Single-Cell Peptidomics of Drosophila melanogaster Neurons

Anal. Chem. 2007, 79, 3690-3694

Single-Cell Peptidomics of Drosophila melanogaster Neurons Identified by Gal4-Driven Fluorescence Susanne Neupert,*,† Helena A. D. Johard,‡ Dick R. Na 1 ssel,‡ and Reinhard Predel†

Institute of Zoology, Friedrich-Schiller-University Jena, Erbertstrasse 1, 07743 Jena, Germany, and Department of Zoology, Stockholm University, SE-10691 Stockholm, Sweden

Neuropeptides are widespread signal molecules that display a great chemical and functional diversity. Predictions of neuropeptide cleavage from precursor proteins are not always correct, and thus, biochemical identification is essential. Single-cell analysis is valuable to identify peptides processed from a single precursor, but also to determine coexpression of further neuropeptides from other precursors. We have developed an approach to isolate single identified neurons from the fruit fly Drosophila melanogaster for mass spectrometric analysis. By using Gal4 promoter lines to drive green fluorescent protein under UAS control, we identified specific peptidergic neurons. These neurons were isolated singly under a fluorescence microscope and subjected to MALDI-TOF mass spectrometry. Two Gal4 lines were used here to identify pigment-dispersing factor (PDF) and huginexpressing neurons. We found that the large PDF expressing clock neurons only give rise to a single peptide, PDF. The three different classes of hugin expressing neurons all display the same mass signal, identical to pyrokinin-2. The other peptide predicted from the hugin precursor, hugin γ, was not detected in any of the cells. Single-cell peptidomics is a powerful tool in Drosophila neuroscience since Gal4 drivers can be produced for all known neuropeptide genes and thus provide detailed information about neuropeptide complements in neurons of interest. Neuropeptides, which present the largest class of messenger molecules in the CNS, are cleaved from larger precursor molecules. Especially in invertebrates, single precursor proteins often contain multiple paralogs and other putative peptides. Although the cleavage sites may be predicted from the precursor sequence (see ref 1), the peptides that are expressed in the organism can differ from these predictions. In Drosophila melanogaster, this fact is well illustrated by the differential expression of CAPA peptides (periviscerokinins, PVKs) and pyrokinin (PK) in the CNS.2,3 The capa gene (CG15520) contains two introns; CAPA-PVK-1 is * To whom orrespondence should be addressed. Tel: +49 3641 949191. Fax: +49 3641 949192. E-mail: [email protected]. † Friedrich-Schiller-University Jena. ‡ Stockholm University. (1) Souhey, B. R.; Amare, A.; Zimmerman, T. A.; Rodriguez-Zas, S. L.; Sweedler, J. V. Nucleic Acids Res. 2006, 34, 267-272.

