Anal. Chem. 2004, 76, 5523-5533
Discovery and Neurochemical Screening of Peptides in Brain Extracellular Fluid by Chemical Analysis of in Vivo Microdialysis Samples William E. Haskins,†,‡ Christopher J. Watson,§ Nicholas A. Cellar,§ David H. Powell,† and Robert T. Kennedy*,§,|
Department of Chemistry, University of Florida, Gainesville, Florida 32611, and Department of Chemistry and Department of Pharmacology, University of Michigan, Ann Arbor, Michigan 48109
Endogenous peptides from brain extracellular fluid of live rats were analyzed using capillary liquid chromatography (LC)-tandem mass spectrometry (MS2). A 4-mm-long microdialysis probe perfused at 0.6 µL/min implanted into the striatum of anesthetized male rats was used to collect 3.6 µL dialysate fractions that were injected online into the capillary LC-MS2 system for analysis. A total of 3349 MS2 spectra were collected from 13 different animals under basal conditions and during localized depolarization evoked by infusion of a high-K+ solution through the microdialysis probe. Subtractive analysis revealed a total of 859 MS2 spectra that were observed only during depolarization. From these spectra, 29 peptide sequences (25 were peptides not previously observed) from 6 different protein precursors were identified using database searching software. Proteins identified include precursors to neuropeptides, synaptic proteins, blood proteins, and transporters. The identified peptides represent candidates for neurotransmitters, neuromodulators, and markers of synaptic activity or brain tissue damage. A screen for neuroactivity of novel proenkephalin fragments that were found was performed by infusing the peptides into the brain while monitoring amino acid neurotransmitters by microdialysis sampling combined with capillary electrophoresis. Three of the six tested peptides evoked significant increases in various neuroactive amino acids. These results demonstrate that this combination of methods can identify novel neurotransmitter candidates and screen for potential neuroactivity. Neuropeptides are important neurotransmitters and neuromodulators that play roles in many functions including learning, memory, appetite regulation, and sleep. Peptidergic signaling molecules are produced by processing of prohormones through cleavage and modification of select amino acids within neuronal secretory vesicles. Further processing of peptides can occur in the extracellular space after secretion to produce new peptides * Corresponding author: (e-mail)
[email protected]; (phone) (734) 6154363; (fax) (734) 615-6462. † University of Florida. ‡ Current affiliation: Sandia National Laboratory, Livermore, CA 94551. § Department of Chemistry, University of Michigan. | Department of Pharmacology, University of Michigan. 10.1021/ac049363y CCC: $27.50 Published on Web 08/06/2004
© 2004 American Chemical Society
with altered activity compared to the original released transmitter. A prohormone may only contain a single active fragment or it may contain the sequence of several neuropeptides, enabling the production of multiple transmitters further increasing the diversity of signaling capability of the neuron. General rules for processing of prohormones, such as cleavage at dibasic sites, are known;1 however, actual active peptides produced from a prohormone, when or where they may be produced, and physiological regulation of such processing are not easily predicted. Fragments of other proteins that accumulate in the extracellular fluid may also be of interest as markers of brain injury2 or disease.3 In this work, we describe an approach to identifying peptides in the extracellular space of the brain and screening for neurochemical activity using a novel combination of chemical analysis methods. A “peptidomic” route for studying neuropeptides has recently been introduced wherein peptides isolated from brain tissue are analyzed by capillary liquid chromatography (LC)-tandem mass spectrometry (MS2), and resulting MS2 spectra are correlated to sequences by searching protein databases.4 This method is similar to “shotgun” proteomics in which a protein mixture is pretreated by selective proteolysis to form a collection of peptides that is characteristic of the proteins and the protease (e.g., trypsin) prior to analysis.5 Application of this method to endogenous peptides in tissue is complicated by lack of control over the proteases used for digestion; i.e., the peptides produced in tissue are not tryptic, creating at least two difficulties. Peptides produced by other proteases may be less sensitively detected than tryptic peptides that have basic sites resulting in good ionization efficiencies and efficient cleavage along the peptide backbone promoting a high yield of b- and y-type ions.6 In addition, without knowledge of the proteases involved, a larger library of peptides must be searched, thus increasing the probability of finding a random match. Despite these difficulties, many new peptides have been found by this and similar approaches.4,7,8 (1) Devi, L.; Gupta, P.; Fricker, L. D. J. Neurochem. 1991, 56, 320-329. (2) Franz, G.; Beer, R.; Kampfl, A.; Engelhardt, K.; Schmutzhard, M.; Ulmer, H.; Deisenhammer, F. Neurology 2003, 60, 1457-1461. (3) Mills, J.; Reiner, P. B. J. Neurochem. 1999, 72, 443-460. (4) Skold, K.; Svensson, M.; Kaplan, A.; Bjorkesten, L.; Astrom, J.; Andren, P. E. Proteomics 2002, 2, 447-454. (5) Yates, J. R.; Morgan, S. F.; Gatlin, C. L.; Griffin, P. R.; Eng, J. K. Anal. Chem. 1998, 70, 3557-3565. (6) Olsen, J. V.; Ong, S. E.; Mann, M. Mol. Cell. Proteomics, in press.
