Development 108, 59-71 (1990)Printed in Great Britain © T h e Company of Biologists Limited 1990

59

Immunological, biochemical and physiological analyses of

cardioacceleratory peptide 2 (CAP2) activity in the embryo of the tobacco

hawkmoth Manduca sexta

KENDAL S. BROADIE1*, ANDREW W. SYLWESTER11, MICHAEL BATE2 and

NATHAN J. TUBLITZ1

1 Institute of Neuroscience, University of Oregon, Eugene, OR 97403, USA2 Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, UK

•Present address: Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, UKt Present address: Department of Bioengineering, University of Iowa, Iowa City, IA 52246, USA

Summary

The cells in the embryonic CNS of the tobacco hawk-moth, Manduca sexta, that synthesize a cardioaccelera-tory peptide 2 (CAP2)-llke antigen were identified usingimmunohistochemical techniques. Two distinct neuro-secretory cell types were present in the abdominalventral nerve cord (VNC) that contain CAP2-like immu-noreactivity during late embryogenesis: a pair of large(diameter range 15-20 fan) cells lying along the pos-terior, dorsal midline of abdominal ganglia A4—A8, anda bilateral set of four smaller (diameter range 6-11 /on)neurons which lie at the base of each ventral root inabdominal ganglia A2-A8. CAP2-like accumulation ap-peared to follow independent patterns in the two celltypes. CAP2-like immunoreactivity began at 60% ofembryo development (DT) in the medial cells, accumu-lated steadily throughout embryogenesis, and droppedmarkedly during hatching. Lateral cells synthesized theCAP2-like antigen later in development (70 % DT) andshowed a sharp drop in antigen levels between 75 % and80 % of embryonic development.

Extracts from developing M. sexta embryos werefound to contain a cardioactive factor capable of acceler-ating the contraction frequency of the pharate adultmoth heart in a fashion similar to CAP2. Immuno-precipitation with a monoclonal antibody that specifi-cally recognizes the two endogenous Manduca cardioac-celeratory peptides and purification using high pressureliquid chromatography identified this factor as cardioac-

celeratory peptide 2 (CAP2). Using an in vitro heartbioassay, the levels of this cardioactive neuropeptidewere traced during the development of the M. sextaembryo. As with the immunohistochemical results, twoperiods during embryogenesis were identified in whichthe level of CAP2 dropped markedly: between 75 % and80 % development, and at hatching. Embryo bioassaysof CAP2 activity were used to identify possible targettissues for physiological activity during these two puta-tive release times. CAP2 was found to accelerate contrac-tion frequency in the embryonic heart and hindgut ofManduca in a dose-dependent fashion. Of these twopossible targets, the hindgut proved to be more sensitiveto CAP2, having a lower response threshold and a longerduration of response to a given concentration of theexogenously applied peptide.

Based on these immunocytochemical, pharmacologi-cal and biochemical results, and on a previously pub-lished detailed analysis of Manduca embryogenesis, weconclude that CAP2 is probably released from a specificset of identified neurosecretory cells in the abdominalVNC to modulate embryonic gut activity at 75-80 % ofembryo development during ingestion of the extra-embryonic yolk.

Key words: neuropeptides, neurohormones, insectneurobiology, developmental endocrinology, invertebrateneurodevelopment, insect gut, invertebrate neuropeptides.

Introduction

Neuropeptides are universally recognized as an import-ant class of intercellular messengers throughout theanimal kingdom. In mature systems, neuropeptideshave been demonstrated to act as neurotransmitters,

neurohormones, and paracrine factors to regulate thefunction of most tissues including modulating CNSactivity (Pickering et al. 1987; O'Shea and Schaffer,1986; Mayeri et al. 1985; Jan and Jan, 1982; Guillemin,1978). Despite this wealth of information in maturesystems, the functional significance of neuropeptides

60 K. S. Broadie and others

during development has been largely ignored. In par-ticular, little attention has been given to participation ofpeptides during embryogenesis. To date, efforts in thisdirection have been largely limited to tracing peptidelevels and mapping peptide distribution using a varietyof immunological and hydridization assay techniques(Pickering et al. 1987). In most cases, the functional roleof these embryonic neuronal factors has proven verydifficult to establish.

Of the few investigations concerning neuropeptidefunction in developing systems, most have utilizedinvertebrate preparations. In simple freshwater coelen-terates, for example, Head/Foot Activating Peptidehas been demonstrated to possess neurotransmitterand/or neuromodulatory properties that interact toestablish determinate fate of regenerating and/or de-veloping tissues (Schaller, 1979). In insects, the shed-ding of the embryonic cuticle is correlated to a drop inthe storage levels of a neurosecretory peptide, eclosionhormone (Truman et al. 1981). Based on this evidence,eclosion hormone has been postulated to play a role inembryonic moults similar to its role during larval andadult development.

Given the diversity of neuropeptide roles post-embryonically, coupled with the few reports in theliterature on their putative embryonic function, it seemsplausible that peptides play a greater variety of roles inthe developing embryo than previously thought. Oneapproach to the study of peptides during developmentis to investigate the embryonic role of peptides that arepresent and functionally important in the mature ner-vous system. Such studies are best accomplished in adeveloping system that is relatively well characterizedand experimentally tractable. One preparation thatmeets these criteria is the tobacco hawkmoth, Manducasexta, whose CNS has been the subject of intensivestudy (Weeks, 1987; Levine, 1986; Tublitz et al. 1986;Truman, 1985).

Our focus in the present study is two cardioregulatoryneuropeptides, cardioacceleratory peptide 1 and 2(CAP1 and CAP2), that were originally isolated in thepharate adult stage of M. sexta (Tublitz and Truman,1985a). Using in vitro and in vivo heart bioassays, thesepeptides were shown to be released into the haemo-lymph from individually-identified neurosecretory cellsin the CNS, accelerating heart contraction frequency ina dose-dependent manner (Tublitz and Truman,1985a, b). Biochemical and physiological studiesdemonstrated that both peptides play important physio-logical roles in the adult moth, acting as cardioexcit-atory hormones during wing-spreading behavior andflight (Tublitz and Truman, 1985b; Tublitz and Evans,1986; Tublitz, 1989). Further work has recently shownthat CAP2 strongly excites the hindgut in fifth instarlarvae, apparently to aid gut emptying during thisdevelopmental stage (N.J. Tublitz, unpublished obser-vations). The CAPs thus have several different stage-specific functions.

Since the CAPs perform several different rolesthroughout post-embryonic life, we were interested inidentifying other CAP roles even earlier in develop-

ment, during embryogenesis. Results from preliminaryimmunocytochemical studies using an anti-CAP anti-body identified several sets of CAP-immunopositiveneurons in the latter stages of embryogenesis (Broadieet al. 19896). In the present study, we use pharmaco-logical, biochemical, and immunocytochemical tech-niques to trace the acquisition and distribution of theCAP peptides during embryo morphogenesis. Ourresults provide several lines of evidence strongly sup-porting a functional role for one of the CAPs duringembryo formation.

Materials and methods

AnimalsTobacco hawkmoths, M. sexta, were reared on an artificialdiet (Bell and Joachim, 1978) in a controlled temperature(27°C [light], 25°C [dark]) room with a 17h photoperiod.Humidity was kept above 50% at all times. Under thesestandard conditions, the time duration from oviposition tohatching was 95±3h (mean±s.E.M.; ^=100). For this study,embryonic development is expressed as a percentage of thetotal development time (DT) using morphological develop-ment characteristics as observed in our laboratory (Broadie etal. 1989a) and elsewhere (Dorn et al. 1987).

Antibody production and specificityA mouse monoclonal antibody, 6C5, shown to be highlyspecific for both CAP, and CAP2 (Tublitz and Evans, 1986),was used in all trials. 6C5 isolation and preparation has beendescribed previously (Taghert et al. 1983, 1984). In short, thecrude extract of 3000 pharate adult Manduca perivisceralorgans, the neurohaemal release site in the VNC for CAPs,was used as the inoculum. Antibody specificity was estab-lished by several independent criteria including morphologi-cal staining of known CAP-containing cells in pharate adultabdominal ganglia, in vivo immunoneutralization, in vitroELISA assays and immunoprecipitation of both CAPs (Tub-litz and Evans, 1986: Taghert et al. 1983, 1984). 6C5 was alsoshown to bind specifically both CAP! and CAP2 that had beenpurified to homogeneity using high pressure liquid chroma-tography (HPLC; Tublitz and Evans, 1986).

