A diversity of neuron types express sNPF
We have investigated the widespread neuronal expression of a Drosophila neuropeptide gene snpf and its peptide products, sNPFs, in the larval and adult nervous system of the fruitfly. This mapping was made by in situ hybridization, a snpf-Gal4 line and immunocytochemistry with antiserum specific for the snpf-derived precursor protein. A large population of diverse neurons types was revealed with the different sNPF/snpf markers: a few neurosecretory cells, numerous interneurons of many kinds as well as olfactory receptor neurons (ORNs) of the antennae. This wide expression of sNPFs in diverse neuron types, many of which may colocalize different neurotransmitters, suggests a diversity of functions beyond that of any insect neuropeptide studied so far. Especially the finding of numerous small neurons such as Kenyon cells of the mushroom bodies and olfactory receptor neurons (ORNs) of the antennae and small neurons of the optic lobe suggest that sNPFs play important roles in local cotransmission and/or neuromodulation.
Since the intensity of immunolabeling with the sNPFp antiserum was variable, especially for the smaller neuronal cell bodies, it was hard to provide definite numbers of smaller sNPF expressing neurons, in spite of analyzing many specimens. Possibly part of this variability in immunolabeling reflects bona fide fluctuations in sNPF levels related to activity levels of neurons. Another drawback with the immunolabeling was that it was mostly punctate within neuronal processes and thus did not allow resolution of neuronal morphology beyond location of cell bodies, major tracts and axon terminations. Therefore, the majority of the sNPF expressing neuron types remain anonymous cell bodies that cannot be connected to their release sites. Apart from ORNs, a prominent exception to the anonymity is seen for sNPF expression in the numerous Kenyon cells that send sNPF-IR processes into the α, β and γ lobes, but not the α' and β' lobes and αcβc portions of these lobes . Two other sets of neurons that could be revealed in detail are the unique pair of DP neurons in the first abdominal neuromere (A1) in larvae that co-express ILP7 and the lateral protocerebral cells (LNCs) coexpressing corazonin.
One aim here was to obtain data that could reveal functional roles of sNPF expressing neurons. Does sNPF(s) have distinct functions or is the function dependent on the context of the neurons releasing the peptide(s). Generally neuropeptides are diverse and multifunctional signal substances [3, 14, 47–50]. In insects more than 30 genes encode precursors of neuropeptides that are known to regulate many aspects of development, growth, reproduction, metabolism, homeostasis and behavior [3, 14, 49–54]. Most of these regulatory roles are played by neuropeptides released as hormones into the circulation and some may act after episodic bulk release within the CNS [15, 55]. In addition, neuropeptides in interneurons are likely to act as local cotransmitters or neuromodulators in central neural circuits [16, 48]. Peptidergic neurons thus are of several different major types [16, 15]: (1) large neurosecretory cells with peripheral release sites that produce hormonal peptides, (2) a variety of circuit interneurons, some of which are relatively small, produce peptidergic cotransmitters (or local neuromodulators), (3) a separate type of peptidergic interneurons of intermediate to large size are likely to be responsible for episodic bulk transmission within the CNS and (4) subsets of motoneurons are known to express the peptide proctolin as cotransmitters [1, 17, 18]. To these types of peptide expressing neuron types we can now add sNPF expressing chemosensory neurons (ORNs) of the Drosophila antennae and maxillary palps.
