Identification of a Novel Hemolymph Peptide That Modulates Silkworm Feeding Motivation*

Phytophagous insects do not constantly chew their diets; most of their time is spent in a non-feeding quiescent state even though they live on or around their diets. Following starvation, phytophagous insect larvae exhibit enhanced foraging behaviors such as nibbling and walking similar to the sequential behavior that occurs prior to each meal. Although extensive physiological studies have revealed regularly occurring feeding behaviors in phytophagous insects, little has been elucidated regarding the mechanism at the molecular level. Here, we report identification and characterization of a novel 62-amino acid peptide, designated as hemolymph major anionic peptide (HemaP), from the hemolymph of Bombyx mori larvae that induces foraging behaviors. The endogenous HemaP levels are significantly increased by diet deprivation, whereas refeeding after starvation returns them to basal levels. In larvae fed ad libitum, hemolymph HemaP levels fluctuate according to the feeding cycle, indicating that locomotor-associated feeding behaviors of B. mori larvae are initiated when HemaP levels exceed an unidentified threshold. Furthermore, administration of exogenous HemaP mimics the starvation-experienced state by affecting dopamine levels in the suboesophageal ganglion, which coordinates neck and mandible movements. These data strongly suggest that fluctuation of hemolymph HemaP levels modulates the regularly occurring feeding-motivated behavior in B. mori by triggering feeding initiation.

Phytophagous insects do not constantly feed; rather most of their time is spent in a quiescent non-feeding state even though they live on or around their diets. This means that despite continuous stimulation by host plant chemicals, the trigger for phytophagous insects to switch from quiescent state to the initiation of feeding behaviors is host plant independent. It is clear that phytophagous insects alternate between feeding and quiescent non-feeding states based on their dietary intake. These repetitive changes from feeding state to the quiescent state and back generate feeding cycles. Recently, we demonstrated that the silkworm, Bombyx mori, a monophagous, phytophagous lepidopteran, has a regular feeding cycle of about 2 h (1). Similar cyclic feeding rhythms are also observed in other phytophagous insects such as locusts (2,3) and caterpillars (4 -6). This suggests that, cyclic feeding behavior might be a conserved phenomenon among phytophagous insects. It also implies that some endogenous system strictly regulates the initiation and termination of feeding in phytophagous insects (7,8).
In the case of B. mori larvae, the probability of feeding initiation (hereafter referred to as feeding motivation) drastically increases about 1 h post feeding from the previous meal (1). In addition, B. mori larvae exhibit repetitive feeding behavioral cycles independent of circadian rhythms (1,9), unlike other phytophagous lepidopteran species, which have feeding cycles that are influenced by other general factors, such as circadian rhythms and visual light stimuli (10). Because B. mori larvae have regularly occurring feeding cycles that are independent of circadian rhythms, it is unlikely that these factors are involved in regulating B. mori feeding cycles. This suggests that B. mori larvae are a good experimental animal for investigating the endogenous regulatory mechanisms underlying regularly occurring feeding behaviors in phytophagous insects.
From observations of B. mori larvae, we found that feeding motivated B. mori larvae exhibit increased foraging activities, including head-swaying or swinging, nibbling, and walking, whereas B. mori larvae rarely move during the periods between meals. Such foraging behaviors are also generally observed immediately before each meal (Fig. 1A). In particular, small head-swaying behavior always occurs immediately before each meal, which sometimes triggers exaggerated behaviors related to feeding initiation. These excited feeding related behaviors associated with locomotor activation of legs, mouth parts, and mandibles following starvation, or prior to feeding, have also been observed in other phytophagous lepidopteran species such as the tobacco hornworm, Manduca sexta (11,12). It appears that activation of foraging behaviors and the subsequent initiation of consumption are necessary for a feeding motivated condition. It also appears that some endogenous factor(s) drives the initiation of feeding and the accompanying activation of locomotor activities in phytophagous insects.
Several physiological studies have proposed that some unknown endogenous factor(s) regulate these behaviors in phytophagous insects, such as locusts (3,13) and caterpillars (5,8). The factor may detect the imbalance of some nutrients or metabolites in hemolymph. In addition, physiological studies have demonstrated that the hemolymph is the site where nutritional condition of the insect is perceived. However, because of the scarcity of accepted bioassay systems for evaluating feeding motivation in phytophagous insects, these factors have not been characterized.