To reconfirm the specificity of our anti-CAP antibody 6C5,a series of incubations was performed to establish the natureof the antigen(s) capable of immunoprecipitating the anti-body. Both the suspect neuropeptide antigen, CAP2, and arange of other known invertebrate cardioexcitatory neuro-peptides were incubated with 6C5 (diluted 1:1000 in 0.4%saponin-PBS+1.0% BSA) for 30min prior to applicationonto a fixed embryo. Neuropeptides used were as follow:HPLC-purified CAP2, peptide F (thr-asn-arg-asn-phe-leu-arg-phe-amide) kindly supplied by Dr B. A. Trimmer (Trim-mer etal. 1987), and two molluscan cardioexcitatory peptides;FMRFamide (phe-met-arg-phe-NH2; Greenberg and Price,1979), and one of the small cardioactive peptides (SCPD, met-asn-tyr-leu-ala-phe-pro-arg-met-NH2; Lloyd, 1978). Thestaining protocol used with these putative 6C5-antigen com-plexes was in all other ways identical.

ImmunohistochemistryEmbryos were dissected free of the chorion and yolk with fineforceps and the ventral CNS exposed. This dissection wasaccomplished by securely fastening the specimen in a Sylgard(Dow-Corning) dish containing Manduca saline, making an

Peptidergic modulation of the insect embryonic gut 61

anterior-directed incision through the dorsal body wall fromtail horn to head capsule, pinning open the body wall, andremoving the embryonic gut to expose the ventral nerve cord(VNC). Dissected embryos were incubated at 4°C with gentleagitation in a modified Bouin's/glutaraldehyde fixative (2 %glutaraldehyde, 25% saturated picric acid and 1% glacialacetic acid) for lh,washed three times in 0.4% saponin-PBSfor 30min each, and taken through an ethanol dehydrationseries at 4°C. Fixed specimens were incubated in collagenase( lmgmr1; Sigma, type XI) for lh, followed by extinction ofendogenous peroxidase activity with 0.75 % hydrogen per-oxide in methanol. Specimens were then blocked with goatserum (5 nig ml"1; Sigma lot no. 28F-9401) and bovine serumalbumin (1% w/v) in 0.4% saponin-PBS for 2h.

A three-tier antibody system using the peroxidase-anti-peroxidase (PAP) method was employed to identify cells withCAP-like immunoreactivity. Primary antibody (6C5; dilution1:1000), secondary antibody (whole molecule, goat anti-mouse IgG; dilution 1:100), and tertiary antibody (mouseperoxidase anti-peroxidase (PAP); dilution 1:100) were sus-pended with 1 % BSA in 0.4% saponin-PBS. Each antibodysuspension was incubated for 24 h at 4°C. Specimens werewashed five times in five hours with 0.4% saponin-PBScontaining 1 % BSA between successive incubations. Follow-ing equilibration in 3,3'- diaminobenzidine (DAB), immuno-reactivity was visualized by incubation in a solution of 0.8 %saponin-PBS containing DAB (O.Smgml"1), hydrogen per-oxide (0.003 % v/v) and NiCl2 (0.001 % w/v) until complete,usually 10-15 min. Visualized embryos were taken through anethanol dehydration series, equilibrated in xylene, andmounted in Permount for observation with Nomarski optics.

CAP extractionDevelopmentally staged embryos (Dorn et al. 1987; Broadie,1989e) were heat-treated for 5 min at 80°C, ice-cooled, andhomogenized in double-distilled H2O (ddH2O; 10^1/embryo)in a ground glass homogenizer. The homogenate was centri-fuged (15 min, 12000g, at 4°C) and the supernatant collected.The pellet was re-suspended in ddH2O (20 A«l/embryo), re-ground, centrifuged, and the supernatants from both extrac-tions pooled. The combined supernatant fraction was loadedonto a MeOH-activated, water-rinsed Waters C-18 Sep-Pakcartridge and washed in five times its volume with ddH2O.This was followed by step-wise applications of 20 % and 80 %acetonitrile (HPLC grade; J.T. Baker no. 9017-2) in ddH2O.From earlier studies (Tublitz and Truman, 1985a; Tublitz andEvans, 1986), it was known that both CAP] and CAP2 elute inthe 80% acetonitrile fraction. Accordingly, this fraction wascollected, frozen in dry ice, and lyophilized to powder.Lyophilized samples were stored at —20°C for up to onemonth before use. Immediately prior to bioassay, sampleswere brought to room temperature and re-hydrated in normalManduca saline.

CAP bioassayAn in vitro heart bioassay was used to quantify relative CAPlevels in each fraction as described earlier (Tublitz andTruman, 1985a). In short, a portion of the abdominal heartwas dissected from a pharate adult male immediately prior toeclosion. One end of the heart tissue was pinned in a smallsuperfusion chamber; the other end was attached with finesuture thread (Ethicon, 6-0) to a Bionix F-200 isotonic-displacement transducer powered by a Bionix ED1-1A Pow-erpack. The signal was amplified and displayed on a HitachiVC-6026 oscilloscope. Concurrently, the signal from the forcetransducer was passed through a frequency converter with awindow discriminator to determine instantaneous heart rate.

Heart amplitude and contraction frequency were recordedcontinuously on a Gould 2200 chart recorder for later analysis.

Manduca saline of the following composition was used in allexperiments: Pipes biological buffer (dipotassium salt;Sigma), 5mM; CaCl2, 5.6mM; NaCl, 6.5mM; KC1, 28.5mM;MgCl2, 16 mM; dextrose, 173 min. The final pH was adjustedto 6.7±0.1 using a concentrated solution of HC1. During eachbioassay. saline flow rate was maintained at approximately80mlh through the open superfusion chamber containingthe isolated heart. 100//I test samples were directly injectedinto the saline flow with a gas-tight Hamilton syringe.

Each injection of embryonic extract was bracketed withseveral graduated injections of known adult CAP activity. Theresultant dose-response curves of adult CAP activity enableda precise determination of an adult CAP activity equivalentfor the embryonic extract. Thus, the CAP level in eachembryo fraction was expressed as a percentage of the standardCAP levels in the abdominal nerve cord of the adult moth.Using these adult activity equivalents, measurements of theamount of CAP in embryo development stages could bequantitatively compared within the same bioassay, or be-tween different heart preparations.

Immunoprecipitation75 % DT Manduca embryos were extracted using the CAPextraction procedure as described above. The lyophilizedsample was resuspended in Manduca saline and half of thesample incubated with the 6C5 antibody at a dilution of 1 partantibody: 10 parts saline. The other remaining aliquot wasincubated with a non-reactive protein (1.0% BSA) as acontrol. Each aliquot was incubated at 4°C for 30min. Thesupernatant from each fraction was collected and bioassayedfor cardioacceleratory activity on the isolated pharate adultManduca heart as described above. Other controls included6C5 alone, BSA alone, 6C5+serotonin (Sigma) and6C5+peptide F (a crustacean cardioactive neuropeptide;Trimmer et al. 1987). Bioactivity of each treatment wascompared to that of an untreated control.

HPLCSep-Paked embryo extracts were chromatographed on aBrownlee Alltech C-18, reverse-phase HPLC column(4.6x250mm, 300jan particle size). An acetonitrile-watersolvent gradient with 0.1% trifluoroacetic acid (TFA)counter-ion was used in all trials. A linear acetonitrile-watergradient was used with the acetonitrile concentration increas-ing at 1.5% per min (Tublitz and Evans, 1986). For eachchromatography run, 30 separate 1 ml fractions were collectedat lmin intervals. Each fraction was lyophilized, stored at—20°C, and later re-suspended in Manduca saline for bioassayon the isolated heart.

Embryonic heart and gut bioassaysManduca embryos were dissected free of their chorion andyolk, and development stage was determined (Dorn et al.1987; Broadie et al. 1989a). Specific target organs, either theembryonic hindgut or heart, were exposed by making in-cisions through the body wall and pinning back the epidermiswith minute steel pins. When necessary, minimal surgery wasperformed to clearly expose the target organ. This semi-intactpreparation was placed in a small superfusion chamber andperfused with Manduca saline at 60 ml h"1.