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004 5523
A limitation of tissue analysis for studying neuropeptides is that no discrimination is made between intracellular peptides and those that are actually released into the extracellular space. In addition, tissue analysis cannot be used for correlating behavior to a given set of released peptides. Analysis of peptides in brain extracellular fluid of live animals would be a useful complement to tissue studies. Neuropeptides in the extracellular space may be continuously sampled in vivo by microdialysis9,10 or pushpull perfusion.11 Mass spectrometry and LC-MS have been coupled to microdialysis for peptide analysis;12-14 however, the main application has been targeted analysis of select peptides or determination of processing of exogenous peptides infused into the brain. While such studies are invaluable, they preclude discovery of novel endogenous species. Besides the difficulties of endogenous peptide tissue analysis mentioned above, analysis of endogenous peptides in microdialysis samples is hindered by the extremely low levels available. Dialysate sample volumes are a few microliters and peptide concentrations are in the 10-1210-10 M range;9 therefore, identification must be made on 0.1 were considered significant.23 Significant spectra were analyzed using Mascot with default settings (v. 1.8) and the same database. For both searches, no protease specificity was selected. Sequences were considered confirmed by Mascot if the score indicated homology with greater than 95% probability and the sequence matched that found by SEQUEST. Confirmed spectra were further analyzed by de novo sequencing using Lutefisk (v. 1.3.2) with a precursor ion tolerance
of 2.0 and a product ion tolerance of 0.8 (consistent with Mascot). Lutefisk generally only provided partial sequences that could then be compared to those obtained from SEQUEST and Mascot. If the first ranked de novo-derived partial sequence matched the first ranked database-derived sequence with greater homology than it matched the second ranked database-derived sequence peptide sequence, then the de novo sequencing was considered a successful confirmation. Comparison of Data Sets, Background Subtraction, and Library Matching of Spectra. Identification of matching spectra in different data sets was performed using the Xcalibur program Ionquest5 (a binning algorithm) set at default tolerances to determine matching. This program was used to subtract matching spectra obtained during basal conditions from those obtained during depolarizing conditions. This program was also used to compare data sets (such as different tryptic digest samples or results from different animals) to determine the reproducibility of spectra acquisition for a given sample type. Validated MS2 were examined for incorrect assignments by searching the MS2 spectra found to be significant by SEQUEST. Inspection of the difference spectrum (subtraction of matching MS2 spectra regardless of precursor ion mass or product ion threshold) was used to confirm a high degree of similarity between matching MS2 spectra. This “library” searching was performed using Xcalibur (v. 1.2) in conjunction with NIST MS Search (v. 1.7). The similarity search was selected with the default NIST MS Search options. Microdialysis On-Line Capillary Electrophoresis with Laser-Induced Fluorescence Detection (CE-LIF). Tests of the effects of peptide infusion were performed using CE-LIF coupled on-line to the microdialysis probe as described in detail elsewhere.24 In this instrument, sample exiting the microdialysis probe is mixed on-line with o-phthaldehyde/β-mercaptoethanol, a fluorogenic reagent that reacts with amines to form a fluorescent product, to form a continuous stream of derivatized dialysate. Samples from this stream are injected at 10-s intervals onto the electrophoresis capillary (10 µm i.d. by 10 cm long fused silica) where they are separated under an electric field of 2000 V/cm. Analytes exit the column into a sheath-flow cuvette where they are detected by fluorescence excited with the 354 nm line of an Ar ion laser. Microdialysis for this portion of the work was performed using 200 µm o.d. by 3 mm long (sampling length) probes fabricated in-house. Probes were implanted into the striatum as described above and perfused with balanced salt solution (this solution was 145 mM NaCl, 2.68 mM KCl, 1.01 mM MgSO4‚7H2O, 1.22 mM CaCl2, 1.55 mM Na2HPO4, and 0.45 mM NaH2PO4 set to a pH of 7.4 25 at 1 µL/min for 60 min. After this equilibration time, electropherograms were recorded for at least 5 min. The perfusion fluid was then switched to the same fluid spiked with 10 µM peptide for 10 min prior to switching back to control solution. If a change in amino acid level was observed during this 10-min period, then 10 µM tetrodotoxin (TTX) was perfused for 20 min followed by a mixture of 10 µM peptide and 10 µM TTX for 10 min. The average dialysate concentration for each amino acid obtained during the initial 5-min collection time was considered
(23) Ducret, A.; Van Oostveen, I.; Eng, J. K.; Yates, J. R.; Aebersold, R. Protein Sci. 1998, 7, 706-719.
