Expression of nitric oxide synthase and soluble guanylyl cyclase in the developing olfactory system of Manduca sexta

Animals

Male Manduca sexta (Lepidoptera: Sphingidae) were reared on an artificial diet, as described previously (Sanes and Hildebrand, 1976; Prescott et al., 1977). Animals were staged according to external morphology, as described by Tolbert et al. (1983).

RNA isolation

Antennae and antennal lobes were dissected from male pupae. Total RNA was isolated by using Trizol reagent (Life Technologies, Gaithersburg, MD). Poly (A⁺) RNA was obtained from antennal total RNA by using Oligotex beads (Qiagen, Valencia, CA).

Northern blot analysis

Antennal poly (A⁺) RNA (2 μg/lane) and antennal lobe total RNA (10 μg/lane) from a range of pupal stages were separated on formaldehyde-1% agarose gels and blotted onto Zetaprobe membranes (BioRad, Hercules, CA). Blots were hybridized overnight at 42°C in hybridization buffer containing 50% formamide, 5× sodium-saline-phosphate ethylenediamine tetraacetic acid (EDTA) buffer (SSPE), 5× Denhardt's solution, 1% sodium dodecylsulfate (SDS), 10% dextran sulfate, 100 μg/ml sheared salmon sperm DNA, and 10⁶ cpm/ml ³²P-labeled, random-primed probe made from gel-purified cDNA clones by using a Decaprime labeling kit (Ambion).

NOS Western blot analysis

For each lane, one stage 5 brain was harvested and homogenized in homogenization buffer (50 mM Tris-HCl, pH 7.4; 0.1 mM EDTA; 0.1 mM ethylene glycol-bis (β-aminoethyl ether) N,N,N′,N′-tetraacetic acid EGTA; 12 mM β-mercaptoethanol; 1 μM leupeptin; 1 μM pepstatin A; 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride AEBSF; all from Sigma, St. Louis, MO). The samples were then separated by SDS-polyacrylamide gel electrophoresis and were transferred to Immobilon-P membrane (Millipore, Bedford, MA). The blots were probed with a polyclonal universal NOS antibody (uNOS; Oncogene Research Products, Cambridge, MA) made in rabbit against the peptide sequence KRYHEDIFG. This sequence is almost identical to the Manduca sexta NOS sequence, NRYHEDIFG. The primary antisera were used at a dilution of 1:1,000, and the secondary antibody (goat anti-rabbit AP; Jackson Laboratories, West Grove, PA) was used at a dilution of 1:10,000. The blot was developed by using an Opti-4CN substrate kit (Bio-Rad).

NOS immunocytochemistry

Animals were anesthetized on ice. Antennae for cross sections were fixed in situ in 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for 20 minutes; cut into ten-segment lengths, fixed for an additional 3 hours at room temperature, and embedded in 5% agarose.

Antennae for fillets were dissected into insect saline adjusted to 360 mOsm with mannitol, cut lengthwise with iridectomy scissors, pinned out, fixed overnight in 4% paraformaldehyde, then handled as for sections. Brains were dissected into insect saline and fixed overnight at 4°C in 4% paraformaldehyde. Both brains and antennae were washed in phosphate-buffered saline (PBS), embedded in 5% agarose, and sectioned on a Vibratome at 100 μm. Sections were soaked in PBS with 0.5% Triton X-100 (PBST) for varying periods of time, then blocked in PBST containing 5% normal goat serum at room temperature for 2 hours. Primary antiserum (uNOS; Oncogene Research Products) was added at a dilution of 1:300 (1:200 for antennae). After incubation overnight at 4°C, sections were washed, blocked, and incubated for 2 hours at room temperature with goat anti-rabbit antibody conjugated to Alexa 546 or 488 dyes (Molecular Probes, Eugene, OR) at 1:250. Sections that also were labeled with propidium iodide (Sigma) were treated first with 100 μg/ml RNAse (Sigma) for 5–15 minutes at room temperature, washed, then incubated in 25 μg/ml propidium iodide for 15 minutes at room temperature, then washed. Sections were mounted in PBS with 80% glycerol and viewed on a Nikon PCM 2000 laser-scanning confocal microscope (Nikon, Tokyo, Japan) on an E800 microscope using Simple 32 software (Compix Inc., Cranberry Township, PA). Controls in which no primary antibody was used gave no discernible image at the confocal microscope settings used for the experimental images.