HPLC-purified CAP2, and a range of other known neuro-hormones and transmitters, were applied to the preparationand the effect on myogenic contraction frequency in the heartand gut quantified. The applied substances included: smallcardioactive peptideB (SCPB; Lloyd, 1978), FMRFamide

62 K. S. Broadie and others

(Greenberg and Price, 1979), peptide F (Trimmer et al. 1987),proctolin (Sigma), serotonin (5-HT; Sigma), octopamine(Sigma), and acetylcholine. All substances were dissolved in100 fi\ of Manduca saline and injected directly into the salineflow with a gas-tight Hamilton syringe.

The target tissue was observed under high magnification(x800) with a Wild binocular dissecting microscope, andmyogenic contractions counted during 30 s intervals for up to8min following an injection. From these observations, con-traction frequency over 30 s intervals was computed. A secondinjection was applied if and only if the organ returned rapidlyto a basal contraction frequency.

Results

Spatial distribution of CAP-like immunoreactivityDistinctive patterns of CAP-like immunoreactivitywere observed repeatedly in the posterior abdominalganglia during the late stages (70-100 % DT) of Man-duca embryonic development. Cells showing CAP-likeimmunoreactivity fit into two spatially distinct groups.The first group consisted of a pair of large neurons thatlie along the posterior midline of the caudal abdominalganglia (A4-A8; Figs 1 and 2). The axons from thesecells bifurcate and exit the ganglion via each ventralnerve, ultimately projecting posteriorly to both ipsilat-eral and contralateral transverse nerves (data notshown). The second group consisted of lateral neuro-secretory cell clusters that lie at the base of each ventralroot in the abdominal ganglia (A2-A8; Figs 1 and 2).These lateral cells also have processes in the ventralnerve leading to the transverse nerve. Midline stainingwas observed in an average of two medial neurons(range: 1-3 cells) in all four posterior abdominalganglia. The mean number of CAP-like immuno-reactive cell bodies in each lateral cluster varied greatlydepending on position in the VNC (range: 2-8 cells/cluster). In general, more caudal ganglia contained ahigher number of immunoreactive somata in the lateralclusters as compared to the rostral ganglia (Fig. 2). Thecellular morphology of both types of CAP-like immuno-reactive cells was commensurate with observations ofother insect neurosecretory cells (Rowell, 1976): largeand clearly evident cell bodies displaying the TyndallBlue effect characteristic of neurosecretory function.

CAP-like immunoreactivity appeared earliest in de-velopment and at the highest intensity in the fusedterminal abdominal ganglion (A7/A8; Figs 3A and4A). Both immunoreactive midline and lateral cellswere observed in the terminal ganglion. The midlinecells in A8 appeared anterior and dorsal, just posteriorof the transverse nerve branching, whereas the midlinecells in A7 were found very posteriorly, just anterior tothe juncture between A7 and A8. These midline cellswere very large relative to other neurosecretory cells inthe ganglion, with cell body diameter ranging from15 fim to 20 /im. The cell bodies of these cells displayedstrong CAP-like immunoreactivity throughout late em-bryonic development (Figs 1, 3A and 4A).

The lateral cell clusters in the fused terminal ganglionalso stained intensely throughout the last quarter of

embryonic development. The lateral cells in A8 showeda distribution different from that of A7 and the anteriorunfused ganglia, in that the more numerous cell bodiesin A8 were less clustered and tended to be isolated intoone or two cells lying at the bases of the major posteriorventral root branches (Figs 1 and 2C). In this ganglion(A8), an average of six cell bodies containing CAP-likeimmunoreactivity were observed, ranging from 2-10cells per preparation. In contrast, the lateral cellclusters in A7 formed close-knit, bilaterally symmetri-cal clusters at the bases of the posterior ventral root(Figs 1 and 2C). Both groups of lateral cells in theterminal ganglion (A7/A8) had relatively small somata,with diameters in the range of 6^m to 11 fun each. In allpreparations, only the cell bodies of the lateral cellswere intensely immunoreactive.

In general, more CAP-like immunoreactive cellbodies were located in ganglion A8 than in the moreanterior ganglion. We surmise that this is due to the factthat ganglion 'A8' is actually a condensation of multipleneuromeres found during the initial formation of theCNS (Jacobs and Murphey, 1987). This unique develop-ment explains the different distribution of CAP-likeimmunoreactive cells in A8 relative to A7 and the moreanterior, unfused abdominal ganglia (Fig. 2).

Unfused abdominal ganglia showed immunoreactivepatterns similar to A7 (Figs 1 and 2). An average of twomidline cells per ganglion stained in ganglia A4 to A6(range: 1-3 immunoreactive cells per ganglion). Thecell bodies lay along the posterior, dorsal midline andwere often positioned asymmetrically across theganglion midline, with one cell body lying slightlyanterior relative to the other. The morphology and sizeof these cells was comparable to the midline cells in thefused terminal ganglion (Figs 1 and 2). The midlinecells in the unfused ganglia were less immunoreactivethan in the fused terminal ganglia, with intensity of theDAB reaction product progressively decreasing fromA6 to A4 (Fig. 4A). No midline cells showed CAP-likeimmunoreactivity in the first three abdominal gangliaduring any stage of embryo development.

The lateral cell bodies in the unfused abdominalganglia remained clustered at the base of the posteriorventral root in a pattern similar to A7 (Figs 1 and 2).The number of cells in the lateral clusters decreased inthe more anterior unfused ganglia: an average of fourcells per cluster in A6 (range: 4-5), three cells in A5(range: 2-4), and two lateral cells, one cell at the baseof each ventral root, in A2-A4 (range: 1-2). As withthe midline cells, immunoreactive staining intensityprogressively decreased in the lateral cells of the moreanterior abdominal ganglia: A6 lateral cells were themost intensely stained, and A2 lateral cells were theleast immunoreactive (Fig. 4B). Lateral cell body mor-phology and size were similar in all the abdominalganglia. CAP-like immunoreactivity was not detectablein the lateral cells of ganglion Al.

In summary, CAP-like immunoreactivity during em-bryogenesis was restricted to midline and lateral neuro-secretory cells in the abdominal ganglia, predominantlythe posterior five ganglia (A4-A8). Immunoreactivity

§

Fig. 1. Immunocytochemical labeling of neurosecretory cells in the embryonic abdominal ganglia of Manduca sexta.(A) Terminal, fused ganglion (A7/A8) in a 75% DT embryo. Notice intensely immunoreactive medial and lateral cellbodies. (B) Terminal ganglion (A7/A8) at 80% DT. Immunoreactivity in lateral and medial cells is maintained only in theterminal ganglion. (C) Unfused (A5) ganglia in a 75% DT embryo showing immunoreactive midline and lateral cell bodies.(D) Ganglion A5 at 80 % DT with selectively decreased immunoreactivity in lateral cells. Note that medial cells still displaystrong CAP-like immunoreactivity. (E) Terminal ganglion (A7/A8) in the 100 % DT embryo prior to hatching, intenselyimmunoreactive lateral and midline cells. (F) Terminal ganglion (A7/A8) 1 h after hatching of first instar larvae. Notice theselective decline in medial cell immunoreactivity relative to staining in lateral cells. (G) Unfused ganglion (A5) at 100 % DTprior to hatching. (H) Unfused (A3) ganglion 1-2 h after hatching. Observe selective decrease in medial cell labelingcompared to (G). (I) Pre-incubation of the anti-CAP antibody with HPLC-purified CAP2 abolished observedimmunoreactivity patterns. Pre-incubation with four other invertebrate cardioexcitatory peptides (see Materials andmethods) had no effect on observed immunoreactivity. Scale bar, 50/on.