(24) Bowser, M. T.; Kennedy, R. T. Electrophoresis 2001, 22, 3668-3676. (25) Bert, L.; Parrot, S.; Robert, F.; Desvignes, C.; Denoroy, L.; Suaud-Chagny, M. F.; Renaud, B. Neuropharmacology 2002, 43, 825-835.
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
5525
Figure 1. (A) Ion maps of MS2 spectra collected from a tryptic digest of conalbumin, carbonic anhydrase II, and lysozyme spiked with deuterated Leu-enkephalin. Two maps are shown from identical samples to illustrate reproducibility. (B) Sample MS2 spectra of acquired during analysis at the time indicated by the arrow. (C) Results from peptide identification by SEQUEST, Mascot, and Lutefisk. The top-ranked peptides by SEQUEST and Mascot matched each other. The de novo derived sequence from Lutefisk was incomplete, but the partial sequence matched the top-ranked sequence by database searching better than the lower ranked sequences.
basal, and all subsequent concentrations were reported relative to this value. Electrophoretic data were analyzed using software developed in-house. RESULTS Identification of peptides in brain dialysate samples presents several difficulties not typically encountered in shotgun proteomics including limited sample and lack of enzyme specificity. To identify peptides under these conditions, we adopted a serial approach to data analysis in which MS2 spectra acquired by data-dependent scanning were first analyzed by the database search algorithm SEQUEST5 to determine a set of significant spectra. This subset of spectra was analyzed by a second database search program Mascot,26 and the resulting sequence identifications were compared. Spectra for which identification was the same by both database search tools were then submitted to Lutefisk27 for de novo sequencing. This approach was tested on control samples that consisted of a tryptic digest of three proteins injected onto the capillary LC (26) Perkins, D. N.; Pappin, D. J.; Creasy, D. M.; Cottrell, J. S. Electrophoresis 1999, 20, 3551-3567. (27) Taylor, J. A.; Johnson, R. S. Anal. Chem. 2001, 73, 2594-2604.
5526 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
column at 400 amol, an amount expected for neuropeptides in dialysate samples, and MS2 spectra acquired by data-dependent scanning (Figure 1). In these experiments, 132 ( 9 spectra were collected per sample (n ) 3). Of the 395 total spectra acquired, 62 were found significant by SEQUEST and 24 of these were assigned the same sequence by Mascot. De novo sequencing using Lutefisk was successful for 19 of the 24 spectra found significant by both SEQUEST and Mascot (Table 1). (De novo sequencing never produced a complete sequence; however, this approach was considered successful if the partial sequence was a better match to the top-ranked sequence than the second-ranked sequence found by database searching.28) All 24 spectra identified by both database searching methods were considered true positives as they corresponded to predicted tryptic peptides in the mixture. Furthermore, the remaining 38 spectra not successfully identified by both methods were considered true negatives as they did not correspond to proteins in the digest or plausible contaminants. Deuterated Leu-enkephalin, spiked into the samples at the same concentration, was also detected and identified in each of the samples. (28) Johnson, R. S.; Taylor, J. A. Mol. Biotechnol. 2002, 22, 301-315.