To address the molecular basis of feeding motivation in B. mori larvae, we established a bioassay to characterize the factor involved in feeding motivation by observing larval foraging behaviors. Using this bioassay, we purified and identified a novel peptide from B. mori larval hemolymph that drove stereotypical foraging behavior following injection into B. mori larvae. We also demonstrated that hemolymph levels of the peptide fluctuated according to feeding cycles.
Chemicals and Reagents-Chemicals and reagents used in the present study were purchased from Nacalai-tesk (Osaka, Japan). Dopamine sodium salt, and serotonin (5hydroxytryptamine), octopamine, and tyramine were purchased from Sigma.
Bioassay for Foraging Behavior-Only populations of larvae growing synchronously were used in assays. Short-term starved larvae were deprived of artificial diet blocks during day 2 of the last instar for 6 -8 h. Artificial diet blocks were then placed in front of the larvae. In most cases, larvae fed for 20 -30 min. After feeding, the larvae were anesthetized by submerging in ice-cold water for 10 min. After drying, lyophilized samples dissolved in distilled water were injected into the dorso-abdominal portion (at the fifth thoracic segment). For assays, individual larvae were injected with 100-l samples. After sample injection, larvae were placed in a plastic container or on a large sheet of wax paper facing an artificial diet block (supplemental Movie S1). Larvae were spaced so as not to interrupt the feeding behavior of other animals. Larvae were observed for 1-2 h. Throughout the assay period benchand wind-vibrations were carefully controlled. The time of initiation and termination of foraging behaviors were individually recorded. The first noticeable head movement, which is observed at the beginning of head-swaying and is usually characterized by a figure eight movement pattern consisting of ϳ1-2-mm movements with mandible vibrations, which are generally observed at the beginning of nibbling behavior, were counted as active behaviors. Because of difficulty in observation, antennae and maxillary movements were not counted as active behaviors. Biological activity was evaluated by calculating the proportion of time spent in foraging behaviors. Observations were carried out using at most 10 larvae in a single assay. Because video monitoring as described below limited observations of small mouthpart movements, the general He-maP assay was carried out by one of us (S. N.) and observa-tions were independently repeated by a second author (N. M.) using a different lot of larvae.
HemaP Purification-The 28-amino acid residue peptide initially isolated was purified as follows. Hemolymph was collected from 2000 day-2 last instar B. mori larvae starved for 2 days (ϳ500 ml) and maintained on ice. The collected hemolymph was stored at Ϫ80°C until purification. The hemolymph was acidified with trifluoroacetic acid (TFA), the final concentration was 0.2%, and then boiled for 15 min. The denatured pellet was removed by centrifugation at 5,000 ϫ g for 15 min. The supernatant was subjected to Sep-Pak Vac C 18 . The column was washed with 0.1% TFA aqueous solution and then eluted with 60% acetonitrile, 0.1% TFA. The eluent was subjected to reversed-phase HPLC (RP-HPLC) on a Hitachi L-6250 (pump with controller; L-6050, pump; L-4000, UV detector; Hitachi, Tokyo, Japan) with a Senshu Pak Pegasil-300 ODS column (10 mm inner diameter ϫ 250 mm, Senshu Kagaku; Tokyo, Japan) using a 25-min linear gradient of 10 -60% acetonitrile containing 0.05% TFA at a flow rate of 2.5 ml/min. The elution was monitored by absorbance at 280 nm. Biologically active fractions were further purified (second to fourth steps) by analytical HPLC (JASCO SC-802, PU-880, UV-875; JASCO; Tokyo, Japan) on a Senshu Pak Pegasil-300 ODS column (4.6 mm inner diameter ϫ 250 mm) with a 25min linear gradient of 10 -60% acetonitrile containing 0.05% TFA at a flow rate of 1.0 ml/min. The elution was monitored by absorbance at 225 nm. In the third RP-HPLC purification step, heptafluorobutyric acid replaced TFA as the ion-pairing reagent. The amount of the finally purified peptide (an initially purified peptide cleaved by acidic boiling) was ϳ1.5 g. The isolated peptide was sequenced using a Procise cLC (Applied Biosystems). HemaP was separately isolated in a similar procedure. The amount of isolated HemaP was ϳ100 g from the hemolymph of 10 larvae.
Monitoring Larval Behavior-Video monitoring was carried out using a digital video camera, F1.2 GZ-MG40-P (Victor, Tokyo, Japan). High speed video recordings were generated with Video Studio version 12 (Corel, Tokyo, Japan). Larval movement traces were performed manually by stopping the video record every 30 s.