Peptidergic modulation of the insect embryonic gut 63

A . Abdominal VentralNerve Cord (VNC)

A1

A2

Midline Cells Lateral Cells

VN

Midline'Cell

Lateral'Cell

Unfused AbdominalGanglion (A4)

A3

A4

A5

A6

A7

A8

Fused, TerminalGanglia (A7/A8)

Fig. 2. Spatial location of CAP-like immunoreactive cells inthe abdominal VNC of Manduca sexta embryo.(A) Schematic representation of the abdominal VNCshowing all cells with CAP-like immunoreactivity presentduring embryogenesis. Mean number and relative positionof medial (midline) and lateral cells demonstrating CAP-like immunoreactivity are indicated. (B) Camera lucidadrawing of lateral and medial immunoreactive cell bodies inan unfused abdominal ganglion (A4) of a 85 % DT embryo,tn, transverse nerve; dn, dorsal nerve; vn, ventral nerve.(C) Camera lucida drawing of immunoreactive cell bodiesin the fused, terminal abdominal ganglion (A7/A8) of a85 % DT embryo.

was limited to the cellular cytoplasm with no observednuclear staining. In most preparations, CAP-like immu-noreactivity was limited to the cell body; however, thebase of the transverse nerve also demonstrated CAP-like immunoreactivity in approximately 10% of ourtrials. No CAP-like reactivity was observed in thethoracic ganglia or in the brain during embryonicdevelopment.

Temporal pattern of CAP-like immunoreactivityCAP-like immunoreactivity first appeared in the 60%developed embryo. The first cells to demonstrate im-munoreactivity were the two pairs of large, dorsalmidline cells in the terminal abdominal ganglion (A7/

A1

v A2Vn

i wco3 A4cooo AS

1 «A7 : i

JO K «U

iii\r

••••R S 8 8 S S R 8 3 8

Percent Embryonic Development

Fig. 3. Percentage of trials possessing CAP-likeimmunoreactive cells in the abdominal ganglia during lateembryonic development (60-100% DT). (A) Percent ofembryo preparations with immunoreactive midline cells.Notice the gradual increase in the number of ganglia withimmunoreactive cells and the percent increase in positivetests throughout embryogenesis, followed by a drop athatching. (B) Percent of embryo specimens withimmunoreactive lateral cells. Notice the drop in positivestaining trials between 75 % and 85 % development.

A8; Figs 3A and 4A). Immunoreactive intensity wasvery weak in these cells in the 60 % developed embryos,and positively stained cells were only observed in asmall fraction of our animals (12%; Figs 3A and 4A).

The number of immunoreactive midline cells and theimmunoreactive intensity of these cells increased gradu-ally during the later stages of embryonic development(70-100 % DT; Figs 3A and 4A). After the first appear-ance of CAP-like immunoreactivity in the midline cellsof the fused, terminal ganglion, the homologous cells inan unfused abdominal ganglion (A6) first appeared inthe 70% developed embryo. However, only a smallfraction (10%) of the preparations exhibited positivestaining. By 75 % DT, the percentage of preparationswith immunoreactive midline cells in A6 had increasedto 25 %, and the immunoreactive intensity of these cellshad become more pronounced (Figs 3A and 4A).Midline cells in A5 first demonstrated immunoreactivityat 80% development. From 80 to 95% DT, no newmidline cells with CAP-like immunoreactivity appeared(Fig. 3A). Immunoreactive intensity also remainedfairly constant from 80 to 90% DT, but began toincrease in the midline cells of the last four abdominalganglia by 95% DT (Fig. 4A). Immunoreactive A4midline cells, the most anterior midline cells to haveCAP-like immunoreactivity during embryogenesis,were first detected in the 100%, fully developed em-bryo, just prior to hatching. In the fully developedembryo, a high percentage of preparations had immu-noreactive midline cells in the last five abdominal

64 K. S. Broadie and others

A Midline Cells B Lateral Cells

F 1 " J13

i!i40

Iir

J

!

n

100 (»

3

2 K R 8 2

Percent Embryonic Development

Fig. 4. Intensity of immunocytochemical reaction inembryonic abdominal VNC cells. (A) Midline cells.Relative intensity of immunochemical reaction scaled withthe most intense reaction arbitarily assigned maximumintensity (1=1) and immunoreactive intensity in othermedial cells scaled relative this standard. Notice theincrease in relative reaction intensity during embryodevelopment and the drop in intensity at hatching.(B) Lateral cells. Relative intensity scaled to most intenselateral cell immunoreactivity (1=1). A large drop in CAP-like immunoreactivity occurs between 75 % and 80 %development.

ganglia, ranging from 75 % positive staining in A8 downto 10% staining in A4 (Fig. 3A). In addition, theintensity of CAP-like immunoreactivity peaked in themidline cells within the last two hours of embryonicdevelopment.

A marked decrease in midline cell CAP-like immuno-reactivity was observed at hatching. During larvalemergence, the number of immunoreactive midlinecells and the intensity of their staining decreased(Figs 1E-H, 3A, and 4A). Specifically, midline cellCAP-like immunoreactivity in ganglia A4 and A5disappeared completely and the midline cell immuno-reactive intensity decreased markedly in A6 through A8(Figs 3A and 4A).

By way of contrast, the lateral neurosecretory cellsacquired CAP-like immunoreactivity in a developmen-tally unique pattern, distinct from the midline cells.First appearing at 70% DT, lateral cell immunoreac-tivity peaked in the 75 % developed embryo (Figs 1A,Cand 4B). At this stage we saw the largest number ofCAP-like immunoreactive lateral cells (Fig. 3B), themost intense CAP-like reactivity in these cells(Fig. 4B), and the greatest number of abdominalganglia (A2-A8) possessing immunoreactive lateralcells (Fig. 3B). A sharp decline in lateral cell immuno-reactivity was observed between 75 % and 80 % DT:lateral cell immunoreactivity in the more anterior twoabdominal ganglia (A2-A3) disappeared completely,while the percentage of CAP-like reactive cells and the

immunoreactive intensity of these cells declined mark-edly in all the remaining posterior abdominal ganglia(A4-A8; Figs 3B, 4B and 1A-D). By 85% DT, bothimmunoreactive intensity and the total percentage ofCAP-like immunoreactive lateral cells began to recoverin A5 through A8, whereas no recovery was detected inthe lateral cells of ganglion A4 at this stage of develop-ment (85 % DT).

CAP-like immunoreactivity of the lateral neuro-secretory cells continued to recover during the last 15 %of development. By 95 % DT, a large number (Fig. 3B)of intensely immunoreactive lateral cells were onceagain observed in A4-A8. Less intense staining wasseen in ganglia A3. The immunoreactivity of the lateralcells plateaued at this level for the remainder ofembryonic development (Figs 3B and 4B). Unlike themidline cells, lateral cell immunoreactivity remainedconstant during hatching behavior (Fig. 1E-H).

In summary, we found that levels of the CAP-likeantigen were modulated independently in the twoabdominal ganglia cell types. Expression of the CAP-like antigen in the midline cells was characterized byearly development (60% DT), a gradual continuousincrease in immunoreactivity during embryogenesis,and a marked drop in the levels of the immunoreactiveantigen during hatching. In contrast, lateral cells ex-pressed the CAP-like antigen later at 70 % DT, rapidlyaccumulated levels of the antigen during the next 5 % ofdevelopment, and demonstrated a dramatic drop inCAP-like immunoreactivity between 75 % and 80 % ofembryo development.

Antibody specificityTo reaffirm its specificity, our primary antibody (6C5)was pre-absorbed with a variety of antigens to helpverify the nature of molecule responsible for the ob-served immunoreactivity patterns. All CAP-like immu-noreactivity was completely abolished after pre-incu-bation of the primary antibody in CAP2 (Fig. li). Incontrast, pre-incubation of 6C5 with other invertebratecardioactive peptides (peptide F, SCPB and FMRF-amide) in no way altered the immunoreactivity patternsobserved with the free antibody (data not shown).