GC immunocytochemistry

Animals were anesthetized on ice. Brains were dissected into insect saline and fixed and sectioned as described above. Sections were blocked in PBST containing 10% normal goat serum at room temperature for 2 hours. Primary antiserum (anti-MsGCα1; made in chick) was added at a dilution of 1:5,000. After incubation overnight at 23°C, sections were washed, blocked, and incubated for 2 hours at room temperature with goat anti-chick antibody conjugated to Alexa 546 dye (Molecular Probes) at 1:250. Sections were mounted and viewed as described above. Controls in which preimmune serum was used gave no discernible image at the confocal microscope settings used for the experimental images.

cGMP immunocytochemistry

Brains and antennae were dissected as described above then incubated for 60 minutes in insect saline at room temperature followed by incubation for 20 minutes in insect saline alone or in insect saline containing 1 mM sodium nitroprusside (SNP; Sigma) plus 10 mM 3-isobutyl-1-methylxannthine (IBMX; Sigma), IBMX alone, or SNP alone, then rinsed, and fixed in 4% paraformaldehyde for 2 hours at room temperature. After fixation, brains were washed in PBS, embedded in 5% agarose, soaked in PBST (0.5% Triton X-100) at 4°C overnight, and sectioned on a Vibratome at 200 μm. Sections were blocked in PBST containing 5% normal donkey serum at room temperature for 2 hours. Sheep polyclonal anti-cGMP antisera (H. Steinbusch, Maastricht, The Netherlands) was added at a dilution of 1:10,000, and sections were incubated overnight at 4°C. Sections were then washed, blocked, and incubated for 2 hours at room temperature with donkey anti-sheep antibody conjugated to indocarbocyanine dye (Jackson Laboratories, West Grove, PA) at a dilution of 1:250. Sections were then processed and viewed as described above. Controls in which no primary antibody was used gave no discernible image at the confocal microscope settings used for the experimental images.

NADPH-d staining

Brains were dissected in saline and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for 2 hours at 4°C. Frozen sections (20 μm) were cut and mounted onto subbed slides. After a rinse in 50 mM Tris-HCl, pH 7.5, sections were incubated in 50 mM Tris-HCl, pH 7.5, containing 0.1% Triton X-100, 0.1 mM nitro blue tetrazolium NBT, and 0.1 mM β-NADPH (all from Sigma) for 2 hours at room temperature.

Diaminofluorescein labeling

Stage 5 antennae (with attached brains) were dissected in insect saline and then filleted along part of their length to increase contact of the luminal surface with the medium. Tissue was transferred to insect saline containing 3 μl/ml of the NO indicator diaminofluorescein (DAF-2 DA; Calbiochem, San Diego, CA) and 3 μl/ml Pluronic F-127 (Molecular Probes) and incubated at room temperature in the dark for 4 hours. The tissue was incubated in insect saline for 1 hour, then fixed overnight in 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, at 4°C. The tissue was then viewed on the Nikon laser-scanning confocal microscope using appropriate filters.

In situ hybridization

In situ hybridization was performed as described in Nighorn et al. (1998). Briefly, brains were dissected and fixed in 4% paraformaldehyde in PBS overnight at 4°C. Frozen sections (20 μm) were then cut and incubated overnight at 42°C. After pretreatment, the slides were hybridized overnight at 42°C in 50% formamide, 10% dextran sulfate, 1× Denhardt's, 4 × SSPE, 500 μg/ml yeast tRNA, 250 μg/ml sonicated salmon sperm DNA, and ≈5 ng of digoxigenin-labeled riboprobe. Equal amounts of a sense riboprobe were used as a negative control. After washing, the probes were visualized by using alkaline phosphatase-labeled sheep antidigoxigenin antibody (Boehringer Mannheim), and the staining was developed by using a 5-Bromo-4-chloro-3-indoyl phosphate BCIP/NBT solution. Slides were then viewed with a Nikon E600 microscope with a Spot CCD camera (Diagnostic Instruments, Sterling Heights, MI).