Analytical Indices of HemaP Activity-To evaluate the biological effects of HemaP, we used several biological indices related to feeding behavior. The amount of ingested diet was measured by the difference in weights of individual artificial diet blocks before and after sample injection. The approximate digestibility (AD) 2 (14) is a conventional index representing the uptake efficiency of ingested diet and is calculated as follows: AD ϭ {(weight of ingested diet) Ϫ (weight of fecal pellets)}/(weight of ingested diet).
MALDI-TOF MS Analysis-Hemolymph and other samples were desalted with Sep-Pak Vac C 18 (100 mg) used as described above (eluted with 60% acetonitrile, 0.1% TFA). Mass spectra were measured on a matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometer (Voyager-DE TM STR, Applied Biosystems) with ␣-cyano-4-hydroxycinnamic acid as matrix in the positive ion mode. Matrix solution was prepared by saturating ␣-cyano-4-hydroxycinnamic acid in 60% acetonitrile containing 0.1% TFA. The samples were applied after mixing with matrix solution in a 1:1 ratio.
Preparation of Recombinant HemaP (rHemaP)-To construct GST-fused HemaP, RT-PCR was carried out to produce two cDNA fragments, one encoding GST with a C-terminal factor Xa recognition site and the other encoding HemaP with a factor Xa recognition site (primer sets used are described below). The full-length GST-factor Xa-HemaP (GST-X-HmP) was generated via overlap extension PCR. We used the NdeI and XhoI sites of GST-X-HmP to subclone into the NdeI and XhoI sites of pET28a. The resulting plasmid (pGST-X-HmP) was transformed into BL21(DE3) bacterial cells to produce recombinant GST-X-HmP. Production of GST-X-HmP was initiated following addition of isopropyl 1-thio-␤-Dgalactopyranoside (0.5 mM final concentration) at a bacterial density of 0.5-0.7 at 600 nm. After a 5-h incubation, cells were collected by centrifugation and sonicated in ice-cold PBS containing 5 mM EDTA and 0.1 mM PMSF. Sonicated cells were re-centrifuged. GST-X-HmP was predominantly located in the insoluble/inclusion body fraction. The precipitate was solubilized by addition of a small volume of PBS, 1% SDS. The solubilized protein was passed through Sephadex G-15 using 50 mM Tris-Cl (pH 8.0) as a mobile phase. Fractions containing proteinaceous substances were collected and digested with Factor Xa (Merck, Novagen, Darmstadt, Germany) at room temperature for 16 h after addition of 100 mM NaCl, 5 mM CaCl 2 . The reaction mixture was passed over a glutathione column (GE Healthcare) to remove the digested GST portion. The flow through was then subjected to RP-HPLC using the same linear gradient program as described for He-maP isolation. The resulting recombinant HemaP (rHemaP) was confirmed by MALDI-TOF MS analyses and N-terminal sequencing. The yield of rHemaP was ϳ150 g from a 1-liter culture. First RT-PCR was performed by following primers using cDNA derived from B. mori larval fat body as a template DNA: Fw-1 primer: 5Ј-TCTGATCGAAGGTCGTGATGCT-CCTAAAGAAG-3Ј, Rv-1 primer with a XhoI restriction site: 5Ј-CCGCTCGAGTTATTCAACTTTTTCCCT-3Ј. Secondary PCR for GST amplification was performed by the following primers using pGEX-5X3 as a template DNA: Fw-2 primer with a NdeI restriction site: 5Ј-CAGCCATATGTCCCCTAT-ACTAGGTTATTGG-3Ј, Rv-2 primer: 5Ј-CTTCTTTAGGA-GCATCACGACCTTCGATCAGT-3Ј. A third PCR was performed with primers Fw-2 and Rv-1 using a mixture of the first and second PCR products as DNA template.
Antiserum against HemaP-Rabbit anti-HemaP antiserum was generated using a synthetic peptide corresponding to the N-terminal portion of HemaP from Asp 1 to Phe 40 .