Table 1. Peptide Sequences Identified from 400-amol Injection of Three-Protein Tryptic Digesta protein precursor (% sequence coverage, no. of observationsb) lysozyme (29%, 3) carbonic anhydrase II (11%, 3) conalbumin (3%, 3)
peptide sequence
de novo derived sequence
mass observedc
δd
amino acid position
FESNFNTQATNR NTDGSTDYGILQINSR GTDVQAWIR DFPIANGER ALVYGEATSR VGDANPALQK DGKGDVAFVK AQSDFGVDTK
[276.1]LAKTNFAMTR [262.1]LLKLLGYDTVNSR none [262.1]PLANGER none VGDAN[168.1]LKK DGAGGD[154.1]YVK [199.1]SDFRD[229.0]
1428.7 1754.0 1045.7 1017.5 1066.7 1011.7 1739.0 1666.8
1.0 1.2 1.1 0.0 1.2 0.1 0.2 0.5
34-45 46-61 117-125 18-26 47-56 148-157 200-209 270-279
a Three samples were subjected to LC-MS2 and resulting spectra analyzed by the sequential database searching and de novo sequencing protocol. Underlined portions of de novo derived sequence indicate homology with database derived sequence.27,28 b Three samples were analyzed. Number of observations indicates the number of samples that each of the peptides was observed in. c The molecular mass calculated from the m/z and charge state of the precursor ion used for identification. d Difference between calculated molecular mass for the peptide sequence and that observed.
Figure 2. Ion maps of MS2 spectra collected from a microdialysis sample collected from the striatum of an anesthetized rat. Collection was while control (A) and high-K+ (B) saline solutions were pumped through the microdialysis probes. Differences in the maps are reflective of the changes in the chemical composition of brain extracellular fluid evoked by depolarizing neuronal membranes with high K+. (C) Example of identified mass spectrum that was collected at the time indicated by the arrow.
These results suggest that the criteria used with database searching by two different methods can positively identify peptides at attomole levels. The de novo sequencing program has some utility for confirming a sequence; however, its high false negative rate suggests that its results should not be used to eliminate an identified sequence. These results are in good agreement with a
previous study that indicated that the most accurate sequences were obtained by database searching.28 These experiments also allowed reproducibility of the method to be assessed. Reproducibility of the acquired spectra was high as illustrated by the comparison of ion maps in Figure 1. Quantitative comparison of the spectra acquired for each sample Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
5527
Figure 3. Representative MS2 spectra acquired from dialysate samples with identified peptide sequences. Representative peptides are shown for proenkephalin A (A), neurogranin (B), fibrinogen R (C), fibrinogen β (D), excitatory amino acid transporter 1 (E), and brain acidic membrane protein (F).
revealed that 75% of the spectra in any two samples matched. Peptide identification was also consistent with the same eight peptides being identified in all samples. Peptide Sequencing and Protein Precursor Identification in Vivo. Samples of dialysate from basal conditions and during K+-induced depolarization were obtained from 13 animals and analyzed by capillary LC-MS2 with data-dependent scanning. These analyses yielded an average of 121 ( 28 (n ) 13) MS2 spectra per chromatogram from basal conditions and 144 ( 26 (n ) 13) during stimulation. Clear differences in the chromatograms obtained under basal and stimulated conditions indicate detection of many changes in the chemical environment of the brain extracellular fluid as a result of this manipulation (Figure 2). Our primary interest was in identifying peptides that would be released by K+ stimulation since neuropeptides are released under these conditions. To simplify analysis of these data, the spectra obtained under basal conditions that matched those obtained during K+ stimulation were subtracted from the K+stimulated data set for each animal, resulting in removal of 75% of the spectra from the K+-stimulated data sets. To ensure that no meaningful spectra were lost by this strategy, the basal data sets were analyzed for three animals. In no case was successful 5528 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
peptide identification made from the basal spectra. From this result, it was concluded that the basal data set consisted primarily of spectra that contained insufficient information for identification, were due to nonpeptidergic compounds, or were due to contaminants in the system components. The 859 background-subtracted MS2 spectra (66 ( 11 from each animal) were then analyzed by SEQUEST, Mascot, and Lutefisk as described above in order to identify peptides and confirm their sequence. SEQUEST analysis yielded 322 (average of 25 ( 3 per animal) MS2 spectra with a significant sequence match, and 55 of these spectra were found to have the same match by Mascot. A total of 39 of the 55 spectra were further confirmed by de novo sequencing. Sample spectra used for identification are given in Figures 2 and 3 to illustrate the quality of spectra obtained under these conditions. Of 13 animals tested, 3 produced no identified spectra and the rest produced from 1 to 11 identified peptides. The total of 55 spectra found significant by both database searching protocols corresponded to 29 different peptides, 25 of which were novel. The 29 peptides were from 6 different parent proteins and yielded sequence coverage of 3-28% for those proteins as summarized in Table 2.
