Separation and Identification of Peptides in Single Neurons by

Physics and Biophysics, University of Münster, Robert-Koch-Strasse 31, D-48149 Münster, ..... We acknowledge financial support from the Research Ins...
0 downloads 0 Views 93KB Size
Download PDF
Anal. Chem. 1998, 70, 1847-1852

Separation and Identification of Peptides in Single Neurons by Microcolumn Liquid Chromatography-Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry and Postsource Decay Analysis Showchien Hsieh,‡ Klaus Dreisewerd,† Roel C. van der Schors,† Connie R. Jime´nez,† Jianru Stahl-Zeng,§ Franz Hillenkamp,§ James W. Jorgenson,‡ Wijnand P. M. Geraerts,† and Ka Wan Li*,†

Department of Molecular and Cellular Neurobiology, Graduate School Neurosciences Amsterdam, Research Institute Neurosciences, Vrije Universiteit, Faculty of Biology, De Boelelaan 1087, NL-1081 HV Amsterdam, The Netherlands, Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290, and Institute of Medical Physics and Biophysics, University of Mu¨ nster, Robert-Koch-Strasse 31, D-48149 Mu¨ nster, Germany

Microcolumn liquid chromatography (LC) was interfaced with matrix-assisted laser desorption/ionization time-offlight mass spectrometry (MALDI-TOF-MS) for separation and identification of peptides present in single neurons from the brain of the snail Lymnaea stagnalis. The nanoliter microcolumn LC effluent, mixed off-line with nanoliter matrix solution, was deposited onto the sample target every 60 s, producing fractions of approximately 145 nL in volume, which, upon drying, produced spots of approximately 1 mm in size. At the end of the chromatographic separation, fractions from the sample target were scanned by MALDI-TOF-MS. Identification of peptide peaks was achieved on the basis of LC elution order and mass information. Further identification based on sequence information was carried out for a native peptide fractionated by microcolumn LC from a single neuron with the postsource decay technique. The analysis of chemical messengers at the single-cell level is important for neuroscience research, since even closely apposed, morphologically similar neurons may use different messengers for cell-to-cell communication.1 Generally, two main classes of messengers exist, the classic transmitters and neuropeptides, and different methodologies are used for their detection. In the past decade, several innovative techniques have been applied to the studies of these molecules contained in single cells. In most cases, attention was focused on structurally simple electroactive transmitters such as dopamine, 5-hydroxytryptamine, norepinephrine, epinephrine, and histamine, which, after separation by microcolumn liquid chromatography (LC) or capillary electrophoresis, were detected electrochemically, with sensitivities at the attomole level.2-5 Microcolumn separations with electrochemical detection †

Vrije Universiteit Amsterdam. University of North Carolina. § University of Mu ¨ nster. (1) de Lange, R. P. J.; van Golen, F. A.; van Minnen, J. J. Neurosci. 1997, 78, 289-299. ‡