HemaP Quantification by ELISA-Hemolymph was collected after centrifugation to remove hemocytes at 4°C. The supernatant was partially purified with Sep-Pak Vac C 18 as described above. HemaP-containing fractions were lyophilized prior to quantification. Lyophilized samples were dissolved in 10 mM PBS (pH 7.4). Samples (100 l) were adsorbed to wells of a 96F MAXISORP BLACK MICROWELL plate (Nunc, Thermo Scientific, IL) for 1 h at room temperature. Wells were then washed with PBS-T (0.05% Tween 20, 10 mM PBS, pH 7.4) three times and then blocked with PBS-T containing 1% BSA for 20 min at room temperature. After washing with PBS-T three times, the wells were incubated with an anti-HemaP antiserum (100 l; 1:10000 dilution in PBS-T) for 1 h at room temperature. After washing as before, wells were incubated with goat anti-rabbit IgG conjugated to HRP (Thermo Scientific, IL) (100 l; 1:3000 dilution in PBS-T) for 45 min at room temperature. Wells were washed as before and 75 l of Super Signal West Pico Chemiluminescent substrate (Pierce, Thermo Scientific, IL) was added to each well. Chemical luminescent signals were measured using a Luminescencer-JNR II (ATTO, Tokyo, Japan). For rHemaP standards, varying concentrations were prepared by diluting with PBS (0.01-10 ng). Quantification was calculated using standard curves generated from at least three different concentrations of rHemaP.
Quantification of HemaP Levels from Single Larvae-Day 2 last instar larvae were used for quantification. Hemolymph was collected by piercing the ventral thoracic portion of the fifth segment with a 31-gauge needle every 20 min for 3 h. Hemolymph (1 l) was collected with a micropipette and diluted immediately with 50 l of ice-cold distilled water. The diluted hemolymph was centrifuged at 12,000 ϫ g at 4°C to remove hemocytes, and then the supernatant was boiled for 5 min to inactivate contaminating proteases, and then centrifuged at 12,000 ϫ g again to remove denatured proteins. The supernatant using 0.05 l of hemolymph equivalents was quantified by ELISA in triplicate.
Iodination of HemaP-rHemaP was iodinated using radioactive iodine and the chloramine T method described previously (15). rHemaP (10 g) in PBS was mixed with 2 l of NaI (1 mg/ml), 2 l of Na 125 I (GE Healthcare), and 15 l of chloramine T (2 mg/ml). The mixture was incubated on ice for 90 s. The reaction was stopped by addition of 50 l of sodium metabisulfite (5 mg/ml). The resulting iodinated rHemaP was purified using reversed-phase HPLC (Jasco SC-802, PU-980, UV-970) and eluted using the same linear gradient program as used in the final HemaP purification step. Iodinated rHe-maP ( 125 I-rHemaP) was isolated based on retention time and UV absorbance at 225 nm. The retention time and iodination of 125 I-rHemaP were confirmed by comparing with MALDI-TOF MS analyses of unlabeled rHemaP subjected to the chloramine T method in parallel. Based on data from cold iodinated HemaP, the major chloramine T reaction product was confirmed to be di-iodinated HemaP by MALDI-TOF MS analysis. The biological activity was confirmed using cold diiodinated rHemaP and the radioactivity of 125 I-rHemaP was approximated at 3-15 ϫ 10 6 cpm/g. 125 I-rHemaP was used within 1 week of radiolabeled iodination. The stability of 125 I-rHemaP stock maintained at Ϫ80°C was confirmed with no substantial degradation by RP-HPLC.

125
I-rHemaP Tracing-125 I-rHemaP was injected into 12-h starved larvae following a 10-min ice-cold water anesthetization. Tissues from 125 I-rHemaP-injected larvae were dissected 30 min after injection when larvae had terminated feeding, and 125 I-rHemaP distribution in refed larvae was analyzed by comparing with that of continuously starved larvae. The tissue distribution was evaluated by dividing the weight of measured tissues. Radioactivity was measured using a ␥-counter (ALOKA, Tokyo, Japan).
Immunohistochemistry of HemaP-Fat body of B. mori larvae on day 2 of last instar was dissected and fixed in 4% paraformaldehyde for 16 h at 4°C, and then washed with PBS three times for 10 min each. The cells were permeabilized with PBS containing 0.1% collagenase type I (Sigma) and 0.2% Triton X-100 for 1 h at room temperature. After blocking with 2% BSA for 30 min at room temperature, tissues were incubated with an antiserum against HemaP (1:100 dilution) for 16 h at 4°C. Goat anti-rabbit HRP (diluted to 1:500) was used as a second antibody and incubated for 16 h at 4°C. The positive signal was detected by staining with 0.1 mg/ml of 3,3Ј-diaminobenzidine (DAB) for a 15-min incubation at room temperature. Negative control was incubated with preimmune serum instead of the antiserum against HemaP.