Image processing

All of the confocal images and CCD camera images above were imported into Corel Photopaint software, in which false color was applied to the confocal images, and the contrast was adjusted for maximum clarity. The images were then put together into figures in Corel Draw, and annotations were added.

Manduca sexta olfactory system expresses both NOS and sGC during development

Northern blot analyses using total RNA isolated from antennal lobes at different stages of pupal development were performed to determine whether MsNOS and NO-sensitive sGC (MsGCα1 and MsGCβ1) were expressed in the developing antennal lobe. Pupal development is divided into 18 stages, each lasting approximately 1 day (Tolbert et al., 1983). The following stages were examined: stage 2, just prior to the major developmental events initiated by the ingrowth of ORN axons; stages 3 and 4, when ORN axons first enter the developing antennal lobe but just before protoglomeruli start to form; stage 6, after projection neuron dendrites have entered protoglomeruli and the glomeruli have begun to stabilize; stage 7, whenglomeruli are stabilized and waves of synaptogenesis are occurring; stage 12, when the organization of the antennal lobe is established; and stage 18, just prior to adult emergence (Oland and Tolbert, 1996). MsGCα1 mRNA remained constant, whereas MsGCβ1 mRNA increased somewhat throughout development (Fig. 1A). The MsNOS message was developmentally regulated very strongly: Amounts were high at developmental stages 2–4, declined sharply at stages 6 and 7, and were undetectable at stages 12 and 18. The peak of MsNOS expression was at stage 4, just prior to the time protoglomeruli form and when interactions between ORN axons and antennal lobe neuronal and glial processes are likely to be very important. The expression of MsNOS, MsGCα1, and MsGCβ1 also were examined in the developing antennae and were found to be present throughout development. A Northern blot showing their expression at stages 5 and 7 is shown in Figure 1B.

MsNOS is expressed in perineurial cells and ORN axons

Having established that the genes for Manduca NOS and sGC are expressed in the developing antennal lobe, we sought to identify the cells that were expressing NOS during development by using NOS immunocytochemistry, NADPH-d histochemistry, and in situ hybridization. Pupal stage 5 was chosen as a starting point, because this was close to the peak of the developmental expression of NOS in both the antennae and the antennal lobes. It also is a stage in which the interactions between the olfactory receptor axons and the antennal lobe glial cells and neurons are important for the formation of glomeruli.

The commercially available NOS antipeptide antiserum (uNOS antibody) was checked for specificity on a Western blot of stage 5 pupal brain homogenates and produced a strong band of approximately 138 kD with weak staining of two higher molecular weight bands around 225 kD (Fig. 2). The measured size of 138 kD is a good match to the predicted size of 137 kD for MsNOS (Nighorn et al., 1988) and suggests that this antiserum can be used to specifically localize MsNOS.

NOS immunocytochemistry revealed strong NOS expression in the perineurium of stage 5 antennal lobes (Fig. 3A). This staining was not restricted to the antennal lobe area but was present in perineurium covering all regions of the brain. There also was strong NOS immunoreactivity in a subset of ORN axons that could be seen traveling together in the antennal nerve. Two ordinary protoglomeruli near the entrance region of the antennal nerve also were labeled strongly, whereas several adjacent protoglomeruli that were just beginning to form showed weak staining (Fig. 3A). Protoglomeruli near the nerve are known to be the first to develop, with those farther away from the nerve forming later in relation to their distance (Malun et al., 1994). Although it is not seen in the image in Figure 3, similar sections have revealed that protoglomeruli of the macroglomerular complex (Hildebrand, 1995) also are NOS immunoreactive at this stage, but the staining is very diffuse, perhaps reflecting a more diffuse arborization of the axon terminals. In addition to the strong staining of antennal nerve axon terminals, there was very faint staining of neuronal nuclei surrounding the antennal lobe neuropil. This staining of the nuclei was relatively constant throughout antennal lobe development, even though the levels of mRNA message for NOS was not. Thus, we suspect that this staining reflects background, especially in view of its nuclear location. The problem of faint background staining with this antibody has been described previously in lobster by Scholz et al. (1998). It is possibly due to the fact that the epitope recognized by the antibody corresponds to the NADPH-binding region of NOS and, thus, may cross react with other enzymes that use NADPH.