S0003-2700(97)00829-9 CCC: $15.00 Published on Web 04/01/1998

© 1998 American Chemical Society

are less suitable for the detection of neuropeptides, since electrochemical detection is profoundly less sensitive for this class of messenger molecules than for biogenic amines. Furthermore, electrode surfaces easily get fouled from peptide and protein molecules. More recently, laser-induced fluorescence detection coupled with microcolumn separation has been used to examine amine-containing molecules from complex biological samples.6 This technique is very sensitive, with detection limits in the zeptomole range for derivatized molecules. However, there are several drawbacks: N-terminally blocked peptides with no other free amine group are not amenable to derivatization, since most fluorescent tagging techniques rely on reactions of the tagging reagent with primary or secondary amines. On the other hand, peptides with several possible amine tagging sites in the side chains of amino acid residues may give rise to multiple products with different degrees and sites of derivatization. Moreover, the high structural complexity among the various peptide families (e.g., in the mammalian brain, over 100 neuropeptides have been reported) implies that the mere separation and detection of the molecules may not be sufficient to unequivocally reveal the identities of the peptides. It is apparent that an alternative technique capable of providing structural information is desirable for the analysis of peptides. Mass spectrometry (MS), especially tandem MS (MS/MS), enables one to obtain structural information on the analytes, such as peptides. Recent advancements in desorption/ionization techniques, i.e., electrospray ionization (ESI) and matrix-assisted laser desorption ionization (MALDI), allow the mass spectrometric detection of peptides and proteins at the femtomole level. Among (2) Cooper, B. R.; Jankowski, J. A.; Leszczyszyn, D. J.; Wightman, R. M.; Jorgenson, J. W. Anal. Chem. 1992, 64, 691-694. (3) Pihel, K.; Hsieh, S.; Jorgenson, J. W.; Wightman, R. M. Anal. Chem. 1995, 67, 4514-4521. (4) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 436-441. (5) Olefirowicz, T. M.; Ewing, A. G. J. Neurosci. Methods 1990, 34, 11-15. (6) Larmann, J. P., Jr.; Lemmo, A. V.; Moore, A. W., Jr.; Jorgenson, J. W. Electrophoresis 1993, 14, 439-447.

Analytical Chemistry, Vol. 70, No. 9, May 1, 1998 1847

the two techniques, MALDI-MS is generally more straightforwardly applied to the analysis of complex biological sample due to its high sensitivity, easily interpretable mass spectra, and relatively low susceptibility to salts and detergents. The detection limit in MALDI-MS can further be lowered to the low attomole level by simply reducing the sample volume.7 Based on these advantageous properties of MALDI-MS, it has been applied to the direct analysis of peptides contained in single neurons8-11 and single erythrocyte.12 Recently, we have shown the feasibility of structural characterization of a small peptide (3000 Da) may not yield informative fragment ions, which restricts the approach of direct single-cell peptide sequencing to the analysis of small peptides, whereas on-target enzymatic digestion14 of the fractionated peptides and subsequent postsource decay (PSD) or tandem MS analysis of the product peptides may also permit the identification of large peptides and proteins. In the work presented, the giant identifiable visceral dorsal 1 (VD1) neuron (100-µm cell body diameter), located in the visceral ganglion of the freshwater snail, Lymnaea stagnalis, and known to play an important role in cardiorespiratory processes of Lymnaea, was used to demonstrate the feasibility of peptide fractionation from single neurons by microcolumn separations and their mass measurements by MALDI-MS. The structural information of one of the peptides found in the microcolumn LC fractions was furthermore (partially) elucidated by MALDI-PSD analysis. Together, the combined three techniques can define the identity of the peptides contained in single cells by (1) their elution positions from the microcolumn, (2) their molecular weights, and (3) the structurally characteristic fragments generated by PSD. EXPERIMENTAL SECTION Chemical Reagents. 2,5-Dihydroxybenzoic acid (DHB) and R-cyano-4-hydroxycinnamic acid (4-HCCA) were obtained from Sigma Chemical Co. (St. Louis, MO). The synthetic test peptide (DRADILYNIAQ) used in the microcolumn LC, run as an internal (7) Jespersen, S.; Niessen, W. M.; Tjaden, U. R.; van der Greef, J.; Litborn, E.; Lindberg, U.; Roeraade, J. Rapid Commun. Mass Spectrom. 1994, 8, 581584. (8) Jime´nez, C. R.; van Veelen, P. A.; Li, K. W.; Wildering, W. C.; Geraerts, W. P. M.; Tjaden, U. R.; van der Greef, J. J. Neurochem. 1994, 62, 404-407. (9) van Veelen, P. A.; Jime´nez, C. R.; Li, K. W.; Wildering, W. C.; Geraerts, W. P. M.; Tjaden, U. R.; van der Greef, J. Org. Mass Spectrom. 1993, 28, 15421546. (10) Li, K. W.; Hoek, R. M.; Smith, F.; Jime´nez, C. R.; van der Schors, R. C.; van Veelen, P. A.; Chen, S.; van der Greef, J.; Parish, D. C.; Benjamin, P. R.; Geraerts, W. P. M. J. Biol Chem. 1994, 269, 30288-30292. (11) Gardon, R. W.; Moroz, L. L.; Moroz, T. P.; Shippy, S. A.; Sweedler, J. V. J. Mass Spectrom. 1996, 31, 1126-1130. (12) Li, L.; Golding, R. E.; Whittal, R. M. J. Am. Chem. Soc. 1996, 118, 1166211663. (13) Li, K. W.; Kingston, R.; Dreisewerd, K.; Jime´nez, C. R.; van der Schors, R. C.; Bateman, R. H.; Geraerts, W. P. M. Anal. Chem. 1997, 69, 563-565. (14) Glocker, M. O.; Bauer, S. H. J.; Kast, J.; Voly, J.; Przybylski, M. J. Mass Spectrom. 1996, 31, 1221-1227.