Quantitative PCR (qPCR)-Extraction of total RNA and RT reaction were performed as described above. The qPCR primers for HemaP were: forward 5Ј-CTTGGCCCCGGAGA-CTGT-3Ј; reverse 5Ј-CCGCGGGCTTCAGATTTT-3Ј. As a control, the following primers for ribosomal protein L3 (RpL3) were used: forward 5Ј-TGGGAGGTTTCCCCCATT-3Ј; reverse 5Ј-CCATGCAGCAACCCTTGAT-3Ј. PCR was performed using a 7300 Real Time PCR system (Applied Biosystems). Thermal cycling conditions were as follows: initial denaturation at 95°C for 10 min and then 40 cycles of 95°C, 15 s; 60°C, 1 min. The single targeted qPCR product was confirmed by constructing a melt curve to all reactions. The reaction was performed using Power SYBR Green PCR Master Mix (Applied Biosystems) and 50 nM of each primer. Data were analyzed using the 7300 Real Time PCR system software version 1.3 (Applied Biosystems). Results are represented as a ratio of HemaP and RpL3 by C t .
Circular Dichroism Analyses-HemaP and rHemaP were lyophilized and dissolved in PBS at a final concentration of 0.4 mg/ml. Circular dichroism (CD) spectra of HemaP were recorded from 200 to 260 nm using a spectropolarimeter (J-720, Jasco) at room temperature with a 1-mm path length cell.
Quantification of Biogenic Amines (BAs)-The conditions for HPLC and pulse amperometric detection were modified from previous reports (16,17). HPLC was carried out using a Shimadzu HPLC (Class VP) system (controller, SIL10AD; Pump, LC-10AT; Degasser, DGU-20A 3 ; Shimadzu, Kyoto, Japan). Samples were injected using an autoinjector (SIL-10AD; Shimadzu) through an injection valve with a 100-l sample loop. Samples were extracted from lyophilized brain or suboesophageal ganglion (SOG) using an ultrasonicator (Ultrasonic processor UP200S, Ultrasound Technology, Heilsher, Germany) in 50 l of 0.1 M perchloric acid containing 0.15 mM EDTA. Samples were injected onto an HPLC system using a column oven (CTO-6A, Shimadzu) at 40°C and a Capcell-pak C 18 (250 ϫ 4.6 mm inner diameter, 3-m average particle size, Shisei-do, Tokyo, Japan) fitted with a pre-column (35 ϫ 4.6 mm inner diameter, packed with the same resin of Capcell-Pak C 18 ). BAs were detected with a Coulochem II amperometric detector (ESA, Inc., MA) using a high sensitivity analytical cell containing an enhanced response amperometric electrode coupled to a graphite coulometric electrode as a reference electrode. The detector potential was maintained at 900 mV for octopamine and tyramine, and 400 mV for dopamine (DA). The mobile phase contained 0.1 M citric acid, 0.1 M sodium acetate (pH 4.5), EDTA-2Na (6 mg/liter), sodium 1-octanesulfonate (250 mg/liter), and 7.5% acetonitrile. The flow rate was 1.0 ml/min. Standards were prepared using commercial BAs diluted into extraction buffer and ranged from 2 to 100 pg/30 l. Amounts of BAs were calculated by referencing peak areas from the standard curve. Under these conditions, we confirmed that there were no other contaminating peaks derived from brain and SOG extracts, and all detected BA peaks were resolved as single peaks with no overlap on the chromatogram.

RESULTS
Isolation and Identification of a Novel Peptide Driving Foraging Behaviors from B. mori Larvae-To address the molecular basis of feeding motivation in B. mori larvae, we first established a bioassay to characterize the factor involved in feeding motivation. Foraging behaviors including head-swaying, walking, nibbling, and ingesting were observed in feeding-motivated larvae, such as starved larvae and in the larvae prior to each meal (Fig. 1A). Based on these observations, we then evaluated feeding motivation by measuring the amount of time spent in foraging behaviors (Fig. 1B). By screening extracts of several body parts from B. mori larvae, we found that a fraction of hemolymph from starved larvae stimulated foraging behavior upon injection into satiated B. mori larvae (supplemental Movie S1). This result indicated the presence of a factor that drives foraging behavior. A single homogenous fraction corresponding to a peptide composed of 28 amino acid residues (Fig. 1, C and D) was obtained following four reversed-phase high performance liquid chromatography (RP-HPLC) purification steps. A search of the B. mori EST data base, Silkbase (morus.ab.a.u-tokyo.ac.jp/), revealed that the foraging behavior-driving factor was the N-terminal portion of an anionic peptide composed of 62 amino acid residues (calculated pI 4.7) (Fig. 1D, a cDNA sequence, and supplemental Fig. S1). The truncated biologically active peptide was likely cleaved during the acidic boiling purification process from the native peptide by acid-catalyzed hydrolysis of the peptide bond between Glu 28 and Ser 29 .