Levels of cardioacceleratory activity during embryonicdevelopment in ManducaThe immunohistochemical observations describedabove clearly demonstrated the presence of a CAP-likeantigen in a well defined subset of neurosecretory cellsin the embryonic caterpillar. We next carried out aseries of biochemical tests to establish the relationshipbetween this antigen and the CAPs. Embryos at variousdevelopmental stages were extracted for cardioacceler-atory activity as described in Materials and methods,with the crude homogenate partially purified through aC-18 Sep-Pak. The resultant fractions then were as-sayed for biological activity on the isolated Manducaheart. Using this procedure, we determined that allembryonic development stages possessed some degreeof cardioacceleratory bioactivity. Even samples of

Peptidergic modulation of the insect embryonic gut 65

o

i

Io

a.

i"c««

Q-

20-•

10--

un 0 10 20 30 40 50 60 70 75 80 85 90 95 100pro/post

Percentage Embryo Development Time

Fig. 5. Changes in cardioacceleratory levels duringManduca embryogenesis. CAP2 activity was assayed on thein vitro pharate adult Manduca heart as described inMaterials and methods. An ANC unit refers to the CAPactivity level found in the standardized pharate adult mothabdominal nerve cord. Activity was first detected at 40% ofembryo development. Two drops in cardioacceleratorylevels occurred during development: between 75% and80% development, and at hatching. Cardioacceleratoryactivity in embryo extracts of all developmental stages wascorrected for the minor yet detectable activity found in theunfertilized eggs (UN), as described in the Results. Pre/Post 100% DT refers to 1-2 h before hatching and 1-2 hfollowing hatching, respectively. Each bar represents themean±s.E.M. of at least ten independently pooleddeterminations.

unfertilized Manduca eggs, purified with the identicalpurification scheme, contained very low yet measurablelevels of cardioacceleratory activity (0.9% of the ac-tivity in an adult abdominal nerve cord (ANC); «=10).Thus, all measurements of the embryonic cardioexcit-atory levels found at later development stages werecorrected for this background cardioacceleratory ac-tivity by subtracting the level of unfertilized egg bioac-tivity from the observed response of the pharate adultheart to a given sample.

Cardioacceleratory activity substantially above back-ground levels was first detected in the 40 % developedembryo (Fig. 5). Extracts of this stage had 2.0±0.8%(mean±s.E.M.) of the cardioacceleratory activity in anadult ANC as determined by quantitative bioassay onthe isolated pharate adult heart (Fig. 5). This activityincreased rapidly between 40 % and 75 % DT, where itpeaked at 14±3.8% of the adult ANC levels. Cardioac-celeratory bioactivity in the entire embryo decreasedmarkedly between 75 % and 80 % DT, dropping to4.5±1.2% of adult ANC cardioacceleratory activity(Fig. 5). A gradual increase in bioactivity was observedfrom 80 % to 100 % DT, with fully developed embryoscontaining 8.2±1.8 % adult ANC units of CAP activity.A second, substantial drop in cardioacceleratory ac-tivity was recorded at hatching when bioactivity levelsdecreased by approximately 50 %, with newly-hatchedfirst instar larvae having 4.3±0.9 % of adult ANC levels(Fig. 5).

oDoin

100-

75-

50-

2 5 - •

75X DT EmbryoExtract

7 5 * DT EmbryoExtract

+6C5 Antibody

Fig. 6. Immunoprecipitation of embryonic cardio-acceleratory activity by the anti-CAP antibody, 6C5.Bioactivity is expressed as a percentage of cardioexcitatoryactivity in the experimental sample compared to that of theuntreated embryonic extract as determined on the in vitroheart. 6C5 had no effect on the activity of biogenic amines(5-HT) or another invertebrate cardioactive peptides(peptide F). 6C5 alone did not effect cardiac contractionfrequency. Each bar represents the mean±s.E.M. of at leastfive determinations.

Table 1. An anti-CAP antibody (6C5) selectivelyprecipitated the cardioactive factor found in the

partially-purified Manduca embryo extracts. 6C5 hadlittle or no effect on a biogenic amine (5-HT) or on

another arthropod cardioactive neuropeptide(peptide F)

Treatment % Bioactivity

CAP2 (control)CAP2+6C55-HT (control)5-HCT+6C5Peptide F (control)Peptide F+6C5

Each value represents the mean±s.E.M.determinations.

100±4%19±5%

100±2%100±3%100±4 %96±7%

of at least five

Biochemical identification of cardioacceleratory activityin the Manduca embryoAs a first step in identifying the cardioactive factorfound in the partially-purified embryonic extracts, amonoclonal anti-CAP antibody, 6C5, (Taghert et al.1983, 1984) was tested for its ability to precipitate thebioactive factor. This antibody has been previouslydemonstrated to specifically recognize an epitope com-mon to CAP1 and CAP2 (Tublitz and Evans, 1986).When extracts from 75 % developed embryos wereincubated with the 6C5 antibody for 30min prior tobioassay on the in vitro Manduca heart, bioactivity wasreduced by 81 % when compared to untreated controls(Fig. 6; Table 1). The 6C5 antibody alone produced nodetectable effect when applied to the isolated pharateadult heart. Furthermore, the 6C5 antibody had noeffect on the bioactivity of a biogenic amine (serotonin)

66 K. S. Broadie and others

10

a

6-

4

2

CAPn

rh

PHARATEADULT

CAP,

60

50

12 13 14 15 16 17 18 19 20 21 22

Qution Time (min)

30

ftoo

EZ

°1

8

6'

4-

2

0

7 5 * DTEMBRYOS

rh

T 60

50

3012 13 14 15 16 17 18 19 20 21 22

Qution Time (min)

Fig. 7. Cardioacceleratory activity in Manduca embryoextracts co-elute with CAP2 when purified with highpressure liquid chromatography (HPLC). (A) The elutionprofile of both CAPs isolated from the abdominal VNC ofthe pharate adult moth. An acetonitrile/water gradient wasused starting with 20% acetonitrile at t=0 and withacetonitrile concentration increasing at 1.5%/minute. CAP2elutes at 15 min (42.5% acetonitrile) and CAPi at 19 min(49% acetonitrile). After separation on the HPLC column,fractions were bioassayed on the in vitro heart as describedin Materials and methods. (B) Elution profile ofcardioacceleratory activity from 75 % DT embryos. Allbioactivity co-eluted with CAP2 (15 min; 42.5 %acetonitrile) in a single peak. No other cardioactivity wasfound to co-elute with CAPi or elsewhere on the column.Bars represent the mean±s.E.M. of at least tendeterminations.

or on peptide F (Trimmer, 1987), another arthropodcardioactive peptide (Table 1).

To unequivocally ascertain the relationship betweenthe CAPs and the cardioactivity in embryos, partially-purified material from embryos at various stages waschromatographed on a HPLC using a reverse phaseC-18 column, and the resultant fractions bioassayed onthe in vitro Manduca heart. Chromatographic profiles,such as the one shown in Fig. 7, indicated the presenceof a single peak of cardioexcitatory activity during lateembryonic development. This activity co-eluted withpurified CAP2 from the pharate adult moth. No detect-able cardioregulatory bioactivity eluted in the fractionsassociated with CAPi, or elsewhere during the chroma-tographic run (Fig. 7). This sole activity peak accounted

Table 2. Sensitivity of the Manduca embryonic heartand hindgut to pulse applications of CAP2 and other

selected myoactive factorsThreshold (M)

SubstanceEmbryonic

heartEmbryonic

gut

SerotoninOctopaminePeptide FFMRFamideSCPBProctolinAcetylcholineCAP2

io-9

10"8

io-7io-5

>10"5

>io-4>10"3

0.2 ANC

io-9io-710"8

>10"4

>10~3

>10"5

>10"3

0.05 ANC

Threshold is defined as the lowest concentration of the substancethat produces a measurable (5%) increase in contraction frequencyin 50% of the trials. CAP2 thresholds are given in terms of thestandard CAP2 levels in the abdominal ventral nerve cord (ANC)of the adult moth.

for all the cardioacceleratory activity present in theembryonic crude extracts. Cardioacceleratory activitycould not be detected in FIPLC samples from embryosless than 50% developed, even with sample sizes tentimes larger than those assayed with older embryos.

Pharmacology of the in vitro Manduca embryonicheart and gutOur biochemical results, combined with our previouslydescribed immunocytochemical observations (Figs 1, 3and 4), clearly demonstrate that the level of CAP2fluctuates dramatically during embryogenesis, and thatthe observed drops in peptide levels may be related toCAP2 release from individual neurosecretory cells.Hence, we were interested in the possible physiologicalrole(s) of this neuropeptide in the embryonic system.As a first step in addressing this question, we analyzedthe effect of exogenously applied CAP2, as well asseveral other known invertebrate cardioregulatory fac-tors, on the beat frequency of myogenically activeembryonic tissues utilizing an in vitro preparation asdescribed in Materials and methods. In particular, wewere interested in the effects of CAP2 on the heart andhindgut, as both tissues are known targets of CAPmodulation during larval (N. Tublitz, unpublished ob-servations) and adult life (Tublitz and Truman,1985a,b,c; Tublitz and Evans, 1986; Tublitz, 1989).