Staining of antennal lobes at later stages of development revealed that, by stage 6, all protoglomeruli were NOS immunoreactive and that, by the end of stage 7, the perineurium had lost its NOS immunoreactivity (not shown). At stage 18, just prior to adult eclosion, uNOS staining revealed that NOS was present in the axon terminal portion of all of the glomeruli (Fig. 3C), where it presumably plays a role in modulation of odor detection in the adult (Nighorn et al, 1998). At this stage, the macroglomerular complex can be seen to be NOS immunoreactive, although the intensity increases from the cumulus, to toroid 1, to toroid 2 (Homberg et al., 1995), the last showing an intensity similar to that of the ordinary glomeruli.

NADPH-d staining was used as another way to localize NOS in the antennal lobe. NADPH-d, like NOS immunoreactivity, labeled the perineurial cells strongly at stage 5. However, it also stained some antennal lobe neuronal cell bodies and produced faint staining around the edges of the developing antennal lobe neuropil (Fig. 3B). There was no detectable staining of the antennal nerve. Thus, whereas both NOS immunocytochemistry and NADPH-d staining methods labeled the perineurial cells, only the NADPH-d method labeled antennal lobe neurons, whereas only the NOS antibody labeled the axons in antennal nerve. The difference between the two staining methods was even more striking at stage 18 (Fig. 3C,D). The two methods were most similar in the perineurial sheath, which was completely unlabeled using NOS immunocytochemistry and only weakly labeled with NADPH-d staining. Unlike NOS immunocytochemistry, however, NADPH-d failed to label ORN axons either in the antennal nerve or in the apical regions of the glomeruli. Instead, we saw strong staining of a few antennal lobe neuronal cell bodies and faint staining of the bases of the glomeruli.

In situ hybridization was used to investigate the localization of MsNOS mRNA. At stage 5, the NOS in situ probe labeled the perineurial cells strongly (Fig. 3E,F), whereas no other cell bodies in the antennal lobe were strongly positive for NOS mRNA. There were one or two cells that were very weakly positive, but these were difficult to discern from background. At stage 18, no NOS mRNA expression can be detected in antennal-lobe neurons (Nighorn et al., 1998). Overall, the pattern was much more consistent with the NOS immunocytochemical staining pattern than the NADPH-d staining pattern.

To summarize, all of the methods used, including Northern blot analyses, showed that NOS is expressed in a transient fashion in cells of the developing antennal lobes. NOS immunocytochemistry and in situ hybridization suggest that this localization is limited to the perineurium, whereas NADPH-d staining suggests strong labeling of antennal lobe neuronal cell bodies as well. In the antennal nerve, NOS immunocytochemistry showed strong labeling of the ORC axons, whereas NADPH-d staining was absent.

Because NADPH-d measures enzymatic activity and NOS immunocytochemistry measures only the presence of the protein, we used the NO-sensitive dye DAF-2 (Kojima et al., 1998) in dissected stage 5 antennae-plus-brain primary olfactory system preparations as another measure of NOS activity. In the antennal lobe, penetration problems prevented the use of this dye. In the antennae, however, the dye penetrated the tissue well and responded to the presence of NO. This method clearly showed the presence of NO in ORNs in the antennae (Fig. 4). We detected high levels of NO in the ORN axons and low levels of NO in the cell bodies and dendrites. This result, which suggests that NOS protein is present and can generate NO in the ORN axons, supports the immunocytochemical localization of NOS.

MsGCα1 and MsGCβ1 are expressed in antennal lobe neurons

Because NOS is expressed transiently in the perineurium and continuously at afferent axon terminals, identification of target cells should help to identify NO-mediated developmental processes. Although NO can have many different targets, one of the best characterized is sGC, which mediates the formation of cGMP. Upon binding heme, the activity of the sGC heterodimer can be stimulated strongly by NO, resulting in an increase in cGMP. Therefore, the ability of NO to stimulate an increase in cGMP levels in the developing antennal lobe was examined by using cGMP immunocytochemistry.