Identification of the Undamaged Peptide from Hemolymph-To confirm the presence of the undamaged peptide deduced from the cDNA sequence in the hemolymph of B. mori larvae, we analyzed peptide fractions of larval hemolymph. RP-HPLC and MALDI-TOF MS analyses of B. mori larval hemolymph demonstrated that the full-length native peptide precursor of the biologically active peptide described above was predominantly present in the hemolymph (Fig. 2, A and B). The N-terminal amino acid sequence of the isolated native peptide was DAPKEDNSINTLAESK, which was consistent with that of the initially isolated 28-amino acid peptide fragment. MALDI-TOF MS analysis also demonstrated that the observed molecular weight coincided with the calculated average mass of the peptide. We consequently designated the un-damaged peptide as hemolymph major anionic peptide, HemaP. Consistent with the transcript abundance profile in the EST data base, which contained multiple fat body-derived HemaP encoding cDNAs, RT-PCR analysis of hemap showed that the fat body was the main HemaP producing tissue in the larval body (Fig. 2C). The hemap expression in the fat body was also confirmed by immunohistochemistry using antiserum against HemaP (supplemental Fig. S2).
Preparation of Recombinant HemaP-To investigate whether HemaP influences foraging behaviors, we prepared recombinant HemaP (rHemaP) using an Escherichia coli expression system. Analysis by SDS-PAGE and MALDI-TOF MS revealed that the prepared peptide was consistent with those of native HemaP (Fig. 3, A and B) and was thus suitable for further experiments. In addition, we confirmed that rHe-maP and the HemaP-containing hemolymph fraction had identical retention times by RP-HPLC (Fig. 3C), and that the native HemaP-including fraction induced foraging behavior at the 10-l hemolymph equivalent (Fig. 3D).  Biological Activity of HemaP-To test whether HemaP can induce feeding-related activities in B. mori larvae, we observed HemaP-injected B. mori larvae. Injection of both native HemaP and rHemaP into satiated B. mori larvae induced foraging behavior (Fig. 4A), whereas vehicle-and BSA (as a control nonspecific protein)-injected larvae showed little mobility despite the presence of artificial diet blocks. rHemaP injection influenced foraging behavior in a dose-dependent manner; a higher dose of rHemaP had no effect on foraging behavior (Fig. 4A).
We next analyzed several indices related to feeding behavior by rHemaP injection (Fig. 4, B-D). Foraging behaviors in rHemaP-injected larvae were increased (Fig. 4B and supplemental Movie S2) and they ate a larger amount of diet compared with vehicle-injected larvae (Fig. 4C). In addition, rHe-maP-injected larvae had a higher efficiency of ingestion (14) (AD) after 12 h (Fig. 4D). Because the increased AD was comparable with that of the starvation-experienced larvae, we compared indices with starvation-experienced larvae. As shown in Fig. 4, C and D, the effects of rHemaP injection are similar to those of starved larvae, indicating that exogenous rHemaP administration can mimic the state of the starvationexperienced larvae.
Quantification Analysis of Hemolymph HemaP Level-Next, we sought to determine whether HemaP levels fluctuate in relation to the feeding state by measuring endogenous He-maP in the hemolymph by ELISA using an anti-HemaP antiserum (supplemental Fig. S3, A and B). HemaP levels in starved larvae were significantly higher than those in larvae fed ad libitum (Fig. 5A). In periods of prolonged starvation, increased HemaP levels were maintained, whereas no significant change in HemaP levels was observed in larvae fed ad libitum at any developmental time (Fig. 5B). These data also indicate that starvation-induced HemaP levels plateaued. Furthermore, HemaP levels in starved larvae returned to basal levels following resumption of feeding (Fig.  5C), indicating that increased HemaP levels are re-established in response to dietary cues. These changes of HemaP  Fig. S3). D, biological activity of the fraction containing HemaP (fraction 22 as indicated in C) separated from hemolymph with RP-HPLC. Vehicle and rHemaP (5 g) were used as negative and positive controls, respectively. Biological activity was confirmed using a 10-l hemolymph equivalent. The sample from fraction 22 was lyophilized to dry the mobile phase of RP-HPLC. No influence of the trace amount of TFA on the biological activity was confirmed. **, p Ͻ 0.01 (compared with vehicle-injected larvae); one-way analysis of variance, then a post hoc Tukey HSD test (mean Ϯ S.D., n ϭ 5).
according to feeding states were also observed in larvae fed on the mulberry leaves, an original host plant of B. mori (supplemental Fig. S4).