The contraction frequency of both the embryonicheart and hindgut were found to be particularly sensi-tive to three known neuroactive substances: two bio-genic amines, serotonin and octopamine, and peptideF, a small neuropeptide isolated from crustaceans (thr-asn-arg-asn-phe-leu-arg-phe-amide; Trimmer et al.1987; Table 2). In both embryonic tissues, serotonin (5-HT) had the lowest threshold (10~9M), where thresholdis defined as the lowest concentration required to evokea measureable increase (5 %) in contraction frequencyin 50% of the trials (Table 2). The thresholds foroctopamine and peptide F were at least an order ofmagnitude higher for both assays. The heart was more

Peptidergic modulation of the insect embryonic gut 67

sensitive to octopamine (threshold 10 8M) than topeptide F (threshold 10~7M). In contrast, the hindgutwas more sensitive to peptide F (threshold 10~8M) thanto octopamine (threshold 10~7M; Table 2).

The embryonic heart and hindgut were significantlyless sensitive to the other substances applied during thisstudy. These included three other invertebrate bioac-tive peptides, e.g. FMRFamide, small cardioactivepeptide (SCPB), and proctolin, as well as acetylcholine,another putative insect cardioregulatory substance(Miller, 1979). The thresholds for the three peptideswere of the order of 10~5M or greater when pulse-applied onto either in vitro preparation (Table 2).Acetylcholine had no detectable effect in either assay atconcentrations up to 10~3M.

CAP2 was found to increase contraction frequency inboth the embryonic heart and hindgut. Of the twoorgans, the hindgut was more sensitive to CAP2 appli-cation, with a lower threshold (0.05 adult ANC equival-ents of HPLC-purified CAP2 activity) and a moreprolonged response (response to 1.0 ANC equivalentCAP2: 5.0±0.3min; Fig. 8A). The dose-responserelationship followed a sigmoidal curve with maximalresponse occurring at a pulse application of 0.24 ANCCAP equivalents (Fig. 8B).

In contrast, the embryonic heart had a thresholdsensitivity of 0.2 ANC equivalents (Table 2). The re-sponse latency to exogenously applied CAP2 (1.0 ANC)was similar in both organs, but the duration of thecardiac response was significantly reduced(3.4±0.4min) relative to that of the hindgut (Fig. 9A).A similar sigmoidal dose-response relationship wasobserved in both organs. However, the lower sensitivityof the embryonic heart resulted in a displacement of thecurve to the right, with a maximal increase in cardiaccontraction frequency observed to 0.52 ANC CAPequivalents (Fig. 9B).

Development of spontaneous activity and CAP2sensitivity of the embryonic Manduca gutThe spontaneous activity of the embryonic gut wasmeasured in vivo and in vitro during the last 50% ofembryogenesis. In vivo observations of gut activitywere possible prior to 80% DT through a semi-translucent embryonic cuticle. However, the increase ofcuticle pigmentation after this stage prevented furtherin vivo observations. Comparison of contraction ratesbetween the in vivo gut and our semi-intact in vitropreparation indicated that contractile activity was notperturbed during our manipulations. Specifically, con-traction rates of the two preparations remained inagreement during early development and gut activitydid not change in response to direct physical stimu-lation. Consequently, both assays were used interchan-geably to track embryonic gut activity.

The gut was quiescent in all embryos prior to 65 %DT (Fig. 10A). Spontaneous gut contractions began at70% of embryonic development, increased dramati-cally during the next 10% DT and reached peakcontraction rates by 80% of development (Fig. 10A).Gut contraction frequency remained relatively constant

B

co

uo

Adult ANC Equivalents of CAP2

Fig. 8. (A) Response of the embryonic Manduca hindgut topulse application of 1 ANC equivalent of HPLC-purifiedCAP2. Vertical error bars represent S.E.M. of the meanheart response; horizontal error bars represent the S.E.M. tothe mean heart response time. (B) Dose-responserelationship of the Manduca embryo hindgut to pulseapplication of HPLC-purified CAP2. In each case, thepeptide was dissolved in 100 u\ Manduca saline and directlyapplied to a semi-intact in vitro preparation as described inMaterials and methods. Each point represents themean±s.E.M. of at least ten determinations.

at this level for the remainder of embryogenesis. Aslight increase in contraction rate was observed immedi-ately following hatching in the first instar larvae(Fig. 10A).

HPLC-purified CAP2 was applied to each of thedevelopmentally staged hindgut preparations, and re-sponse in contraction frequency quantified. We foundthat 0.25 ANC units of exogenously applied CAP2 hadno discernible effect on the gut prior to 70 % develop-ment (Fig. 10B), stages when the gut is normallyinactive (Fig. 10A). In contrast, the spontaneouslyactive gut was found to be very responsive to pulseapplications of CAP2. Application of 0.25 ANC units ofCAP2 at 75 % DT increased contraction frequency by40±4% (Fig. 10B). Sensitivity to CAP2 did not changesignificantly during the remainder of embryonic devel-opment.

68 K. S. Broadie and others

o.c

B

A

Adult ANC Equivalents of CAP2

Fig. 9. (A) Response of the embryonic Manduca heart topulse application of 1 ANC equivalent of CAP2 (see Fig. 5for details). (B) Dose-response relationship of theembryonic heart to pulse application of HPLC-purifiedCAP2. In each case, the transmitter was dissolved in 100/JManduca saline and directly applied to the dissected in vitrosystem as described in Materials and methods. Each pointrepresents the mean±s.E.M. of at least ten determinations.

Discussion

Comparison of CAP distribution in Manducathroughout developmentTaylor and Truman (1974) were the first to identify twodistinct groups of neurosecretory cells in the abdominalganglia of the adult moth. Their work demonstrated theexistence of four pairs of large (25-30 (ion) somata lyingalong the dorsal midline and a bilateral set of foursmaller neurons that lay at the base of each ventral root.Later work (Taghert and Truman, 1982) showed thatboth sets of neurons projected to the transverse nerve,the major neurohaemal release site in the insect ventralnerve cord (Raabe, 1982). Additional investigations,using a variety of techniques including intracellularstimulation of single identified peptidergic neurons(Tublitz and Truman, 1985c) and immunocytochemistry(Taghert et al. 1984, 1985; Tublitz and Sylwester, 1988)revealed that the four pairs of segmentally-reiteratedmedial cells in the adult moth synthesized and releasedhigh levels of cardioacceleratory peptide (CAP) ac-tivity, whereas the lateral cells expressed another

Notch ling

Percent Embryonic Development

B

hatching

Percent Embryonic Development

Fig. 10. (A) Spontaneous hindgut contractions in theManduca embryo during the last 50 % of embryogenesis.The gut was inactive prior to 70 % DT and increasesmarkedly in spontaneous activity between 70 % and 80 %development. (B) Response of the in vitro embryonichindgut to pulse application of 0.25 ANC equivalents ofCAP2 during the last 50 % of embryo development. Theinactive gut did not respond to the exogenously appliedpeptide. Each point represents the mean±s.E.M. of 5independent determinations.

neuropeptide, bursicon (Taghert and Truman, 1982). Incontrast, studies in larvae indicated that both cell typescontained CAP activity including the lone pair ofmedial cells that arises embryonically and the fourlaterally-situated pairs lying at the base of each ventralroot. Our results with the anti-CAP antibody in theManduca embryo (Figs 1 and 2) closely resembled thatobserved during the initial stages of post-embryonicdevelopment. As in early larvae (Tublitz and Sylwester,1988, 1989), staining was limited to a single pair ofmedial cells and the lateral clusters at the base of theventral root. All these cells had projections out theventral nerve, apparently extending at least to the baseof the transverse nerve, which also expressed strongCAP immunoreactivity. Given the highly specificnature of the anti-CAP antibody (Fig. II; Tublitz andEvans, 1986; Taghert et al. 1983, 1984) and that thisantibody stains a set of cells in the Manduca embryothat closely resemble the well-characterized CAP-con-taining neurons in larvae, we conclude that these

Peptidergic modulation of the insect embryonic gut 69

embryonic cells are likely to express one or both of theCAPs.