A recent paper  addressed the role of the transcription factor DIMM in determining functional phenotypes of peptidergic neurons. Expression of DIMM was seen only in a portion of the known peptidergic neurons of Drosophila, namely a set referred to as Large cells that Episodically release Amidated Peptides (LEAP cells) . These LEAP neurons express both DIMM and PHM, an enzyme required for alpha-amidation of neuropeptides , and have large cell bodies and/or extensive arborizations and axonal processes. Many peptide-expressing neurons were found not to co-express DIMM (non-LEAP cells)  and we indicate that the sNPF producing neurons in the larva are of this kind. Using the c929-GAL4 as a marker for dimm expression  and antiserum to the sNPF precursor we found no neurons displaying coexpression in the larval CNS. We hade expected a set of 3 pairs of LNCs displaying sNPF and corazonin immunoreactivity to co-express DIMM, but this was not the case. In the paper by Park et al.  a set of five pairs of neurons were shown to co-express DIMM and sNPF (antiserum to sNPF-2) in the larval brain. Four of these have cell bodies medially and colocalize dromyosuppressin (DMS) an FLRFamide-like peptide and another set of six are located among the LNCs and coexpress the neuropeptide corazonin. Since these authors used an antiserum to the full peptide sNPF-2 from Lee et al.  it is likely that the DMS expressing neurons cross react with the sNPF-2 antiserum (due to the shared RFamide). The corazonin-containing LNCs on the other hand may indeed colocalize sNPF and DIMM. It could be that we missed this colocalization in our preps while the c929-driven GFP was too weak in the cells of interest. As mentioned, we did reveal sNPFp and corazonin colocalization in three pairs of LNCs in the position described for the DIMM coexpressing ones  and if these neurons are true neurosecretory cells they would qualify as LEAP cells (however, see discussion below on relation to AKH cells).
The finding of sNPF exclusively (maybe with exception above) in small neurons that do not express DIMM suggest that the sNPF expressing neurons might be signaling locally and primarily in capacity of cotransmitter or local neuromodulator. Another neuropeptide gene, dtk and its DTK products, are also expressed mainly in small interneurons [12, 56] and Park et al.  found that in the larva only one pair of dtk expressing neurons co-expressed DIMM. This pair of neurons also expresses allatostatin B (myoinhibitory peptide; MIP) and have extensive axonal projections all along the ventral nerve cord. Thus it is likely that both sNPF and DTKs are primarily cotransmitters and/or local neuromodulators.
Colocalization of sNPF and other neurotransmitters
To seek support for a role of sNPFs in cotransmission we screened for colocalized neurotransmitters in the large population of sNPF immunoreactive neurons. In the mammalian nervous system neuropeptides colocalize extensively with classical neurotransmitters in interneurons [47, 48, 57], but in insects systematic screens of such colocalizations have not been performed (see ). Here, for simplicity, we used Gal4 expression for marking possible neurotransmitter phenotypes. This choice was partly forced by the fact the available antisera to small molecule neurotransmitters or their biosynthetic enzymes require fixation protocols incompatible with sNPF detection, or do not label neuronal cell bodies and thus precludes definite localization to the same neuron. Immunolabeling with sNPFp was made on Gal4-lines likely to reveal neurons expressing GABA, glutamate, acetylcholine, dopamine and octopamine/tyrosine. In this screen we did detect some patterns of colocalization between sNPF and GABA, glutamate and acetylcholine, but the majority on the sNPFp-IR interneurons did not colocalize the markers tested so far. The main findings on colocalized markers are shown in Table 1.
The antennal ORNs are likely to be cholinergic and can thus be displayed by a Cha-Gal4 [33, 34]. We found that all sNPF expressing ORNs in the antennae indeed coexpressed Cha-Gal4-driven GFP. The presence of an identified neuropeptide in arthropod chemosensory neurons is a novel finding, although antisera to FMRFamide-like peptides have been reported to label sensory axons in blowflies, locusts and lobsters [59–61]. Especially intriguing is the finding that only a subpopulation of the ORNs in the antenna and their axon terminations in the glomeruli in the antennal lobe express sNPF. Thus, most, if not all, ORNs are cholinergic, but only a subpopulation employ a putative peptidergic cotransmitter. It will be of interest to identify the complete set of glomeruli receiving sNPF-IR axons of ORNs to determine for which odors additional peptidergic signaling is utilized.
