of isolated DUM neurons. The experiments revealed that PSK affected spiking at pico-to nanomolar concentrations in a biphasic manner. The highly variable early response observed up to 2 minutes after PSK application was characterized by an acceleration of spiking whereas the uniform late response was always a reduction in spike frequency ( Figure 6A). The time courses given in Figure 6B for two cells illustrate the variability of the early response. There was an acceleration of spiking with slow relaxation (upper panel) or oscillations of spike frequencies (lower panel). To estimate the concentration dependence of PSK effects on spike frequency we normalized the maxima of early responses and the means of late responses at 5 minutes presence of PSK ( Figure 6C). Both the early and the late response were concentration dependent (EC 50 of 21 pM for the early response and 11 pM for the late response).
PSK had little effect on the shape of action potentials. While in some cases the overshoot was enhanced by up to 2 mV, the hyperpolarization was never changed. The only significant effect (t-test, p ¼ 0.0093) we registered was a shift of action potential threshold from À28.5 AE 2.0 mV (n ¼ 5) under control to À24.5 AE 1.7 mV in the presence of 10 nM PSK for 5 minutes.
The investigation of mechanisms responsible for the different effects of PSK on DUM neuron spiking was initially focused on the late response characterized by reduced pacemaker shift within the interspike interval. We thus tried to identify the pacemaker conductance that was downregulated by PSK.
PSK attenuates the Ca 2þ background current. One of the ion currents contributing to pacemaking in DUM neurons is a Ca 2þ background current (Heine and Wicher, 1998). This voltage-independent current is constitutively active but negatively regulated by cAMP (Wicher et al., 2004). It is blocked by LOE908, an inhibitor of non-capacitative Ca 2þ entry (Wicher et al., 2004), and by the TRP channel blocker 2-APB (Wicher et al., 2006a). The channel conducting the Ca 2þ background current is most probably composed of pTRPg, an ortholog to the Drosophila TRPg channel subunits which are involved in photoreception (Wicher et al., 2006a). In voltage-clamp experiments with Ca 2þ as the only membrane permeating cation, PSK (1 nM) reduced a current that was Satiety signaling via sulfakinin constitutively active in the voltage range between À60 and À120 mV ( Figure 7A). The PSK effect on the membrane current developed quickly after peptide application and saturated after 3 minutes. The recovery after washing off the peptide was $80% within 3 minutes ( Figure 7B).
A linear extrapolation of the PSK-sensitive current yielded a reversal potential around 0 mV yet the data are compatible with a Goldman-Hodgkin-Katz model for a Ca 2þ conductance ( Figure 7A). When Ca 2þ was replaced by Sr 2þ , PSK still reduced a resting current of comparable size (not shown). After preincubation of neurons with the Ca 2þ background current blockers LOE908 or 2-APB, PSK failed to affect the membrane current ( Figure 7C).
PSK attenuates the Ca 2þ background current by increasing the cAMP concentration. Since cAMP is known to exert an inhibitory control on the Ca 2þ background current (Wicher et al., 2004) and on pTRPg channels (Wicher et al., 2006a) we tested whether the reduction of the Ca 2þ background current by PSK might be mediated by an increase in the cAMP concentration.
Using an antibody raised against formaldehyde-fixed cAMP, the effect of PSK on the cAMP level was studied. Compared with control experiments (Figures 8A and 8C), pretreatment with 10 nM PSK increased cAMP-ir in DUM neuron cell bodies (fluorescence intensity: 25.8 AE 1.1, n ¼ 31-35.5 AE 1.4, n ¼ 19; Figures 8A and 8C, p < 0.001). After relaxation, which is complete after about 3 minutes, the cAMP concentration was stably enhanced within the tested time between 10 and 25 minutes of prolonged PSK incubation. By contrast, in the presence of anti-PSKR (1:100) incubation of DUM neurons with PSK (1 hours) did not result in an increased cAMP level (Figures 8B and 8C). These results confirm that PSK attenuates the Ca 2þ background current via an elevation of the intracellular cAMP concentration.