3690 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

encoded on exon 2, and CAPA-PVK-2 and CAPA-PK are encoded on exon 3.4 Mass spectrometric analyses of capa neurons revealed that abdominal neurosecretory cells5 express all predicted CAPA peptides6 together with a so-called CAPA precursor peptide B (CPPB).7 Neurosecretory cells of the subesophageal ganglion (SEG), however, express only the CPPB and a truncated form of CAPA-PK, the CAPA-PK2-15.6,7 This differential peptide expression might be caused by the presence of different sets of prohormone convertases in CAPA expressing cells. Thus far, such possibly enigmatic expression patterns are often neglected in studies of Drosophila, and putative functions of peptide-expressing neurons are commonly described without exact identification of the specific peptides that are expressed in the cells. Mass spectrometric methods are increasingly used to analyze the peptidome of the CNS even to the single-cell level (see ref 8). Prior to analysis, the tissues or cells of interest have to be identified. The first direct peptide profiling of single neurons by MALDI-TOF MS was performed using large neurons of a mollusk, Lymnaea stagnalis, which are easily identifiable without introducing exogeneous labels.9 Subsequently, single-cell analyses were published from a number of different animal taxa, including arthropods.8-15 Smaller neurons in more complex ganglia may be (2) Predel, R.; Nachman, R. J. The FXPRLamide (pyrokinin/PBAN) peptide family. In Handbook of Biologically Active Peptides; Kastin, A. J., Ed.; Elsevier: San Diego, CA, 2006; pp 207-212. (3) Predel, R.; Wegener, C. Cell. Mol. Life Sci. 2006, 63, 2477-2490. (4) Kean, L.; Cazenave, W.; Costes, L.; Broderick, K. E.; Graham, S.; Pollock, V. P.; Davies, S. A.; Veenstra, J. A.; Dow, J. A. T. Am. J. Physiol. 2002, 282, 1297-1307. (5) O’Brien, M. A.; Taghert, P. H. J. Exp. Biol. 1998, 201, 193-209. (6) Predel, R.; Wegener, C.; Russell, W. K.; Tichy, S. E.; Russell, D. H.; Nachman, R. J. J. Comp. Neurol. 2004, 474, 379-392. (7) Wegener, C.; Reinl, T.; Jaensch, L.; Predel, R. J. Neurochem. 2006, 96, 13621374. (8) Hummon, A. B.; Amare, A.; Sweedler, J. V. Mass Spectrom. Rev. 2006, 25, 77-98. (9) Jimenez, C. R.; van Veelen, P. A.; Li, K. W.; Wildering, W. C.; Geraerts, W. P.; Tjaden, U:R.; van der Greef, J. J. Neurochem. 1994, 62, 404-407. (10) Redeker, V.; Toullec, J. Y.; Vinh, J.; Rossier, J.; Soyez, D. Anal. Chem. 1998, 70, 1805-1811. (11) Neupert, S.; Predel, R.; Russell, W. K.; Davies, R.; Pietrantonio, P. V.; Nachman, R. J. Biochem. Biophys. Res. Commun. 2005, 338, 1860-1864. (12) DeKeyser, S. S.; Li, L. Mass spectrometric charting of neuropeptides in arthropod neurons. Anal Bioanal. Chem. 2007, 387, 29-35. (13) Neupert, S.; Predel, R. Biochem. Biophy.s Res. Commun. 2005, 327, 640645. (14) Rubakhin, S. S.; Churchill, J. D.; Greenough, W. T.; Sweedler, J. V. Anal. Chem. 2006, 78, 7267-7272. (15) Jime´nez, C. R.; Burlingame, A. L. Exp. Nephrol. 1998, 6, 421-428. 10.1021/ac062411p CCC: $37.00