1848 Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

standard, and the synthetic small cardioactive peptide b (pQNYLAFPRMamide), used for PSD analysis, were synthesized inhouse. The mobile phase used for chromatographic separation consisted of an organic gradient. Before use, the mobile phase was filtered with a 0.2-µm nylon membrane filter (Millipore Corp., Bedford, MA). All solutions were prepared with deionized purified water. Chromatographic System. The capillary columns (50 µm i.d. × 30 cm) were slurry-packed in the laboratory with 5-µm octadecylsilane chemically modified silica particles with 120-Å pore size (YMC, Wilmington, NC). The packing procedure was similar to that previously described,15 with minor modifications in making the frit. The frit was formed by tapping the end of the capillary into a pile of 10-µm glass beads (Duke Scientific, Palo Alto, CA). Once a band of beads approximately 250 µm long was in place, it was sintered with a laboratory-built microelectric arcing device similar to the one previously described.16 The microcolumn LC system was similar to that described previously.15 A commercial syringe pump (model 140 B, PerkinElmer/Applied Biosystems Division, Foster City, CA) was used to deliver a flow rate of 0.1 mL/min. To obtain reproducible low flow rates through the packed capillary column, splitter capillaries matched in inner diameter and length were used to maintain a flow rate through the packed capillary column of approximately 85 nL/min. The gradient used was 100% eluent A from 0 to 5 min and a linear gradient from 10% B/90% A from 5 min to 90% B/10% A in 50 min. Eluent A consisted of 10% acetonitrile (ACN)/ 90% of 0.1% trifluoroacetic acid (TFA), and eluent B consisted of 80% ACN/10% of 0.1% TFA. Sample Preparation for MALDI-MS and PSD Analysis. A schematic diagram for the collection of the microcolumn effluent is shown in Figure 1. Either DHB or 4-HCCA, prepared at a concentration of 10 mg/mL in 30% ACN/70% 7.5 mM TFA and 20 mg/mL in 60% ACN/40% 7.5 mM TFA, respectively, was used as matrix. By applying a pressure of 1 bar to a stainless steel pressure reservoir containing the matrix, the matrix effluent was delivered through a capillary (25 µm i.d. × 30 cm) at a flow rate of 60 nL/min. The capillaries were threaded into a 1-mm-i.d. glass capillary, which was held stationary, while the polished stainless steel target was placed on an x-y-z positioning unit. Because of the close proximity between the two capillaries, the matrix and the eluent from the microcolumn fused into a single droplet. The droplet was deposited onto the steel target every 60 s by adjusting the z-axis of the positioning unit such that the capillaries came into contact with the target. Therefore, each fraction consisted of approximately 145 nL and was evaporated within 30 s, producing sample spots of approximately 1 mm in diameter. This mixing procedure is comparable to that of the conventional dried-droplet method. After the deposition of fraction, the capillaries were retracted in the z-direction, and the target was moved by about 2 mm in the x- or y-direction for collection of the next fraction. At the end of the chromatographic separation, the target was transferred to the source of the mass spectrometer. Isolation and Processing of Single VD1 Neurons. Isolation of single VD1 neurons from the visceral ganglion of L. stagnalis (15) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 1128-1135. (16) Hoyt, A. M.; Beale, S. C.; Larmann, J. P., Jr.; Jorgenson, J. W. J. Microcolumn Sep. 1993, 5, 325-330.