Identification of CAP2 in the Manduca embryoThree lines of evidence strongly suggest that CAP2 ispresent during embryonic development in M. sexta, andthat the level of this neuropeptide changes in dramaticand potentially physiologically important ways. First,immunohistological work with an anti-CAP antibodydemonstrated the presence of an immunoreactive anti-gen specifically localized in embryonic neurosecretorycells known to contain CAP2 during larval and/or adultlife (Tublitz and Truman, 1985c; Tublitz and Sylwester,1988, 1989). This embryonic staining pattern was com-pletely abolished after pre-incubation of the anti-CAPantibody with CAP2 (Fig. II). In contrast, pre-incu-bation of the anti-CAP antibody with a range of similarinvertebrate cardioactive peptides resulted in no changein the observed immunoreactivity pattern. Conse-quently, we were inclined to attribute the observedimmunoreactivity patterns to the presence of CAP2 inthe Manduca embryo.

It is, however, very rarely possible to unequivocallyidentify an antigen based on immunoreactivity alone.Absolute identification of an antigen can be achievedonly by separating it from tissue extracts, purifying it,and subjecting the pure sample to specific analysis usingchemical and/or physiological assays. Consequently, asa second line of evidence, extracts of embryos wereprepared according to the isolation procedure for theadult CAPs and quantitatively tested for CAP-likeactivity on a highly sensitive in vitro heart bioassay(Tublitz and Truman, 1985a,b). Our biochemical andpharmacological results showed that Manduca embryoscontained substantial levels of CAP-like cardioaccelera-tory activity (Fig. 5) which changed during embryonicdevelopment in a manner consistent with the changes inimmunohistological patterns and immunoreactivityintensities (Figs 3, 4 and 5).

In general, the fluctuations in activity from immuno-cytochemistry and the in vitro heart bioassay followedcomparable time courses. Although the isolated heartbioassay indicated the appearance of CAP2-like activityat an earlier development time (40%) than the immu-nocytochemical results (60%), it is apparent that thebiological cardioactivity of this factor was very low,below 60% DT (Fig. 5). This discrepancy may be dueto the appearance of another cardioexcitatory sub-stance at stages before 60 % DT which is not recognizedby the 6C5 antibody. Alternatively, endogenous CAP2levels before 60% DT may be below the level ofresolution by our immunocytochemical techniques. It isclear, however, that the first significant increase incardioacceleratory activity (60%) precisely correlateswith the appearance of the first anti-CAP immuno-reactive cells (Figs 3 and 5). Between 60 and 75% ofembryo development, both CAP-like immunoreactivityand cardioacceleratory activity levels increased sharply(Figs 3, 4 and 5). Even more compelling, the radicaldecline in cardioacceleratory activity between 75 % and80% development (Fig. 5) was paralleled by a similar

drop in immunoreactivity of the lateral cell clustersduring this developmental span (Fig. 1). Similarly,immunocytochemical (Figs 3 and 4) and biochemicalevidence (Fig. 5) both indicated a gradual increase inCAP-like activity during the remainder of embryonicdevelopment, culminating in a second drop in CAP-likecardioactivity and midline cell immunoreactivity duringhatching.

As the last line of evidence, the embryonic cardioac-celeratory activity was demonstrated to be attributableto CAP2 using two independent tests. In the first, it wasshown that over 80 % of the embryonic bioactivity wasimmunoprecipitated upon incubation with the anti-CAP antibody 6C5 (Fig. 6). Additionally, CAP2 wasshown to specifically block the CAP-like immunocyto-chemical staining patterns in the Manduca embryo(Fig. II). In the second, the embryonic cardioaccelera-tory activity was shown to precisely co- elute with CAP2purified from pharate adult moths using high pressureliquid chromatogTaphy (Fig. 7). Moreover, all cardioac-celeratory activity present in the late embryo stageseluted in a single peak, with no other cardioexcitatoryfactors detected in our tissue extracts at any embryonicstage (Fig. 7B).

The above data, taken together, strongly support thenotion that the observed biological activities are due toCAP2. Furthermore, the close correlation of the embry-onic CAP2 activity profile and the observed immuno-cytochemical staining patterns, as well as the absence ofany other cardioexcitatory factors in the tissue extracts,strongly argues that the CAP-like immunoreactivity inthe identified neurosecretory cells is directly attribu-table to the expression of CAP2.

What is the physiological role(s) of CAP2 in theManduca embryo?Biochemical, pharmacological, and immunohisto-chemical evidence all argue that a large drop in thestorage levels of CAP2 occurs twice during embryodevelopment: between 75% and 80% development,and during hatching (Figs 1, 3, 4 and 5). These findingsimply that CAP2 may play an important physiologicalrole(s) during the course of embryogenesis. In anattempt to define the role(s) of CAP2 in the embryo,experiments were initiated to identify possible targettissues that CAP2 may modulate during embryo devel-opment. Our results demonstrate that the contractileactivity of the embryonic heart and gut musculature ispharmacologically modulated by exogenous applicationof several myoactive factors (Table 2). In particular,HPLC-purified CAP2, when applied to the in vitroembryo, accelerated contractile rates of both the heartand hindgut (Figs 8 and 9).

Defining the precise physiological role of a neuro-chemical in vivo is always a difficult task. In thisinstance, given the scale of the Manduca embryo,direct, unequivocal identification of a physiological rolefor this peptide during embryogenesis may prove unfea-sible. Nevertheless, several lines of indirect evidencelead us to propose a possible role for CAP2 during itsapparent release between 75 % and 80 % of embryo

70 K. S, Broadie and others

development. First, our work on Manduca develop-ment demonstrated that the hindgut initially becomesmyogenically active at 70% DT (Broadie et al. 1989a),and that the rate of gut contraction increases radically(3-4 fold) between 70% and 80% DT (Fig. 10A).Second, ingestion of the extra-embryonic yolk com-mences approximately at 75 % DT (Broadie, unpub-lished observations). At this stage, the gut is empty ofyolk and, by 85 % DT, ingestion of the extra-embryonicyolk is complete, resulting in a yolk-filled gut (Broadie,unpublished observations). Third, the embryonic gut isvery sensitive to CAP2 (Fig. 8), which, at physiologicalconcentrations, accelerates contraction frequency in afashion similar to the endogenous modulation of the gutobserved between 70% and 80% DT in vivo(Fig. 10A). Fourth, the pharmacological sensitivity ofthe embryonic gut to exogenously applied CAP2 in-creases markedly between 70% and 80% DT(Fig. 10B). It is particularly striking that the inactive gutat stages before 70 % DT is not responsive to exogenousapplication of the peptide and only becomes sensitive toCAP2 immediately prior to the acceleration of gutcontractions seen at 75% DT (Fig. 10). Fifth, ourbiochemical studies of CAP2 levels in the embryoindicated a sharp drop in stored CAP2 levels between75 % and 80 % development (Fig. 5). Lastly, our immu-nocytochemical studies (Figs 1, 3 and 4), showed amarked decrease in the CAP2-like immunoreactivity ofidentified lateral neurosecretory cells in the posteriorabdominal VNC between 75 % and 80% development,confirming our biochemical results. Interestingly, alarge proportion of this CAP2-like immunoreactivitywas localized in the terminal abdominal ganglia, whichother studies have indicated is the primary control siteof the digestive hindgut (Raabe, 1982). Taken together,these data provide strong circumstantial evidence thatCAP2 may be released from identified lateral neurosec-retory cells in the abdominal VNC between 75 % and80% of embryonic development, facilitating ingestionof the extra-embryonic yolk by directly accelerating thefrequency of hindgut contractions.

The role of CAP2 secretion during hatching behavioris less clear. Our biochemical studies have shown that asimilar, if somewhat smaller, decline in CAP2 levels inthe embryo occurs at hatching (Fig. 5). Interestingly,the drop in CAP-like immunoreactivity at hatchingoccurs exclusively in the large medial neurosecretorycells of the abdominal VNC as opposed to the decreasein lateral cell immunoreactivity observed at 75 % DT(Figs 1, 3 and 4). This raises the tantalizing possibilitythat the two cell types may be independently regulated,playing distinctive roles at different stages in the Man-duca embryo. CAP2 secretion at hatching can behypothesized to increase heart contraction providingthe necessary hydrostatic force to aid hatching and/or itmay increase gut contraction to facilitate the embryo'sability to eat its way clear of the egg shell and ingest partof the shell immediately after emergence. However, it isobvious that additional work must be done to betterunderstand the exact physiological role of CAP2 duringeither embryonic period.