We detected a number of further Cha-Gal4 expressing neurons that colocalized sNPF in the larval and adult brain, but very few in the ventral nerve cord. Both in the brain and the ganglia these were primarily small interneurons that could not be individually identified, except for a pair of dorsal median neurons (DP) in A1 of the ventral nerve cord. When analyzing sNPF in relation to Cha expression in neuronal processes in brain neuropils we could not detect colocalization of markers in central body neuropils, mushroom bodies, optic lobe (with one exception in the lobula; Fig. 10H), whereas in the antennal lobes the ORN terminations coexpressed the two. Thus cholinergic interneurons that express sNPF appear to arborize primarily in non-glomerular neuropils.
Fewer sNPF expressing neurons displayed markers for glutamate (vGluT-Gal4) and GABA (Gad1-Gal4). These neurons were in both cases located in the brain and no colocalization was detected in the larval ventral ganglia. Again the neurons were not possible to identify as individuals and we did not see any specific patterns of colocalization in processes in major neuropils. We detected no colocalization between sNPF and markers for dopamine and octopamine/tyramine. However, we did not investigate the possible colocalization of sNPF and the biogenic amines serotonin and histamine, known to be expressed in distinct sets of neurons in the larval and adult CNS of Drosophila [62–65].
In summary, except for the ORNs of the antennae and abdominal DP neurons, it is not clear what specific types of interneurons that coexpress the sNPF and the markers for small molecule neurotransmitters. By screening labeling in different neuropil regions of the adult brain we could exclude peptide/transmitter coexpression in the neurons of the central body, antennal lobes (except ORNs), optic lobe and intrinsic neurons (Kenyon cells) of mushroom bodies. The Kenyon cells are likely to express a small molecule transmitter in addition to sNPF. A screen for such a transmitter was made in a previous paper  but no marker employed so far provided a lead to the Kenyon cell transmitter. Several putative small molecule transmitters remain to be investigated in Drosophila and other insects (e. g. nucleotides/nucleosides, glycine, and aspartate). It cannot be excluded that the small proportion of sNPF neurons where we could see colocalized markers was caused by deficiencies in the Gal4 lines utilized: the gene expressions may be incomplete and thus not reveal the entire populations of neurons expressing the neurotransmitters intended.
Relation between sNPF and other neuropeptides, including ILPs
Since we were mainly interested in the colocalization of sNPF and small molecule transmitters we did not make an extensive screen for colocalization with other neuropeptides.
One exciting finding was that the two unique DP neurons of A1 of larvae co-express the relaxin-like ILP7, as well as Cha. These neurons were described first as ILP7 producing neurons with ascending axons terminating close to the dendrites of the ILP producing MNCs in the larval brain . Here we showed that the ascending ILP7 immunoreactive processes indeed colocalize sNPF and are part of a slightly more extensive supply of sNPF-IR processes impinging on Dilp2-Gal4 expressing MNCs. It was shown that sNPF overexpression causes increased production of ILPs in these MNCs and that the MNCs express the sNPF receptor . Thus, it is possible that sNPF together with ILP7 (and maybe acetylcholine) regulate production of other ILPs. It should be noted that we also could show that sNPF immunolabeled varicose axons (from the three pairs of LNCs) terminate close to AKH-producing cells in the corpora cardiaca of the ring gland. This might suggest that these sNPF producing cells could have a role in regulation of the AKH cells, but it cannot be excluded that sNPF is released as a circulating hormone. The same LNCs have additional axons terminating in the anterior aorta suggesting a hormonal release site.
Apart from the above findings we revealed one pair of neurons in the dorsal protocerebrum (in DLN1 group) co-expressing sNPF and NPF (long NPF) and noted an extensive superposition of neuronal processes containing he two peptides in the dorsal protocerebrum. Another set of about six sNPF-IR neurons in the dorso-median brain (pars intercerebralis) co-express Mai-179-Gal4. These particular neurons with axons in the median bundle and possible terminations in tritocerebrum were not described among the neurosecretory cells innervating the ring gland in Drosophila , but similar neurons were seen with an antiserum to Ion transport peptide (ITP) . Thus, it might be that these PI neurons coexpress ITP and sNPF.