© 2007 American Chemical Society Published on Web 04/18/2007

visualized by retrograde filling with dyes,8,13,16 a procedure that works particularly well when neuroendocrine neurons of larger organisms are used. However, retrograde filling is not a suitable technique to detect specific neurons in insects as small as D. melanogaster. Instead, in D. melanogaster, numerous Gal4 lines are available, which allow the identification of different neuron populations by expression of marker proteins such as GFP under UAS control. In this study, we investigated the suitability of such fluorochrome-labeled neurons (cell body diameter 10-20 µm) for a direct mass spectrometric analysis. Successful analysis of these cells not only reveals the peptides expressed from specific peptide precursors but also gives sufficient information about colocalisation with other neuropeptides. In addition, the peptidomes of transgenic fruit flies with peptide overexpression or knockdown as well as those of knockout mutants may become identifiable at the cellular level. For our experiments, we have chosen two different neuron populations, namely, the pigment-dispersing factor (PDF) expressing neurons of the optic lobes and the hugin neurons of the SEG. The PDF neurons are integrated into the circadian pacemaker system;17-19 the hugin neurons have been suggested to mediate the response to gustatory stimuli by influencing the foregut activity.20 EXPERIMENTAL SECTION Fly Stocks. Adult and larval flies of both sexes were used for experiments. They were raised under a 12-h-light, 12-h-dark photoperiod at room temperature and were fed with Instant Drosophila Medium (Sigma-Aldrich). Two GAL4 lines of D. melanogaster were used: pdf-GAL421 and hug-GAL4,20 kindly provided by J. C. Hall (Brandeis University) and M. Pankratz (Institut fur Genetik, Forschungszentrum Karlsruhe), respectively. The pdf-GAL4 lines were used to identify the PDF-containing neurons by driving GFP, and hug-GAL4 line was used for identifying the hugin neurons in the SEG by driving yellow fluorescent protein (YFP). The GAL4 lines were crossed with flies expressing UAS-cd8-gfp and UAS-cd8-yfp.22 Dissection of Single Cells and Sample Preparation for Mass Spectrometry. Isolated ganglia (SEG and brain/optic lobes) were fixed with microneedles, and the ganglionic sheath was removed using ultrafine scissors. Neuron-specific GAL4 lines driving fluorescent protein expressions (GFP or YFP) were detected by means of a stereofluorescence microscope (SteREO Lumar V12, Carl Zeiss AG, Goettingen, Germany) equipped with a GFP filter (488 nm) and a YFP filter (514 nm). Without any enzyme treatment, single identified cells were isolated and removed by using an uncoated glass capillary (tip diameter 1520 µm; Hilgenberg GmbH, Malsfeld, Germany) and transferred to a stainless steel sample plate for MALDI-TOF mass spectrometry. The insect saline (NaCl 7.5 g/L, KCl 0.2 g/L, CaCl2 0.2 g/L, NaHCO3 0.1 g/L; pH 7.2) was subsequently removed from the sample plate using the same capillary. Approximately 5-10 nL of (16) Rubakhin, S. S.; Greenough, W. T.; Sweedler, J. V. Anal. Chem. 2003, 20, 5374-5380. (17) Kaneko, M.; Hall, J. C. J. Comp. Neurol. 2000, 422, 66-94. (18) Helfrich-Fo ¨rster, C. Microsc. Res. Technol. 2003, 62, 94-102. (19) Helfrich-Fo ¨rster, C. Genes Brain Behav. 2005, 4, 65-76. (20) Melcher, C.; Pankratz, M. J. PloS. Biol. 2005, 3, 1618-1629. (21) Park, J. H.; Helfrich-Fo¨rster, C.; Lee, G.; Liu, L.; Rosbash, M.; Hall, J. C. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 3608-3613. (22) Lee, T.; Lou, L. Neuron 1999, 22, 451-461.

Figure 1. 1. Schematic representation of the circadian pacemaker neurons in the brain of adult D. melanogaster. Three clusters of dorsal neurons (DN1, DN2, DN3) are located in the protocerebrum. In the lateral part of the brain, additional pacemaker neurons exist (dorsal, LNd cells; ventral, LNv cells). The LNv cells express PDF and can be divided into small (s-LNv) and large (l-LNv) neurons. (modified from ref 34).