At stage 5, cGMP was found to be elevated in a subset of lateral cell body cluster neurons; the number of stained cells was increased only slightly by incubation in SNP, an NO donor (not shown). Attempts to potentiate the response by incubating the brain in IBMX, which is a phosphodiesterase inhibitor, produced many stained cells in both the lateral and medial cell body packets, with no visible staining of antennal lobe glia (Fig. 5A). Incubation with SNP, however, did not increase these already high levels of cGMP (not shown). These results suggest either that there are other, NO-insensitive GCs mediating this effect or that there is endogenous stimulation with NO coupled with active phosphodiesterase activity. An NO-mediated increase in cGMP in developing antennal lobe neurons in response to dissection or stimulation of the antennae has been reported recently in some lateral cell body cluster neurons (Schachtner et al., 1998, Schachtner et al., 1999). Although labeling of the lateral cell body neurons was extensive and consistent at stage 5, we found extensive labeling of the medial cell body neurons ≈50% of the time. This result most likely reflects small variations in the extent of development, because stages earlier than midstage 5 did not display medial cell body neuron staining (not shown). At stage 7, the cGMP staining pattern was very similar, with strong, consistent staining of both the medial cell body cluster and the lateral cell body cluster. At this later developmental stage, processes from some of these cells could be seen entering developing glomeruli (Fig. 5B). Again, however, this staining could not be potentiated significantly by incubation in SNP.

Because of the possibility that high levels of endogenous NO stimulation of sGC may obscure the interpretation of the exogenous NO experiments, and because NOS is present in both afferent axon terminals and perineurial cells, the results described above do not provide sufficient information to localize potentially important NO-sGC interactions. In addition, in the presence of IBMX, cGMP in stimulated neurons accumulates and diffuses throughout the length of the neuron, obscuring the location of the initial interaction of NO with sGC. To help localize these interactions, we generated a polyclonal antiserum directed against a unique region of MsGCα1. By using this antiserum at midstage 5, clear staining of antennal lobe neuron dendrites is evident (Fig. 5C), although no cell bodies or output tracts are stained. Because lateral cell body neurons do not extend dendrites into protoglomeruli at this stage, the staining of protoglomeruli indicates that medial cell body projection neurons are MsGCα1-positive. By stage 7, staining clearly shows individual glomeruli, including the macroglomerular complex (Fig. 5D). This pattern is present as early as stage 3 (Fig. 5E), although, at the earlier stages, cell body staining also is apparent. Comparison with earlier neuroanatomic studies (Oland et al., 1990; Malun et al., 1994) suggests that the MsGCα1-positive dendrites belong to cells from both groups of neurons (Lynne Oland, personal communication). The localization of MsGCα1 to the dendrites of antennal lobe neurons positions it favorably for interactions with NO produced by the sensory axon terminals.

The presence of NO-sensitive sGC in the developing antennal lobe also was investigated by using in situ hybridization to localize both the α and β subunits. Subsets of stage 5 antennal lobe neurons from both the lateral and medial cell body packets were found to be positive for MsGCα1 message (Fig. 5F). The pattern was similar for MsGCβ1 (not shown), although fewer cells were detectably positive for the MsGCβ1 message. This localization pattern supports the idea that NO can interact with sGC expressed in both populations of antennal lobe neurons.

NOS and sGC also are expressed in the antennae

The expression of NOS in ORN axons led us to investigate the localization of NOS and sGC in the antenna itself by using NOS immunocytochemistry and NO-stimulated cGMP immunocytochemistry. At stage 5, NOS immunoreactivity was present in a subset of ORN axons all along the length of the antennal nerve (Fig. 6A,B) and in rootlets from each antennal segment as they joined the nerve. By comparing the fillet (Fig. 6B) with a cross section (Fig. 6C), it becomes clear that the subset of antennal nerve axons that are NOS-positive are those that have been added by the joining of nerve rootlets. In addition, a double label of NOS and propidium iodide (to label glial cell nuclei) shows that only the subset of axons in contact with glial cells is NOS-immunoreactive (Fig. 6D). Thus, there is a correlation between NOS-positive axons and migrating glial cells. Glial cells have been shown recently to migrate down antennal nerve rootlets to the nerves during pupal stages 3 and 4, although the cause of their migration is not known (Rössler et al., 1999). Finally, large numbers of NOS-positive cells also can be seen in the nonolfactory epithelium (scalar side of the antenna) at stage 5 (Fig. 6C). The significance of this staining is unknown.