PSK reduces the intracellular Ca 2þ concentration. The Ca 2þ background current not only plays a role in pacemaking but also affects the intracellular free Ca 2þ concentration [Ca 2þ ] i of DUM neurons (Heine and Wicher, 1998;Messutat et al., 2001;Wicher et al., 2004). The reduction of the Ca 2þ background current by PSK is thus expected to attenuate [Ca 2þ ] i . In order to confirm this we performed Ca 2þ imaging experiments in isolated DUM neurons.
The [Ca 2þ ] i distribution under control conditions was not always homogeneous. At the basal pole, i.e., near the primary neurite, [Ca 2þ ] i was below 100 nM ( Figure 9D). This region is equipped with Na þ channels and responsible for the generation of action potentials in the cell body (Grolleau and Lapied, 2000). [Ca 2þ ] i at the apical pole was frequently somewhat enhanced, also hot spots were observed. Application of PSK (10 nM) caused an overall reduction of [Ca 2þ ] i . Within 10 minutes PSK reduced [Ca 2þ ] i from 139 AE 15 to 97 AE 11 nM ( Figure 9C).
The results obtained from the investigations of long-lasting PSK effects on the level of ion currents and [Ca 2þ ] i support the view that the peptide reduces the Ca 2þ background current which was previously shown to be involved in other neuro-modulatory processes . The reduction of a current serving as pacemaker seems to be sufficient to explain the reduction of spike frequency by PSK in the late time domain. However, taken into account that the reduction of Ca 2þ background current starts shortly after peptide application ( Figure 7B), it remains to be shown which mechanism is responsible for the early effect of PSK. The increase in spike frequency obviously requires an over-compensation of the peptide-induced attenuation of Ca 2þ background current by another pacemaker conductance.
PSK neither affects Na þ nor K þ currents. Among the conductances capable of serving as pacemaker in DUM cells there is a sustained component of voltage-gated Na þ current (Grolleau and Lapied, 2000). Thus we asked whether PSK might potentiate this current. However, PSK applied at concentrations of 1 and 10 nM revealed that size and kinetics of voltage-gated Na þ currents were not affected (n ¼ 4 and 6, respectively; not shown).
In DUM neurons, various types of K þ currents occur that not only contribute to action potential shaping but also to the control of pacemaking, as for example, the voltage-gated A-type current (Grolleau and Lapied, 2000). To test whether K þ currents might be affected by PSK, we investigated its effect on total outward currents obtained by depolarizations from À90 mV to potentials ranging from À80 to þ40 mV. PSK, at concentrations of 1 and 10 nM, did not affect these currents within 2 minutes (n ¼ 6, not shown). Similarly, when Na þ channels were blocked by TTX thereby inhibiting Na þ -activated K þ currents, application of 10 nM PSK failed to significantly change size and kinetics of outward currents (n ¼ 6, not shown).
PSK enhances a voltage-gated Ca 2þ current component. The somata of DUM neurons are equipped with voltage-gated Ca 2þ channels activating in the low voltage range  and in the high voltage-range Wicher and Penzlin, 1997). When investigating the effect of PSK on voltage-gated Ca 2þ currents we recognized that PSK (10 nM) enhanced Ca 2þ currents positive to À60 mV ( Figures 10A and 10B) but it did not affect current kinetics ( Figure 10A). The I -V relationships for peak currents taken before and 2 minutes after PSK application ( Figure 10B) showed that the PSK effect is most pronounced in the lower voltage range whereas it is virtually absent positive to 0 mV. PSK shifted the midpoint of the G -V relationship towards more negative potentials and it increased the steepness of its slope ( Figure 10C). Fitting data with the Boltzmann equation yielded V 0:5 ¼ À22.4 mV and S ¼ 3.9 mV (Control) and V 0:5 ¼ À26.8 mV and S ¼ 3.0 mV (PSK). Ca 2þ currents show most pronounced Ca 2þ -dependent inactivation in the voltage range between À30 and 0 mV, i.e., where PSK had the strongest effect on the Ca 2þ conductance ( Figure 10A; cf. Wicher and Penzlin, 1997). Therefore, for a more detailed investigation of the PSK effect Sr 2þ was used as charge carrier to avoid this Ca 2þ -dependent inactivation. However, PSK failed to significantly affect Sr 2þ currents ( Figure 10D). The PSK effect on Ca 2þ channels thus seems to rely on Ca 2þ as charge carrier and is possibly related to Ca 2þ -induced inactivation.