saturated matrix solution (R-cyano-4-hydroxycinnamic acid, Sigma Aldrich) was mixed with 50% methanol and injected onto the dried cell (which attached onto the metal surface) over a period of ∼5 s using a Nanoliter injector (World Precision Instruments, Berlin, Germany). Each preparation was air-dried again and covered with pure water for a few seconds, which was removed by cellulose paper (Fine Science Tools GmbH, Heidelberg, Germany). Rinsing with water is commonly used to reduce high salt content in biological samples (see ref 23). MALDI-TOF/TOF MS. MALDI-TOF mass spectra were acquired in positive ion mode on an Ultraflex II TOF/TOF MS (Bruker Daltonics, Bremen, Germany) equipped with LIFT technology. All acquisitions were taken in manual mode. Initially, the instrument was operated in reflectron mode, in order to determine the parent masses. The fragmentation data obtained in these experiments were handled using the flexAnalysis software package. An external mass spectrum calibration was first performed using synthetic cockroach peptides (Pea-pyrokinin-2/5; SPPFAPRLa/GGGGSGETSGMWFGPRLa). RESULTS AND DISCUSSION Mass Spectrometric Analysis of pdf Neurons. Insect PDFs are neuropeptides with a well-conserved, 18-amino acid sequence and display a high degree of homology to the pigment-dispersing hormones of crustacea.24 It is well-established that PDF plays a role as a major output factor of the circadian pacemaker in the Drosophila brain.25,26 In addition, pdf neurons of cockroaches and Drosophila were found to couple the bilaterally paired pacemaker centers of the accessory medullae.27-30 A null mutation in pdf as well as ectopic expression of PDF in the dorsal protocerebrum disrupts normal circadian locomotor rhythms.25,26 In adult Drosophila, PDF is localized in a subset of the circadian pacemaker neurons (Figure 1) known as the small and large ventral lateral neurons (s-LNvs and l-LNvs respectively; see ref 18). (23) Predel, R. J. Comp. Neurol. 2001, 436, 363-75. (24) Rao, K. R.; Riehm, J. P. Peptides 1988, 1, 153-159. (25) Renn, S. C.; Park, J. H.; Rosbash, M.; Hall, J. C.; Taghert, P. H. Cell 1999, 99, 791-802. (26) Helfrich-Fo ¨rster, C. J. Biol. Rhythms 2000., 2, 135-154. (27) Petri, B.; Stengl, M. J. Neurosci. 1997, 17, 4087-4093. (28) Reischig, T.; Stengl, M. J. Comp. Neurol. 2002, 443, 388-400. (29) Reischig, T.; Petri, B.; Stengl, M. Cell Tissue Res. 2004, 318, 553-564. (30) Park, D.; Griffith, L. C. J. Neurophysiol. 2006, 95, 3955-3960.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3691

Homolog PDF-containing neurons in optic lobe pacemaker centers of cockroaches were found to express a further neuropeptide, leucomyosuppressin.31,32 Since the pdf neurons are in a key position in the regulation of the circadian rhythm, the structural elucidation of further putative neuropeptides in these Drosophila neurons is of great interest for us. Thus, we have chosen the pdf neurons, which were visualized by GFP expression, as model neurons to test the feasibility of mass spectrometric analyses of identified single neurons of D. melanogaster. In a first step, the ganglionic sheath of adult Drosophila brains was disrupted, and the GFP expressing pdf neurons were identified by means of a fluorescence stereomicroscope. Cell clusters containing the large LNv cells and small LNv cells, respectively, were dissected and directly transferred to the stainless steel sample plate. Subsequent mass spectrometric analysis of these preparations yielded many ion signals in the mass range of 9003000 Da, including a number of mass signals identical with known Drosophila neuropeptides such as myosuppressin, MTYamide, DMS, Drm-kinin, and so on. (Figure 2A). Mass spectra of preparations containing the large LNvs also showed an ion signal at [M + H]+, m/z 1909.07 (Figure 2A), which is identical with the theoretical mass of Drosophila PDF (Drm-PDF). In a next step, we tried to isolate these large LNvs individually. The resulting mass spectra always contained the putative PDF signal, but also an inconsistant number of additional mass peaks. A subsequent examination of our “single-cell” dissections in an inverted microscope, which allowed a higher magnification, confirmed that one or more nonfluorescent neurons were regularly attached to the isolated pdf neurons. Since these neurons were very small (1012 µm) and did not express GFP, it was nearly impossible to detect these cells with the fluorescence stereomicroscope, which was used for the cell dissections. Therefore, the dissection protocol was completed by a screening of the dissected large pdf neurons in an inverted fluorescence microscope before the cells were transferred to the sample plate. Only those preparations that had been verified to be true single-cell dissections were used for mass spectrometric analysis. Surprisingly, signal intensity of putative PDF in these preparations was always higher than the signal intensity of PDF in preparations that contained additional neurons. Solely the putative PDF, but no other distinct ion signals, were detected (Figure 2B). Thus, the large LNvs expressed only a single abundant neuropeptide, whose mass was identical to that of PDF. By means of tandem mass spectrometry, this substance was subsequently fragmented using the single-cell preparation shown in Figure 2B, and the resulting fragment series confirmed the identity to Drm-PDF (Figure 2C). The complete procedure used to analyze GFP-expressing Drosophila neurons is summarized in Figure 3. Mass Spectrometric Analysis of hugin Neurons. The hugin gene of Drosophila was cloned based on sequence similarity to the human GDNF (glial derived neurotrophic factor) sequence.33 This gene encodes two putative neuropeptides with the C-terminal (31) Reischig, T.; Stengl, M. Cell Tissue Res. 2003, 314, 421-435. (32) Neupert, S. Pesticides 2006, 3, 183-187. (33) Meng, X.; Wahlstro ¨m, G.; Immonen, T.; Kolmer, M.; Tirronen, M.; Predel, R.; Kalkkinen, N.; Heino, T. I.; Sariola, H.; Roos, C. Mech. Dev. 2002, 117, 5-13. (34) Rieger, D.; Shafer, O. T.; Tomioka, K.; Helfrich-Fo¨rster, C. J. Neurosci. 2006, 26, 2531-2543.