By stage 7, a cross section of the antennal nerve stained with propidium iodide and the uNOS antibody reveals that both glial cells and NOS immunoreactivity invest the full width of the antennal nerve (Fig. 6E). At this stage, there also is strong staining of cell bodies throughout the olfactory sensillar portion of the epithelium (Fig. 6F). This portion of the epithelium contains the cell bodies of the ORNs and related support cells. The staining of the cell bodies in the nonolfactory epithelium is now lost (Fig. 6E). By stage 18, none of the cell bodies in the epithelium is NOS immunoreactive (not shown), although the antennal nerve remains strongly NOS positive (Fig. 3C). Thus, NOS is present in ORN axons throughout development, but it is present in ORN cell bodies only transiently, indicating that there is developmental regulation of NOS localization within these cells, as has been demonstrated in vertebrates (Matsumoto et al., 1993a).

The expression of NOS by different cell types in the antennae raises the possibility that NO affects antennal cells through stimulation of sGC. This possibility was investigated by using NO-stimulated cGMP immunocytochemistry. In stage 5 antennae, a small number of cells, including six or seven large neurons in each antennal segment, stained intensely in the presence of SNP and IBMX (Fig. 7A,D,F). These neurons are in a location similar to, and are likely to be, the receptor neurons of the mechanosensory sensilla chaetica (Lee and Strausfeld, 1990). These neurons do not stain for cGMP when the antennae are treated with IBMX only (Fig. 7G) and, thus, appear to contain NO-sensitive sGC. The other cGMP-positive cells stain equally well in the presence or absence of SNP (Fig. 7D,G). These cells, which include the ORNs, either must contain other types of NO-insensitive GCs or must have been stimulated maximally by endogenous NO. Also identified was a new class of neurons with cell bodies that reside in the lumen of the antenna and with processes that extend within the antennal nerve (Fig 7A–C,E). These cells are likely to be the same as the cGMP-positive luminal cells seen in the adult (Monika Stengl, personal communication). It is noteworthy that, when we examined cGMP staining at stages 6-7 in the presence of IBMX and SNP, we found strong labeling of the ORNs in addition to the cells that were labeled at stage 5 (Fig. 7H,I). This staining, unlike the faint staining seen at stage 5, is dependent on SNP addition. It is intriguing that MsNOS also is expressed highly in ORN cell bodies at this stage (Fig. 6F). The relation between the MsNOS expression and the increase in cGMP levels will have to be investigated further with the use of NOS and sGC inhibitors.

NADPH-d staining is a poor measure of NOS expression

Much of the work investigating NOS expression in insects has focused on adults and has relied on NADPH-d staining as an indirect indicator of NOS activity (Müller, 1994; Müller and Bicker, 1994; Elphick et al., 1995; Müller and Hildebrandt, 1995; Bicker et al., 1996). Although attempts were made to show that biochemical NOS activity was present in tissues that exhibited NADPH-d, the two activities could not be colocalized at the cellular level in most of these studies. Reliance on this technique seemed justified, however, given the frequent convergence of NADPH-d staining and NOS protein expression in vertebrates. Here, too, there has long been controversy (Matsumoto et al., 1993b; Buwalda et al., 1995; Spessert and Claassen, 1998). Recently, Ott and Burrows (1999) addressed this problem in insect histochemistry, arguing that NADPH-d activity actually can arise from overfixation of tissue.

The current results show clearly that NADPH-d staining is a poor measure of NOS expression in the developing Manduca olfactory system. Because NADPH-d is believed to measure the ability of an aldehyde-fixed enzyme to reduce NADPH, NOS should account for at least a portion of NADPH-d activity. In this system, however, NOS staining and NADPH-d staining appear to be independent. Although both methods stained the perineurial sheath cells, the antennal lobe neurons were positive for NADPH-d but negative for NOS immunoreactivity. This may be explained by the idea that NOS is only a subset of NADPH-d activity. The antennal nerve, however, stained strongly with the NOS antibody yet was negative for NADPH-d activity. Moreover, the presence of NOS in the antennal nerve was confirmed by the DAF-2 staining. In situ hybridization results at stage 5, combined with the Northern blot analyses for antennal lobes throughout development, support the conclusion that NOS is expressed only in perineurial cells in the antennal lobe. Thus, we conclude that NADPH-d staining does not reflect accurately the distribution of NOS in this system and that NADPH-d should be used with great caution as a quick measure of NOS in insects.