In contrast, we measured HemaP levels in a long developmental period from the last instar larvae to adults. After the feeding period, the HemaP level was significantly decreased, whereas HemaP levels did not change during the feeding period (Fig. 5D). After pupation, HemaP levels increased again, indicating some function of HemaP different from the feeding regulation. In addition, a similar tendency of temporal change was observed in the transcriptional level of hemap in the fat body (supplemental Fig. S5).

Individual Fluctuation of the HemaP Level by Feeding
Cycles-These findings also suggest that HemaP levels change according to the feeding cycle. To test this, we measured He-maP levels from a single larvae fed ad libitum. All larvae investigated showed fluctuations of HemaP in the hemolymph in accordance with their feeding cycles (Fig. 5E). As observed in Fig. 5C, hemolymph HemaP levels returned to the basal levels after each meal.
Fate of HemaP after Feeding Diet-Because quantitative PCR analyses showed that HemaP transcriptional levels did not change in the HemaP-producing tissue, the fat body during the feeding period (supplemental Fig. S5), we traced the level of HemaP after feeding using radioiodinated rHemaP ( 125 I-rHemaP) (Fig. 6A), which had biological activity comparable with rHemaP (Fig. 6B). Injection of 125 I-rHemaP into B. mori larvae confirmed that HemaP levels after feeding were reduced (p ϭ 0.023; Fig. 6C). To determine whether feeding stimulates translocation or degradation of HemaP, we next traced the location of radioactivity in 125 I-rHemaP-injected larvae after feeding. Although high levels of radioactivity were detected in the fat body and gut, a significant increase was detected in the fat body after feeding (Fig. 6D), implying that the return to and maintenance of basal HemaP levels following feeding results from association of HemaP with the fat body. In contrast, a high level of radioactivity in the gut might be derived from degraded peptides of 125 I-rHemaP, because little radioactivity was detected in the fraction corresponding to 125 I-HeamaP in RP-HPLC (data not shown).
Effects of HemaP on Biogenic Amines-Because exogenous rHemaP administration stimulated locomotor activity, we next examined the effect of HemaP on feeding-related locomotor neurons. The suboesophageal ganglia (SOG) has been identified as a regulator of neck and mandible movements (18 -20), and as a sensory regulator in feeding behavior probably via BAs (21)(22)(23). Also, the frontal ganglia projecting from the brain are also thought to regulate swallowing movement as part of a central pattern generator (24,25). Because swaying and nibbling behaviors were enhanced following exogenous rHemaP injection (Fig. 4, A and B), we measured the levels of BAs in the brain and SOG in response to rHemaP injection. We found that rHemaP had no influence on octopamine or serotonin levels in either the brain or SOG (data not shown), but did specifically reduce dopamine (DA) levels in the SOG but not in the brain (Fig. 7, A and B). A similar reduction of BA levels was likewise seen in starved larvae. Taken together, these data indicate that the HemaP level can influence DAergic neurons in the SOG and that the consequent motivation is likely conditioned via DA in SOG. The recovery of DA levels in the brain and SOG upon resumption of feeding suggests that the neuronal locomotor activity related to foraging behavior is triggered by DA and thus accounts for the depleted levels.

DISCUSSION
In this study, we identified a novel peptide that modulates B. mori feeding behavior. Data base searches showed that He-maP-like peptides are conserved across lepidopteran species (supplemental Fig. S6A), whereas HemaP-like peptides were not observed in other insect orders. Among the HemaP-like peptides, a peptide from the honey moth, Galleria mellonella, has been identified as anionic peptide 2 by peptidomics searching for antimicrobial activity (26). They demonstrated that antimicrobial activity of anionic peptide 2 was very weak, consistent with our preliminary data that HemaP has little antimicrobial activities (data not shown).
Although identities of HemaP-like peptides are less than 30%, their structures appear to have characteristics of ␣-helixrich proteins (supported by CD spectral analysis, supplemental Fig. S6B). In addition, HemaP is abundant in the hemolymph (ϳ20 M). Although the complete amino acid sequence of HemaP shows no significant similarity with other proteins, similar ␣-helical characteristics and partial sequence similarities have been observed in other hemolymph proteins, such as lipophorins (insect lipoproteins) (27) and odorantbinding proteins (28). Therefore, HemaP might associate with some compounds possibly related to nutrients or metabolites. As demonstrated in the present study, in which HemaP hemolymph levels oscillate in response to feeding with HemaP transfer from the hemolymph to the fat body (Fig. 6, C and  D), HemaP might shuttle between the hemolymph and fat body similar to that observed for lipophorins. We, however, have not identified those HemaP-associated components in  the hemolymph. It is intriguing that similar proteins (i.e. lipophorin, odorant-binding protein, and HemaP) are produced and secreted by the fat body, indicating that the fat body might contribute to the responses of some hemolymph nutritional changes such as metabolites and nutrients.