3692 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

Figure 2. Representative MALDI-TOF mass spectra from (A) a protocerebral cell cluster containing GFP-expressing pdf neurons and (B) a single large LNv neuron. Note that the peptides which contaminate the PDF signal in (A) do not necessarily originate from the dissected cell bodies but can also result from fiber tracts crossing this cell cluster. (C) CID spectrum of the peptide at [M + H]+ m/z 1909.07 which was found to be abundant in large l-LNv-neurons (see B). Fragments were analyzed manually, and the resulting sequence confirmed the identity of this substance as Drm-PDF.

motif PRLamide (Figure 4). One of the peptides (Drm-pyrokinin2; Drm-PK-2) belongs to the FXPRLamides,2 the other peptide (hug-γ) is structurally related to ecdysis triggering hormone.33 Hug is thus the second gene, identified in the Drosophila genome, that encodes for a pyrokinin. Pyrokinins are known to be expressed in the SEG of insects2,3 and display diverse activities such as induction of pheromone biosynthesis, hindgut/oviduct contraction, melanization, pupariation, and diapause in insects. In Drosophila, hug expression starts in the latter half of embryogenesis; larvae,

Figure 3. Summary of the dissection protocol. Isolated ganglia (SEG and brain) were fixed with microneedles, the ganglionic sheath was removed, and single identified cells were isolated by means of a glass capillary (inset: l-LNvs neurons in the brain of D. melanogaster. These pdf neurons were visualized by crossing a pdf-Gal4 line with UAS-cd8-gfp flies). To check the success of the single-cell dissection, the preparation was subsequently examined using an inverted microscope. The cell was then transferred with a glass capillary to a stainless steel sample plate for MALDI-TOF mass spectrometry. Following the removal of the insect saline, the cell was washed with pure water for a few seconds and a few nanoliter matrix solution (R-cyano-4-hydroxycinnamic acid) was injected onto the cell. The air-dried preparation was again covered with pure water for a few seconds, which was removed by cellulose paper (not shown). After complete drying of the preparations, the sample plate was loaded in a MALDI-TOF mass spectrometer. Abundant peptides that were observed in the mass spectra could be subsequently fragmented by tandem mass spectrometry to reveal the sequences.