Sr 2þ currents activated positive to À70 mV and reached the maximum at À20 mV. Both potentials were $10 mV more negative than those for Ca 2þ currents which may be accounted for by lower surface charge screening by Sr 2þ (Hille, 1992). The I -V relationship for Sr 2þ peak currents shows a continuous rise from activation threshold to maximum, and it can be well fitted with a GHK equation (Figure 10D, cf. Materials and Methods). By contrast, the I -V relationship for Ca 2þ peak currents increases only moderately on jumps ranging from the activation threshold to À40 mV. Positive to À40 mV there is, however, a steep increase in  Satiety signaling via sulfakinin current amplitude with increasing depolarization. A simple GHK model fits this I -V relationship less well ( Figure 10B).
An explanation for the apparent difference between Ca 2þ and Sr 2þ currents may be that low voltage-activated (LVA) Ca 2þ channels are down-regulated in the presence of extracellular Ca 2þ . The regulation of a LVA Ca 2þ current in DUM neurons by the intracellular Ca 2þ concentration has been reported . Furthermore, the Ca 2þ background current was shown to be involved in controlling [Ca 2þ ] i (Heine and Wicher, 1998). As described above, PSK attenuates the Ca 2þ background current thereby reducing [Ca 2þ ] i (Figure 7).
That PSK affected Ca 2þ currents preferentially in the lower voltage range might indicate a selectivity to distinct Ca 2þ current component(s). PSK has the most pronounced effect on Ca 2þ peak currents at À20 mV ( Figure 10E). At this potential the total current mainly results from a current component sensitive to the P/Q-type Ca 2þ channel blockers vconotoxin MVIIC and v-agatoxin IVA (Wicher and Penzlin, 1997). In order to test whether PSK affects this component it was applied in the presence of v-agatoxin IVA (v-AgaTx). v-AgaTx (50 nM, 2 minutes) blocked a current activating positive to À40 mV reaching a maximum at around À10 mV ( Figure 10E). Figure 10F compares the increase in Ca 2þ current density elicited by 10 nM PSK under control conditions with that in the presence of v-AgaTx at À30 mV where the P/Q-type current is rather low (cf. Figure 10E) and at À20 mV where this current is predominant. At the latter potential, v-AgaTx dramatically reduced the PSK effect (by $80%) whereas it did not significantly change PSK efficacy at À30 mV. The strong reduction of PSK effect by v-AgaTx at À20 mV and more positive In the late response the frequency decreases monotonously until reaching steady state. (C) Concentration dependence of early and late PSK effects on the normalized spike frequency. The early response curve contains frequency maxima recorded up to 3 minutes after PSK application, the late response curve contains frequencies recorded at the end of PSK application where relaxation was complete. Data, means of 6-9 cells; bars, SEM. Doseresponse curves fitted to data are described by EC 50 of 21 and 11 pM as well as Hill coefficient of 0.94 and 0.72 for the early and the late response, respectively. potentials (not shown) shows that PSK indeed affects the P/Q-type current. However, the remaining PSK-induced Ca 2þ current increase in the presence of v-AgaTx indicates that PSK also enhances a LVA current. Thus, PSK does not specifically act on a distinct Ca 2þ current component.
In the continuous presence of PSK (>2 minutes), we found a reduction of Ca 2þ currents activated by voltage commands positive to À40 mV. At the peak of the I -V relationship there was a reduction by 17 AE 12% (n ¼ 6). To exclude that this effect may result from run-down due to Ca 2þ influx during measurements we repeated the experiments using Sr 2þ as charge carrier. Under these conditions there was still a reduction by 15 AE 9% (n ¼ 5). The reduction of voltage-gated Ca 2þ current possibly explains the increase in action potential threshold observed in the presence of PSK for >2 minutes (cf. above) and it may contribute to the termination of spike frequency oscillations.