Developmental role of NOS

The current study demonstrates that NOS is expressed in ORN axons and in the perineurial cells. The NOS in the perineurium is regulated developmentally and is present in these cells around the entire brain. The timing of its decline at pupal stage 7 is intriguing, because that is a time at which levels of the steroid hormone 20-OH ecdysone peak (Weeks and Truman, 1986), and it suggests that this phenomenon may be regulated hormonally. The developmental significance of this expression, however, is unknown. Although perineurial NO production may act to maintain sGC-positive antennal lobe neurons in a developmentally necessary state until stage 7, the localization of the MsGCα1 protein in neuron dendrites suggests that they respond primarily to NO released from axon terminals. It is possible that the presence of migrating glia in the perineurium (Rössler et al., 1999) is an important factor, as it appears to be in the antennal nerve (Fig. 6D,E).

The NOS that is expressed in the axon terminals of the ORNs is expressed in an ideal location to mediate developmental events in the antennal lobe. At the beginning of stage 5, ORN axon terminals form protoglomeruli, which are entered almost immediately by antennal lobe projection neuron processes (Oland et al., 1990; Malun et al., 1994). Glial cells then proliferate and migrate to surround the protoglomeruli, after which antennal lobe interneurons extend processes into the protoglomeruli at stage 6. Stage 6 sees the stabilization of the glomeruli and the beginning of a wave of synaptogenesis (Oland et al., 1990). NOS is present in ORN axon terminals at all of these stages and, thus, may be involved in any or all of these events. Initiation and cessation of antennal lobe neuron dendrite outgrowth, stimulation of glial cell migration, and stimulation of synaptogenesis are all processes with counterparts in vertebrates that have been shown to be modulated by NO (Hess et al., 1993; Duman et al, 1993; Meffert et al., 1994; Meulemans et al., 1995; Rentería and Constantine-Paton, 1996; Hindley et al., 1997).

Developmental role of NO-stimulated sGC

One likely target of NO in axon terminals is sGC. The current study demonstrated by using multiple techniques that MsGC is present in neurons of both the lateral and medial cell bodies during development. Immunocytochemistry demonstrates that MsGCα1 is present in the processes of both lateral and medial cell body neurons as early as stage 3 and clearly is present at stages 5 and 7. However, the presence of MsGCα1 may be necessary but insufficient to confer NO responsiveness to a particular neuron. At stage 5, for example, we find that medial cell body neurons are cGMP positive in the presence of exogenous NO plus IBMX in only about 50% of the preparations, whereas neurons from both cell body clusters appear to be positive for MsGCα1. This suggests that the sensitivity to NO may be regulated developmentally, because these cells are cGMP positive in the presence of NO/IBMX at all of the later stages examined. A developmentally regulated sensitivity to stimulation by NO has been shown in locust motor neurons (Truman et al., 1996). More work will be needed to understand the mechanisms by which the sensitivity of a neuron to NO may be regulated in this system. It is important to note that our results differ from those reported by Schachtner et al. (1998, Schachtner et al. 1999), who found only lateral cell body neurons stained after phosphodiesterase inhibitor treatment. Because the Schachtner et al. studies relied solely on cGMP immunocytochemistry, their results may reflect a threshold effect of the sensitivity of the antibody, they may reflect the influence of as yet unidentified environmental factors, or they simply may be the result of a difference in the effectiveness of the phosphodiesterase inhibitor treatment used in the two different sets of experiments.

In our experiments, the inability of SNP to potentiate the cGMP staining in the presence of IBMX implies either that NO-insensitive GCs are mediating the increased cGMP levels in these cells or that endogenous NO production already stimulates sGC to maximal detectable levels. The current results cannot rule out the possible contribution of active, NO-insensitive GCs. Six NO-insensitive GCs are expressed in the nervous system of Manduca sexta (A.N., unpublished data), although their expression and activity during development have not yet been characterized.