Figure 4. Sequence of the hugin precursor with the two putative gene products, Drm-PK-2 and hug γ.

and adults possess ∼20 hug-expressing neurons in the SEG. These hug neurons can be divided into three distinct subpopulations with axons projecting to different targets (Figure 5A): (1) neurons with neurites extending into the ring gland (cell bodies 1; CB1); (2) another set of neurons projecting to the pharyngeal muscles (CB2); and (3) neurons with extensions in the protocerebrum, the ventral nerve cord (CB3), or both.20 Hug-expressing neurons are likely to be involved in modulation of taste-mediated feeding behavior.20 It was postulated that the different neuropeptides of

the hug gene are translated or trafficked to different targets in subset of hug neurons.20 With a strategy similar to the one described above, we performed a mass spectrometric survey of all hug neurons. These hug neurons were visualized by crossing hug-GAL4 flies with flies expressing UAS-cd8-yfp20 and were found to be considerably smaller than the pdf neurons (5-8 µm). Individually dissected YFP-expressing hug neurons of the three subpopulations within the SEG were analyzed by MALDI-TOF mass spectrometry. The Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3693

of the hugin gene that was detectable in preparations of these neurohemal organs from wildtype flies.6,7

Figure 5. (A) Localization of the three subpopulation of hug neurons CB1-3 in the larval SEG of D. melanogaster (arrow). The hug neurons were visualized by crossing a hugin-Gal4 line with UAScd8-yfp flies. CB, cell bodies; PK, pyrokinin; SEG, subesophageal ganglion. (B) MALDI-TOF mass spectrum from a single hug neuron (CB1) of the SEG. Preparations of the different hug expressing neurons always revealed nearly identical spectra, representing a distinct ion signal which was mass identical with the theoretical mass of Drm-PK-2 ([M + H]+ m/z 942.5). Ions signals of hug γ, the other predicted neuropeptide from the hug gene, were not observed.

different neuron types (CB1-CB3) revealed nearly identical spectra, containing only a single distinct ion signal mass that was identical to that of Drm-PK-2 ([M + H]+ m/z 942.58) (Figure 5B). This substance was subsequently fragmented, and the resulting fragment series (not shown) confirmed the identity with DrmPK2. The mass spectra obtained from the hug neurons did not show any trace of the prediced hug-γ peptide or any other cleavage product of the hugin precursor. These data verify earlier findings obtained on preparations of the retrocerebral complex that stores the products of the hugin neurons. Drm-PK-2 was the only product

3694 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

CONCLUSIONS Our results clearly demonstrate that GFP/YFP expressing neurons of Drosophila are suitable for a direct mass spectrometric analysis; the fluorescence protein expression in these cells did not suppress ion signals in mass spectra. This was in sharp contrast to results obtained on immunostained neurons (see ref 13), which always yielded negative results (Neupert and Predel, unpublished). The results obtained on pdf and hugin neurons in this study and those obtained on CAPA-peptide expressing neurons (which were not studied at single-cell level),6,7 show the importance of our peptidomic approach. Whereas PDF is expressed as predicted, the processing of the hugin precursor results in a single peptide (Drm-PK2) instead of two, and the predicted hug γ was not detectable at all. Possibly a correct cleavage of hug γ is hampered by the introduction of Leu between Lys and Arg, which precede this peptide in the prohormone sequence. Other products of the pdf and hugin genes were not detectable, which demonstrates that intermediate products of these genes are obviously present at a much lower level than the final products that are stored in the vesicles. The capa precursor is not only differentially cleaved in the central nervous system, but an additional peptide with unknown functions (CPPB) was found to be expressed as well.7 These deviations from the predicted cleavage are easily detectable with mass spectrometric techniques but not by well-established methods such as in situ hybridization and immunocytochemistry. A prerequisite for a successful mass spectrometric analysis is the unambiguous identification of the cells of interest. Herein we describe a novel method to approach this challenge. ACKNOWLEDGMENT We acknowledge the financial assistance of the Deutsche Forschungsgemeinschaft (Predel 595/6-3) and the Karl Trygger Foundation (to D.R.N.). The authors thank O. Scheibner (HKI, Department of Biostructure Chemistry Jena, Germany), M. Marx (Carl Zeiss GmbH, Jena, Germany), M. Barth (Carl Zeiss GmbH, Jena, Germany), and C. Wegener (Emmy Noether Neuropeptide Group, Marburg, Germany) for support of this study.

Received for review December 21, 2006. Accepted March 26, 2007. AC062411P