DISCUSSION
Signaling systems involved in general regulation of physiological processes such as the noradrenergic system in vertebrates or the octopaminergic system in insects (Roeder, 2005) supply their messengers throughout the nervous system. The sulfakinin system in Periplaneta is similarly widespread as the projections of the PDS1-neuron run throughout the ventral nerve cord. The arborizations, as demonstrated in the sixth abdominal ganglion, are sufficient to supply the whole ganglion with PSK. Within ganglia, PSK is released both in synaptic as well as in extrasynaptic areas (Figure 3) indicating an important role of volume transmission (e.g., Vizi et al., 2004;Zoli and Agnati, 1996). In addition, since there are 24 PSK-producing neurons in the pars intercerebralis, which project to the corpora cardiaca, a neurohemal organ, a further role of PSK as a hormone is conceivable.
The expression of PSKR in efferent DUM neurons which are involved in the regulation of the animal's activity level (Roeder, 2005) indicates a general activity-regulating role of PSK. Electrophysiological experiments confirmed that PSK indeed modulates the performance of DUM neurons. We have seen that PSK attenuates the spike activity of DUM neurons, which is expected to reduce the octopamine release by these cells. The satiety messenger PSK, which reduces food intake (Maestro et al., 2001), may thus down-regulate the general level of activity. In the opposite physiological context, i.e., under energy-deficient conditions, AKHs are released to mobilize energy reserves. In parallel they up-regulate the activity level that involves stimulation of octopaminergic DUM neurons (Wicher et al., 2006b;Wicher, 2007).
Since there are no synaptic contacts at the somata of insect neurons, the question arises how do chemical messengers approach their receptors localized in the cell body membrane? As the PSK-induced attenuation of the DUM cell spike frequency was observed in isolated somata, PSK receptors are surely present in the cell body membrane as previously reported for other ionotropic and metabotropic receptors such as cholinergic , glutamatergic (Washio, 2002), or peptidergic receptors (Wicher et al., 2006b). Our electron microscopical immunocytochemistry revealed two axon types with differently intense PSK-ir, similar to the situation found in the corpora cardiaca. In both In DUM neurons preincubated with PSKR antisera cAMP immunostaining was not affected by PSK (right). ***Significantly different, p < 0.001, Student's ttest; ns, not significant. Scale bar, 30 mm. cases, we assume that the intense staining indicates the presence of PSK while the weak staining may result from RF-amide cross-reactivity which is in accordance with the ELISA results (cf. Materials and Methods). Furthermore, we obtained evidence for the localization of PSK-ir nerve endings near the extracellular matrix, and-most intriguingly-the presence of PSK-ir material within the extracellular matrix indicates that one way for PSK to reach the DUM cell bodies is a paracrine mechanism via volume transmission through this channel system which has extensions to the DUM surface ( Figure 3). Alternatively, since PSK is stored in the corpora cardiaca (Figure 1), it is released into the circulation and may penetrate the neural sheath of the CNS to approach the DUM neurons as previously shown for the AKH (Wicher et al., 2006b).
The persisting PSK effect leading to reduced spike activity was accomplished by down-regulation of the Ca 2þ background current sensitive to cAMP, LOE908, and 2-APB. We have recently shown that upregulation of this current by an AKH causes acceleration of spike activity (Wicher et al., 2006a). This was accompanied by an increase in [Ca 2þ ] i at the apical pole of the DUM cell indicating that background channels and the signal transduction machinery are localized in this region (Wicher et al., 2004). Interestingly, PSK produced a reduction in [Ca 2þ ] i within the same region (Figure 9).