Although the relative contributions of sGC and NO-insensitive GCs are unknown, the current results show clearly that sGC is present in antennal lobe neurons and is in a location that is optimal for stimulation by NO released by ORC axons. Sensory axons and antennal lobe uniglomerular projection neurons are known to have little direct synaptic contact (Christensen et al., 1993; Malun et al., 1994); the NO/sGC pathway may permit developmentally important, direct signaling between these cell types to stabilize neurite arbors while awaiting synaptogenesis, which commences with the arrival of local interneuron dendrites (Oland et al., 1990). The NO/sGC pathway also may be important for the initiation of synaptogenesis in the antennal lobe, as has been suggested for motor neurons in the embryonic development of Locusta migratoria (Truman et al., 1996; Ball and Truman, 1998).

NO/sGC signaling also may have a role in the development of the antenna. At stage 5, and at every stage examined including the adult (M. Stengl, personal communication), an NO-sensitive GC is present in the mechanosensory cell bodies and axons (Fig. 7A,D,F,H,I). NO may be produced by the ORN axons and interact with the sGC in these cells. The increase in cGMP staining in ORN cell bodies at stage 7 is intriguing, because synaptogenesis occurs in the antennal lobe at stage 7, the same stage at which ecdysteroid levels are at their highest. This may represent a hormonally regulated signal to alter gene expression to allow for increased synapse formation or to begin the process of maturation of the peripheral components of the signal-transduction cascade in the ORNs. The NO-dependence of this effect and its developmental consequences need to be examined further.

The upstream and downstream events surrounding a possible NOS/sGC signaling mechanism have yet to be determined in the antennae and antennal lobe of Manduca. Previous work has shown that MsNOS is likely to be sensitive to activation by Ca²⁺/calmodulin (Nighorn et al., 1998). Visualization of the Ca²⁺ currents present in developing ORNs may yield important insights into the regulation of the production of NO in this system. Cyclic nucleotide-gated channels are possible downstream targets of cGMP. Although such channels have not yet been identified in Manduca sexta, they are known to be present in the olfactory bulb of vertebrates (Kingston et al., 1999) and in the antennae of Drosophila (Baumann et al., 1994), and they may be well placed to modulate the electrical properties of developing neurons in response to changes in cGMP levels. The effect of increased cGMP also may be mediated by the activation of the cGMP-dependent protein kinase. Finally, in the antennal nerve and its rootlets, the correlation between the presence of migrating glial cells and NOS immunoreactivity of ORN axons raises interesting questions of cause and effect that will be pursued in future work.

Developmental role of non-sGC-mediated effects of NO

The NO-sensitive sGC almost certainly is not the only target of the NO produced in the developing olfactory system, especially given the relatively restricted expression of sGC. NO is known to activate directly cyclic nucleotide-gated channels (Broillet and Firestein, 1996) and thereby may cause the release of neurotransmitters from adjacent axon terminals, mimicking an odorant-induced activity that would not occur naturally until approximately stage 16 (Schweitzer et al., 1976). Furthermore, if it is acting on the channels of antennal lobe neurons and glia, NO could cause activity-dependent changes by promoting Ca²⁺ influx (Kossel et al., 1997). NO also has been shown to affect neurite outgrowth, cell migration, synaptogenesis, and synaptic activity by altering acylation, ADP-ribosylation, and nitrosylation of critical proteins (Hess et al., 1993, Hess et al., 1999; Meffert et al., 1994; Schuman et al., 1994; Stamler et al, 1997).

The presence of NOS clearly is not equivalent to the generation of NO, because a number of conditions, particularly high intracellular calcium levels, must be met to activate this enzyme in Manduca (Nighorn et al., 1998). Future work will focus on determining when and where NO actually is generated during development as well as the signaling pathways by which it achieves its actions on a given cell type. This will include investigating enzymes downstream of cGMP, such as protein kinase G and cyclic nucleotide gated channels, as well as the sGC-independent pathways mentioned above. Preliminary experiments with the NOS inhibitor L-NAME during metamorphosis have revealed pronounced effects on glial migration and neurite outgrowth in the antennal lobe (Gibson et al., 1999).

In summary, MsNOS has been shown to be present in or near the antennal lobe at times and at places that would allow it to influence antennal lobe development. Specifically, MsNOS is present at the terminals of ingrowing ORN axons as they innervate the developing antennal lobes, and it also is expressed highly by perineurial cells until stage 7. One possible target of NO generated in the antennal lobe is the NO-sensitive MsGC expressed in the dendrites of both medial and lateral cell body cluster neurons.