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
5529
a All sequences shown were observed in K+-stimulated microdialysis samples and identified as described in the text. Headings are the same as for Table 1 except where indicated. Number of observations refers to the number of animals in which any fragment of this protein was observed.
Table 2. Peptides Observed in Microdialysis Samples Collected in Vivoa
Figure 4. Library matching to determine misidentified spectra. The spectrum shown in (A) was identified by SEQUEST with ∆Cn of 0.24 and Mascot as SPQLEDEAKE. The spectrum shown in (C) was identified by SEQUEST with ∆Cn of 0.11 and Mascot as AIKNGWLSEE. Comparison of the spectra by subtraction (B) reveals that the two spectra are a match for each other.
Library Matching. To prevent misidentification, individual spectra were compared to the library of validated spectra to determine whether any similar spectra had been assigned to different sequences. Using this matching routine, three misidentified peptides were found. Such misidentification was typically due to spurious signals in the misidentified spectrum (Figure 4). Therefore, library matching provides a good final check of sequence identification at the low levels encountered in dialysis samples. Comparison to Tryptic Digest at Similar Levels. Comparison of the results obtained from microdialysis samples and the low-level tryptic digest highlights effects of working with the less controlled environment of in vivo sampling. The fraction of total spectra acquired that were ultimately identified was 6% in both types of sample. A greater portion of the dialysis spectra were retained in the initial SEQUEST screen for the dialysis samples (37%) than the tryptic digest (16%), perhaps reflecting bias toward use of b- and y-type ions typically formed by tryptic peptides. Sample-to-sample reproducibility, measured as the percent of spectra that matched in any two samples, was higher in digest samples (75% matching) than in dialysis samples (33% matching after background subtraction). This could be due to several factors including the greater complexity, animal-to-animal differences, and a larger number of low-quality spectra due to low-level analytes associated with in vivo samples. Neurochemical Activity Testing. An initial screen for neurochemical activity of the six non-opioid peptides derived from proenkephalin A was performed by infusing synthetic peptide into the striatum by microdialysis while monitoring amino acids in the dialysate. In this experiment, the dialysis probe is used to simultaneously deliver the peptide and sample the amino acids from a given brain region. The dialysate solution was monitored by CE-LIF, which allows eight amino acids to be monitored at 5530
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
10-s intervals (Figure 5A). These studies revealed that three of the six peptides significantly altered at least one neuroactive amino acid (Figure 5C-H). In all cases, the evoked responses were rapid as illustrated by the sample trace in Figure 5B. The two peptide I fragments had similar effects of significantly elevating excitatory amino acids while VGRPEWWMDYQ (this peptide corresponds to bovine adrenal medullary peptide fragment 8-18 (BAM 8-18)29) substantially increased GABA and taurine. The regulation of amino acids in the extracellular space is complex and involves not only neuronal release but also control by various transporter.s30,31 To determine whether the changes evoked by the peptides were related to neuronal firing, the Na+-channel blocker TTX, which prevents action potentials, was infused with the active peptides. TTX completely abolished the effect of BAM 8-18 but did not significantly inhibit the effect of the peptide I fragments. DISCUSSION We have described an approach to identify peptides released into the brain extracellular space of live animals and to test them for possible neurochemical activity. This experiment has yielded a novel group of non-opioid peptides derived from proenkephalin A, some of which appear to have neuronal signaling properties. In addition, other peptides that are candidate biomarkers of brain trauma and synaptic activity have been found. Peptide Identification in Microdialysis Samples. Peptide identification at low levels without a priori knowledge of peptidases (29) Lembo, P. M.; Grazzini, E.; Groblewski, T.; O’Donnell, D.; Roy, M. O.; Zhang, J.; Hoffert, C.; Cao, J.; Schmidt, R.; Pelletier, M.; Labarre, M.; Gosselin, M.; Fortin, Y.; Banville, D.; Shen, S. H.; Strom, P.; Payza, K.; Dray, A.; Walker, P.; Ahmad, S. Nat. Neurosci. 2002, 5, 201-209. (30) Miele, M.; Boutelle, M. G.; Fillenz, M. J. Physiol. 1996, 497 (Pt 3), 745751. (31) Baker, D. A.; Xi, Z. X.; Shen, H.; Swanson, C. J.; Kalivas, P. W. J. Neurosci. 2002, 22, 9134-9141.