In developmental analyses of HemaP, HemaP levels were decreased at wandering and spinning periods, and increased after pupation (Fig. 5D). In addition, the changed HemaP levels were similar to those of hemap transcriptional levels (supplemental Fig. S5). These data indicate that hemolymph HemaP levels might be regulated by secretion from the fat body and by association again with this organ in short feeding cyclic periods. In contrast, HemaP levels might be transcriptionally regulated in long developmental periods. Interestingly, HemaP levels were increased after pupation, indicating the different function of He-maP from feeding regulation in vivo.
As observed in Fig. 5A, HemaP levels increased in starved larvae. In contrast, 10 g of rHemaP injection did not result in any foraging behavior (Fig. 4A). These results indicate that different responsiveness to the increased HemaP level might be affected by different physiological feeding states between satiated and starved larvae.
Our results clearly demonstrate that when HemaP levels exceed a yet undefined threshold level, B. mori larvae initiate locomotor activation and exhibit the characteristic behaviors of motivated feeding. The idea of a threshold HemaP level for triggering locomotor activation corresponds well with the simulation model in locust (3,13). These models illustrate a threshold level that separates the initiation level of feeding or locomotor activation. As demonstrated in the present study (Fig. 5E), the hemolymph HemaP level fluctuated as depicted in the locust feeding threshold model. Therefore, HemaP is the first identified molecule to date that has been shown at the molecular level to regulate feeding motivation cycles.
To date, it has been reported that a number of physiological events, including conditioning for pheromone sensing, are involved in the regulation of BAs (29 -32). More specifically, in Drosophila melanogaster, feeding motivation is conditioned by dopaminergic neurons of the mushroom body in the brain (31), and pheromone sensing is conditioned by SOG DA levels (32). Similarly, HemaP levels contribute to the motivated foraging condition via SOG DA as a model of HemaP turnover depicted   The main product was confirmed to be di-iodinated rHemaP (I-rHemaP). *, a peak corresponding to unreacted rHemaP. X, a peak derived from excess reagents. B, bioassay of I-rHemaP. I-rHemaP induced foraging activity at levels similar to rHemaP (5 g) compared with vehicle injection. **, p Ͻ 0.01; one-way analysis of variance, then a post hoc Dunnett test (compared with vehicle-injected larvae) (mean Ϯ S.D. n ϭ 10). C, decrease of 125 I-rHemaP in the hemolymph following a single meal. 125 I-rHemaP was injected into short-term starved B. mori larvae. Data were compared with those of larvae that were not fed after 125 I-rHemaP injection. p ϭ 0.023 (unpaired t test) (mean Ϯ S.D. n ϭ 10). D, tissue distribution of 125 I-rHemaP after feeding. Gray and white bars represent radioactivity in tissues from refed and continuously starved larvae, respectively. **, p Ͻ 0.01; unpaired t test (refed versus continuously starved larvae) (mean Ϯ S.D. n ϭ 10).
in our present study (Fig. 8). HemaP is secreted from the fat body and accumulates in the hemolymph. Once hemolymph HemaP levels reach a certain threshold, DA in the SOG is consumed for conditioning of foraging behavior. After feeding, the dietary cue introduces excess hemolymph HemaP to the fat body with hemolymph HemaP returning to basal levels. In fact, previous reports demonstrated that DAergic neurons in the SOG of B. mori larvae are specifically located on the anterior portion of SOG (23), which are similar to neuronal sites implicated in nerve cord innervations of the mandible locomotor (25). Although it remains to be elucidated whether HemaP can directly affect SOG DA levels, DAergic neurons appear to impact mandible locomotor neurons. More detailed investigations of the regulatory mechanisms in insect feeding behavior, such as DA tracing following feeding or starvation, and analyses of hormonal influences on He-maP levels, as well as identification of the unknown factors that drive HemaP secretion from the fat body are issues awaiting elucidation. However, our discovery of a specific, abundant hemolymph peptide, HemaP, provides a new framework for understanding the endogenous regulation of feeding motivation. Also, the present work confirms earlier reports at the molecular level that the hemolymph plays a crucial role in insect perception of nutritional status (33)(34)(35).