We would like to thank Dr J. C. Weeks for her commentson earlier versions of this manuscript and Dr B. Trimmer forhis gift of peptide F used in this study. We also thank Mr T.Schilling for his input and advice during the preliminary stagesof the immunocytochemical work. This work was supportedby NIH grant no. NS 24613, a NIH Research DevelopmentAward no. NS 01258, and a Sloan Fellowship to N.J.T.

References

BELL, R. A. AND JOACHIM, F. A. (1978). Techniques for rearinglaboratory colonies of tobacco hornworms and pink bollworms.Ann. em. Soc. Amer. 69, 365-373.

BROADIE, K. S., BATE, C. M. AND TUBUTZ, N. J. (1989a).

Quantitative staging of embryonic development of the tobaccohawkmoth, Manduca sexta. J. exp. Biol., (submitted).

BROADIE, K. S., SYLWESTER, A. W., BATE, C. M. AND TUBUTZ, N.

J. (1989). Embryonic expression, modulation and function of aneuropeptide in a moth. Soc. Neurosci. Abstr. 15, 26.

DORN, A., BISHOFF, S. T. AND GILBERT, L. I. (1987). Anincremental analysis of embryonic development of the tobaccohornworm, Manduca sexia. Int. J. Invert. Reprod. Develop. 11,137-158.

FRONTAU, N. AND GAINER, H. (1977). Peptides in invertebratesystems. In Peptides in Neurobiology. (ed. Gainer, H.), pp.359-394. Plenum Press: New York.

GREENBERG, M. J. AND PRICE, D. A. (1979). FMRFamide, acardioexcitatory neuropeptide of molluscs: An agent in search ofa mission. Am. Zool. 19, 163-174.

GUILLEMIN, R. (1978). Peptides in the brain. Science 202, 390-402.JACOBS, G. A. AND MURPHEY, R. K. (1987). Segmental origins of

the cricket giant interneuron system. J. comp. Neurol. 265,145-147.

JAN, L. Y. AND JAN, Y. N. (1982). Peptidergic neurotransmission inthe sympathetic ganglion of the frog. J. Physiol. 327, 219-246.

LEVINE, R. B. (1986). Reorganization of the insect nervous systemduring metamorphosis. Trends Neurosci. 9, 315-319.

LLOYD, P. E. (1978). Distribution and molecular characteristics ofcardioactive peptides in the snail, Helix aspersa. J. comp.Physiol. 128, 269-276.

MAYEJU, E., ROTHMAN, B. S., BROWNELL, P. H., BRANTON, W. D.

AND PADGETT, L. (1985). Nonsynaptic characteristics ofneurotransmission mediated by ELH in the abdominal ganglionoiAplysia. J. Neurosci. 5, 2060-2077.

MILLER, T. A. (1979). Nervous versus neurohormonal control ofinsect heartbeat. Am. Zool. 19, 77-86.

O'SHEA, M. AND SCHAFFER, M. (1986). Neuropeptide function: Theinvertebrate contribution. A. Rev. Neurosci. 8, 171-198.

PICKERING, B. T., WAKERLEY, J. B. AND SUMMERLEE, A. J. S.(1987). Neurosecretion: Cellular Aspects of the Production andRelease of Neuropeptides, Plenum Press: New York, 270 pp.

RAABE, M. (1982). Insect Neurohormones. pp. 121-140. PlenumPress: New York.

ROWELL, C. H. F. (1976). The cells of the insect neurosecretorysystem: constancy, variability and the concept of the unique,identifiable neuron. Adv. Insect Physiol. 12, 63-124.

SCHALLER, H. C. (1979). Neuropeptides in hydra. Trends Neurosci.3, 120-122.

TAGHERT, P. H. AND TRUMAN, J. W. (1982). Identification of thebursicon-containing neurones in abdominal ganglia of thetobacco hornworm, Manduca sexta. J. exp. Biol. 98, 385-401.

TAGHERT, P. H., TUBUTZ, N. J., TRUMAN, J. W. AND GOODMAN, C.

S. (1983). Monoclonal antibodies to insect peptidergic neurons.Soc. Neurosci. Abstr. 9, 314.

TAGHERT, P. H., TUBUTZ, N. J., TRUMAN, J. W. AND GOODMAN, C.

S. (1984). Monoclonal antibodies to insect peptidergic neurons.Soc. Neurosci. Abstr. 10, 152.

TAYLOR, H. M. AND TRUMAN, J. W. (1974). Metamorphosis of theabdominal ganglia of the tobacco hornworm, Manduca sexta.J. comp. Physiol. 90, 367-388.

TRIMMER, B., KOBIERSKI, L. AND KRAVTTZ, E. (1987). Purification

and characterization of FMRFamide-like immunoreactive

Peptidergic modulation of the insect embryonic gut 71

substances from the lobster nervous system: isolation andsequence analysis of two closely related peptides. J. comp.Neurol. 266, 16-26.

TRUMAN, J. W. (1985). Hormonal control of ecdysis. InComprehensive Insect Physiology, Biochemistry, andPharmacology, (ed. Kerkut, G.A., and Gilbert, L.I.), pp.413-440. Pergamon Press: NY.

TRUMAN, J. W., TAGHERT, P. H., COPENHAVER, P. F., TUBLITZ, N.

J. AND SCHWARTZ, L. M. (1981). Eclosion hormone may controlall ecdyses in insects. Nature 791, 70-71.

TUBUTZ, N. J. (1988). Insect cardioactive neuropeptides:peptidergic modulation of the intrinsic myogenic rhythm of aninsect heart is mediated by inositol 1, 4, 5 trisphosphate. J.Neurosci. 8, 4394-4399.

TuBLrrz, N. J. (1989). Insect cardioactive peptides: Neurohormonalregulation of cardiac activity by two cardioacceleratory peptides(CAPs) during flight in the tobacco hawkmoth, Manduca sexta. J.exp. Biol. 142, 31-48.

TUBUTZ, N. J., COPENHAVER, P. F., TAGHERT, P. H. AND TRUMAN,

J. W. (1986). Peptidergic regulation of behavior: An identifiedneuron approach. Trends Neurosci. 9, 359-363.

TUBUTZ, N. J. AND EVANS, P. D. (1986). Insect cardioactivepeptides: cardioacceleratory peptide (CAP) activity is blocked invivo and in vitro with a monoclonal antibody. J. Neurosci. 6,2451-2455.

TUBUTZ, N. J. AND SYLWESTER, A. W. (1988). Plasticity oftransmitter expression: individually identified peptidergic neurons

alter their transmitter phenotype during post-embryonicdevelopment in the moth. Soc. Neurosci. Abstr. 14, 28.

TUBUTZ, N. J. AND SYLWESTER, A. W. (1989). Post-embryonicalteration of transmitter phenotype in individually identifiedpeptidergic neurons. J. Neurosci. (in press).

TUBUTZ, N. J. AND TRUMAN, J. W. (1985a). Insect cardioactivepeptides: Distribution and molecular characteristics of twocardioacceleratory peptides in the tobacco hawkmoth, Manducasexta. J. exp. Biol. 114, 365-379.

TUBUTZ, N. J. AND TRUMAN, J. W. (19856). Insect cardioactivepeptides: Neurohormonal control of heart activity by twocardioacceleratory peptides in the tobacco hawkmoth, Manducasexta. J. exp. Biol. 114, 381-395.

TUBLITZ, N. J. AND TRUMAN, J. W. (1985c). Identification ofneurones containing cardioacceleratory peptides (CAPs) in theventral nerve cord of the tobacco hawkmoth, Manduca sexta.J. exp. Biol. 116, 395-410.

TUBLITZ, N. J. AND TRUMAN, J. W. (1985d). Intracellularstimulation of an identified neuron evokes cardioacceleratorypeptide release. Science 228, 1013-1015.

WEEKS, J. C. (1987). Time course of hormonal independence fordevelopmental events in neurons and other cell types duringinsect metamorphosis. Devi Biol. 124, 163-176.

(Accepted 29 September 1989)