Figure 5. Effect of peptide infusion on amino acid concentrations in brain. (A) A typical electropherogram obtained for on-line analysis of rat striatum dialysate. Peaks a-h in the traces correspond to o-phosphoethanolamine, glutamate, aspartate, GABA, taurine, glutamine, serine, and glycine, respectively. (B) Fluctuation of glutamate, aspartate, GABA, and taurine during the delivery of 10 µM peptide I 1-10 (SPQLEDEAKE), 10 µM TTX, and both through the dialysis probe. Electropherograms were acquired at 10-s intervals and resulting peak heights plotted relative to the average peak height from electropherograms acquired for 5 min prior to infusion. (C-H) Summary of the effect of non-opioid proenkephalin A peptides identified in this work on glutamate, aspartate, GABA, and taurine. Asterisks indicate statistically significant changes at p < 0.05 using Student’s t-test with comparison to control unless shown otherwise (n ) 3-8 depending on the experiment). Other amino acids that were monitored did not show any changes for any of the peptides.
was not as straightforward as that for shotgun proteomics. Application of multiple database searching methods along with
de novo sequencing and library matching minimized the risk of false-positive matches. The use of high-sensitivity capillary LCAnalytical Chemistry, Vol. 76, No. 18, September 15, 2004
5531
MS2 was important for obtaining spectra of sufficient quality for identification of attomole quantities of peptide. In a previous study, we had demonstrated that coupling of capillary LC columns with integrated electrospray emitter tips could detect known peptides at low-attomole levels in microdialysis samples.22 We also demonstrated that data-dependent scanning over a narrow precursor ion window (550-600 m/z) allowed identification of a single new peptide.22 For this work, it was found that doubling the injection volume to 3.6 µL and lowering the chromatography flow rate from 20 to 10 nL/min improved sensitivity, presumably through oncolumn preconcentration and improved ionization efficiency,32 respectively, sufficiently to generate identifiable spectra for trace level peptides in dialysate even while scanning over a wide range, thus enabling detection and identification of more peptides. It is estimated that no more than 1 fmol was available for any of the peptides identified. Significance of Peptides Found. A total of 29 peptides from 6 different proteins were found in dialysates after depolarization. Brain acidic membrane protein (also called NAP-22)33 is localized to synapses and synaptic vesicles;34 therefore, detection of fragments of this protein in the extracellular space after depolarization can be presumed to result from release during exocytosis. The concentration of these peptides may be a good marker for synaptic activity. The presence of fragments of the cytosolic protein neurogranin35 and a membrane-bound amino acid transporter36 in the extracellular space could result from normal degradation of these proteins (stimulated by neuronal depolarization) or from membrane damage associated with implantation of the microdialysis probe or the K+ stimulation. The presence of fragments of the blood protein fibrinogen in the dialysate suggests impairment of the blood-brain barrier. While no blood peptides were detected during basal levels in these experiments, subsequent experiments in our laboratory have revealed that fibrinogen peptides can be detected without K+ stimulation. Furthermore, a recent proteomic study of human microdialysis samples revealed that fibrinogen γ chain was present in extracellular fluid without stimulation.37 These observations suggest that their presence may be due to the dialysis implant. Detection of fragments of proenkephalin A are not surprising as this protein is a precursor to the neurotransmitters Metenkephalin and Leu-enkephalin, and some fragments of this protein would be expected to be co-released with these neurotransmitters during depolarization-induced exocytosis. It is not known if the proenkephalin A peptides detected were those stored in vesicles or if they were degradation products of other fragments that were released. Peptide I 1-12 has been found in tissue, (32) Bateman, K. P.; White, R. L.; Thibault, P. Rapid Commun. Mass Spectrom. 1997, 11, 307-315. (33) Maekawa, S.; Maekawa, M.; Hattori, S.; Nakamura, S. J. Biol. Chem. 1993, 268, 13703-13709. (34) Yamamoto, Y.; Sokawa, Y.; Maekawa, S. Neurosci. Lett. 1997, 224, 127130. (35) Houben, M. P.; Lankhorst, A. J.; van Dalen, J. J.; Veldman, H.; Joosten, E. A.; Hamers, F. P.; Gispen, W. H.; Schrama, L. H. J. Neurosci. Res. 2000, 59, 750-759. (36) Mitrovic, A. D.; Amara, S. G.; Johnston, G. A. R.; Vandenberg, R. J. J. Biol. Chem. 1998, 273, 14698-14706. (37) Maurer, M. H.; Berger, C.; Wolf, M.; Futterer, C. D.; Feldmann, R. E., Jr.; Schwab, S.; Kuschinsky, W. Proteome Sci. 2003, 1, 7.