How many peptides are produced by the snpf gene?
Mass spectrometry identified SPSLRLRFa, which corresponds to both sNPF-14–11 and sNPF-212–19, as the major product of the predicted sNPF-1 to 4. A small peak representing full length sNPF-1 was also detected. Thus, it appears that sNPF-2 is processed to sNPF-212–19, and sNPF-3 and 4 are not liberated from the precursor (see also [43–45]). An antiserum to sNPF-3 used in a previous study  labeled the same neurons as the one to the sNPF precursor used here. Possibly that antiserum recognizes the sNPF-3 and 4 sequences (with RLRWamide) within the precursor, or too little peptide is produced to be detected by mass spectrometry. The snpf genes of the mosquitoes Anopheles gambiae, Aedes aegypti, and Culex pipiens also predict RLRWamides like sNPF-3 and 4, but not the ones in the honey bee or the red flour beetle Tribolium [50, 53, 54]. We propose that the neurons we depicted with the sNPFp antiserum and ribonucleotide probe express at least one peptide derived from the snpf.
Functions of sNPF: diverse or not?
Judging from the abundance of sNPF distribution in diverse sets of neurons in the larval and adult CNS one would anticipate multiple functions of these peptides. An earlier study has addressed the general distribution and role of sNPFs and its receptor sNPFR1 in feeding and growth in Drosophila [20, 31]. These authors showed that sNPF signaling appears to be involved in regulation of food intake and growth, , and also in the regulation of Drosophila ILP expression via the sNPF receptor . The experiments were conducted by rather widespread over-expression and knock-down of sNPF products and did not address sNPF action in specific circuits. Thus, it is not clear at what level the sNPF regulation of ILP producing cells occurs. However, the sNPFR1 is expressed on Dilp2-gal4 expressing MNCs in brain and sNPF immunoreactive processes were detected close to these cells , as also shown here. Whatever the exact circuitry utilizing sNPF to regulate ILP production, the effect of sNPF on growth via ERK (extracellular signal-related kinases) signaling is likely to be only a part of a spectrum of sNPF functions in Drosophila. This function may be limited to a small subset of the sNPF-expressing neurons.
Peptides likely to be orthologs of Drosophila sNPF have been identified in other insects. In fact the first identification of an sNPF (or Aedes head peptides) was in the mosquito Aedes aegypti [67, 68] and subsequently related peptides have been identified in several other insect species [50, 53, 69–71, 54, 72]. Some actions of sNPF have been analyzed by peptide administration: sNPFs are myostimulatory on various visceral muscles [68, 71] induce host seeking behavior in female mosquitoes , they may be important for diapause in a beetle  and in locusts they stimulate ovarian growth and induce increases in vitellogenin levels in the circulation [71, 75]. This spectrum of actions, except the host seeking behavior, may represent hormonal functions of sNPF. Thus, probably a host of functions based on local actions of sNPF in circuits of the CNS remain to be demonstrated.
Only a single Drosophila sNPF receptor (NPFR76F; CG7395; sNPFR1) has been identified [76–78]. This receptor may couple to different signaling pathways, as already suggested from in vitro experiments [31, 76–78]. It has been postulated that the sNPFR1 is related to the mammalian neuropeptide Y type of receptors [76, 77] and the receptor for Drosophila long NPF [51, 79]. Long NPF is encoded on a separate gene (npf), identified in Drosophila and other insects [51, 52, 80]. This peptide is 36 amino acids in length with an ARVRFamide C-terminus and resembles neuropeptide Y of mammals. In spite of the unfortunate similarities in names of the two peptide genes and peptides, they are part of two functionally distinct signaling systems [20, 21, 31, 39].