The early response to PSK involving transient acceleration of spiking was accompanied by an increase in voltage-gated Ca 2þ currents. At least two types of currents were found to be affected, the P/Q-type current and an LVA current. Since up-regulation of the Ca 2þ background current in DUM neurons was seen to be responsible for a reduction in voltage-gated Ca 2þ currents (Wicher, 2001b), the down-regulation of the background current by PSK may simply remove a depressing effect on voltagegated Ca 2þ channels. That the amplitude of LVA and P/Q-type currents are under control of Ca 2þ has already been demonstrated Penzlin, 1997, respectively). Since LVA channels also contribute to pacemaking in DUM neurons, their upregulation is expected to increase the firing rate of DUM neurons. The observed transient increase in spike frequency or its oscillation may thus result from the superposition of two contrary effects on pacemaking, namely the up-regulation of voltage-gated Ca 2þ currents on the one hand and the down-regulation of the Ca 2þ background current on the other.
An experimental analysis of the signal transduction of PSK signaling was beyond the scope of this study. However, two algorithms aimed at predicting the G-protein specificity of receptor coupling were applied to analyze the sequence of PSKR. G protein and receptor interaction feature finding system (GRIFFIN, Yabuki et al., 2005) predicted a coupling to G s . Figure 10. Effect of PSK on voltage-gated Ca 2þ currents. (A) Families of Ca 2þ currents before and 2 minutes after application of 10 nM PSK. Currents were activated by stepping from a holding potential of À90 mV to voltages ranging from À60 to À20 mV in 10 mV increments. (B,C) I-V relationships (B) and G-V relationships (C) for Ca 2þ peak currents before (Control) and 1 minute after application of PSK. Data, means of 5-8 cells; bars, SEM. Curves were fitted to the data according to the GHK equation. (D) I-V relationships for Sr 2þ peak currents before (Control) and 1 minute after application of PSK. Data, means of five cells; bars, SEM. Curves were fitted to the data according to the GHK equation. (E) I-V relationships for the current produced by 10 nM PSK (1 minute) and the current blocked by 50 nM v-AgaTx. Currents were normalized to the maximum of the I-V curve under control conditions. (F) Ca 2þ current density produced by 10 nM PSK (1 minute) under control conditions and in the presence of 50 nM v-AgaTx.
Since this algorithm stops upon identifying the G s -binding type it does not provide any information on possible other G protein coupling. The algorithm PRED-COUPLE (Sgourakis et al., 2005) which is also able to model promiscuous coupling predicted a further coupling of PSKR to G q,11 and G i,0 . The regulation of the Ca 2þ background current in DUM neurons by cAMP may be accomplished via different pathways including receptor coupling to G s , G i , and G q or even by affecting the NO signaling system (Wicher et al., 2004). Activation of the Drosophila SKR heterologously expressed in mammalian cells most probably leads to G q activation (Kubiak et al., 2002).

CONCLUSION
We have seen that the gut peptide PSK which also acts within the central nervous system thereby inhibiting feeding (Maestro et al., 2001;Wei et al., 2000) may be supplied throughout the nervous system and inhibits the activity of octopaminergic DUM neurons. This reduces the octopamine release and thus the general activity of animals (Roeder, 2005). Mammalian peptides mediating satiety signals such as CCK similarly act within the CNS (Nussdorfer et al., 2005). For example, CCK depolarizes rat vagal neurons by acting on background conductances (Peters et al., 2006). Thus the structurally related insect and mammalian neuropeptides PSK and CCK seem to have a similar function in the nervous system in that they down-regulate general activity. The additional involvement of both peptides in the midgut control (not shown) supports the view that various aspects of feeding regulation are evolutionary conserved between insects and mammals.
Under energy deficiency-the opposite physiological condition as PSK is released-insects mobilize energy reserves due to secretion of AKHs. Most intriguingly, AKHs enhance the activity of DUM neurons (Wicher et al., 2006b) that are depressed by PSK. Moreover, not only the Ca 2þ background channel is the common target of antagonistic modulation by AKH and PSK, also the cAMP level is regulated by both peptides in antagonistic manner.
Thus the peptidergic control of metabolism involves general regulatory principles conserved in evolution as well as special mechanisms transducing both the message of hunger or satiety.