5532 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
Figure 6. Depiction of a selection of previously observed proenkepahlin A-derived peptides13,40-54 in relation to those identified in this work by amino acid position. Unshaded bars represent opioid peptides containing either YGGFM or YGGFL. Dibasic sites, representing predicted cleavage sites, are indicated by vertical bars. Solid arrows indicate non-opioid peptides identified in this work. Known peptides are shown with a name where given.
suggesting that at least this peptide was created intracellularly.4,38 Despite extensive studies of prohormone processing, six of the eight proenkephalin peptides detected in this study either had not been observed before or were only recently revealed by peptidomic methods (see Figure 6). This reflects the emphasis of previous methods on use of antibodies that bind enkephalinergic epitopes. Detection of these previously unknown peptides demonstrates the power of a peptidomic approach to study processing of endogenous neurotransmitter precursors. Production of Peptide Fragments. The presence of specific peptides in the dialysate samples raises the issue of proteases involved in generating these fragments. Four of the proenkephalin A peptides identified, including Met- and Leu-enkephalin, were encompassed by dibasic sites consistent with production through prohormone convertase activity (Figure 6). Peptide fragment production from the other proteins is uncertain. A frequent observation was nesting of sequences. For neurogranin, seven peptides were found, but all were within the C-terminal region of 53-78 (out of 78) amino acids (Table 2). A likely route to production of nested peptides is cleavage by an endopeptidase at a specific residue followed by carboxy- and aminopeptidase activity to create the other peptides. Bioactivity Testing. Peptidomic methods enable peptides to be discovered in biological samples; however, these methods do not convey information on peptide function. As a screen for possible bioactivity, select peptides were infused into the brain (38) Lewis, R. V.; Ray, P.; Blacher, R.; Stern, A. Biochem. Biophys. Res. Commun. 1983, 113, 229-234.
while several amino acids were monitored. Because neurotransmitter or neuromodulator signaling frequently alters secretion of other transmitters or metabolites, this method is a general way to test for activity. The CE-LIF method enables eight amino acids to be monitored simultaneously, which is important because it is impossible to predict the neurons, neurotransmitters, or metabolites that will be affected by a novel peptide. Screening focused on fragments of proenkephalin A because it is common for prohormones to contain within their sequence multiple active peptides. The success of this approach was demonstrated when three of the novel peptides were found to alter levels of neuroactive amino acids. The peptide I fragments had similar effects of elevating the excitatory amino acids, suggesting that the mechanism of action is likely to be the same for these peptides. The effects did not appear to require neuronal firing; however, alteration in excitatory amino acid level evoked by these peptides could alter excitability of neighboring neurons and therefore modulate neuronal signaling.39 The effects of BAM 8-18 indicate an ability to alter signaling of the primary inhibitory neurotransmitter (GABA) and a putative (39) Baker, D. A.; Xi, Z. X.; Shen, H.; Swanson, C. J.; Kalivas, P. W. J. Neurosci. 2002, 22, 9134-9141. (40) Howells, R. D.; Kilpatrick, D. L.; Bhatt, R.; Monahan, J. J.; Poonian, M.; Udenfriend, S. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 7651-7655. (41) Liebisch, D. C.; Weber, E.; Kosicka, B.; Gramsch, C.; Herz, A.; Seizinger, B. R. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 1936-1940. (42) Jones, B. N.; Shively, J. E.; Kilpatrick, D. L.; Kojima, K.; Udenfriend, S. Proc. Natl. Acad. Sci. U.S.A. 1982, 79, 1313-1315. (43) Jones, B. N.; Kilpatrick, D. L.; Stern, A. S.; Kimura, S.; Shively, J. E.; Taniguchi, T.; Hullihan, J. P.; Stein, S.; Udenfriend, S. Fed. Proc. 1981, 40, 1641. (44) Jones, B. N.; Shively, J. E.; Kilpatrick, D. L.; Stern, A. S.; Lewis, R. V.; Kojima, K.; Udenfriend, S. Proc. Natl. Acad. Sci. U.S.A. 1982, 79, 2096-2100. (45) Kilpatrick, D. L.; Taniguchi, T.; Jones, B. N.; Stern, A. S.; Shively, J. E.; Hullihan, J.; Kimura, S.; Stein, S.; Udenfriend, S. Proc. Nat. Acad. Sci. U.S.A. 1981, 78, 3265-3268. (46) Kilpatrick, D. L.; Jones, B. N.; Lewis, R. V.; Stern, A. S.; Shively, J. E.; Udenfriend, S. Fed. Proc. 1982, 41, 1077. (47) Rosen, H.; Douglass, J.; Herbert, E. J. Biol. Chem. 1984, 259, 14309-14313. (48) Yoshikawa, K.; Williams, C.; Sabol, S. L. J. Biol. Chem. 1984, 259, 1430114308. (49) Yoshikawa, K.; Maruyama, K.; Aizawa, T.; Yamamoto, A. FEBS Lett. 1989, 246, 193-196. (50) Rao, S. M.; Howells, R. D. Regul. Pept. 1992, 40, 397-408. (51) Goumon, Y.; Lugardon, K.; Kieffer, B.; Lefevre, J. F.; Van Dorsselaer, A.; Aunis, D.; Metz-Boutigue, M. H. J. Biol. Chem. 1998, 273, 29847-29856. (52) Goumon, Y.; Lugardon, K.; Gadroy, P.; Strub, J. M.; Welters, I. D.; Stefano, G. B.; Aunis, D.; Metz-Boutigue, M. H. J. Biol. Chem. 2000, 275, 3835538362. (53) Goumon, Y.; Strub, J. M.; Moniatte, M.; Nullans, G.; Poteur, L.; Hubert, P.; Van Dorsselaer, A.; Aunis, D.; Metz-Boutigue, M. H. Eur. J. Biochem. 1996, 235, 516-525. (54) Lewis, R. V.; Ray, P.; Blacher, R.; Stern, A. Biochem. Biophys. Res. Commun. 1983, 113, 229-234.
neuromodulator (taurine). From these results, we conclude that in addition to its known role in opioid neurotransmission, proenkephalin A processing leads to other peptide products that are released into the extracellular space with neurochemical activity not associated with the opioid receptor. The differential effects of these peptides illustrate the diverse signaling possible by a single precursor protein. The lack of an effect by the other peptides on amino acid levels does not eliminate their consideration as bioactive molecules. Many other neurotransmitters are present in the striatum including acetylcholine, dopamine, and substance P, which could be altered but were not detected by this system. In addition, peptide effects on release of neurotransmitters could be suppressed due to anesthesia or not detected if the peptide exerts its effect on a neuron that projects to another brain region. Therefore, future work could be directed toward assessing the effect of these peptides on other neurotransmitters, in brain regions that receive projections from the striatum and in awake animals. It may also be fruitful to screen these peptides in receptor binding assays. Slight extensions of BAM 8-18 were found to activate an orphan G protein-coupled receptor found in rat small sensory neurons indicating the potential of such assays for elucidating the role of novel peptides.29 CONCLUSIONS This method has the potential to identify novel neurotransmitter candidates, processing of endogenous neuropeptides, and markers of brain injury. Such applications are important in developing drugs, treatments for brain trauma, and methods of preventing biofouling of brain implants. The method may also ultimately have application to behavioral studies wherein peptides in dialysates collected under different physiological states, such as sleep and wakefulness or hungry and satiated, are compared to determine signaling molecules involved in these conditions. Reaching this goal may require further advances in sensitivity, which could be obtained through improved instrumentation or sampling, so that peptides may be identified without artificial stimulation of secretion. ACKNOWLEDGMENT This work was supported by NIH (NS38476). We thank Joanna Peris (University of Florida) for confirming microdialysis probe placement. Received for review April 30, 2004. Accepted July 1, 2004. AC049363Y
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
5533