Figure 1. Schematic diagram of the system for coupling of the microcolumn LC system with MALDI-MS. Splitter capillaries were used to split the 0.1 mL/min pump flow to 85 nL/min flow through the microcolumn. The matrix solution at 60 nL/min was mixed with effluent at the end of the column. Every 60 s, the droplet is deposited onto the target by adjusting the manipulator.

was performed as previously described.8 Briefly, the VD1 neuron was dislodged from the visceral ganglion under a microscope with the aid of tweezers and tiny hooks. Once isolated, the neuron was transferred into a 300-nL microvial constructed in the laboratory. A 20-nL aliquot of a 10 µM solution of a synthetic “internal standard” peptide with a molecular mass of 1376.5 Da in 0.5 N perchloric acid was dispensed into the microvial with the aid of a cell injection apparatus (model IM-100, Narishige, Tokyo, Japan) supplied with 1 bar of argon gas. The microvial was immediately capped with several layers of paraffin-coated wrapping film and placed in dry ice. After dispensing, the micropipet was calibrated by dispensing a drop of the same solution into a dish of mineral oil. The diameter of the droplet was measured with a calibrated reticule in the eyepiece of a light microscope and converted to volume. The capped microvials were centrifuged at 12000g for 8 min. The supernatant from the microvial was injected directly into the microcolumn with the aid of a microinjector pipet attached to the cell injection apparatus. The sample-loaded microcolumn was reconnected to the first tee (tee closest to solvent filter, see Figure 1), and the peptides were fractionated and collected as described above. For direct single-cell MALDI-MS analysis, VD1 neurons were applied directly to a 0.5-µL solution of DHB, similar to the procedure previously described.8 MALDI-MS. MALDI-MS analysis was performed on a laboratory-built laser desorption reflectron-type time-of-flight (TOF) mass spectrometer equipped with a pulsed UV nitrogen laser (337 nm, 5-ns pulse width, Laser Photonics, Orlando, FL). Ions were accelerated to an energy of 10.5 keV and detected by a Venetian blind secondary electron multiplier (Thorn-Emi 9643/4A) equipped with a conversion dynode for postacceleration of ions. This relatively slow, albeit robust, secondary electron multiplier set the achievable mass resolution to about 500. A gridless reflector compensates for the initial energy dispersion of the desorbed ions. A 500-MHz transient recorder (Le Croy 9350AM, Le Croy, Chestnut Ridge, NY) was used for data acquisition. The mass

spectra were transferred to a PC for further signal processing. The instrument was equipped with a CCD camera for sample observation/identification and laser spot localization with a resolution of approximately 20 µm. The laser spot size on the target was 200 µm in diameter, and the laser intensity profile was homogeneous and “flat-top”, due to the use of a fiber-optical illumination system similar to the one previously described.17 Depending on the intensity of analyte signals, between 20 and 100 spectra were summed for each fraction. Generally, the instrument settings were chosen for high detection sensitivity, i.e., high laser fluency and a high postacceleration voltage of about 10 kV. The latter was accompanied by the onset of the production of secondary ions at the postacceleration dynode and led to some peak broadening for ions above 1000 Da. The mass spectra were obtained using the software ULISSES (Chips at Work, Bonn, Germany). Each MALDI-MS spectrum was processed with a background subtraction involving median filter and smoothed with a box smoothing similar to a moving average. The median filtering and moving average were performed using the software IGOR (Wavemetrics, Lake Oswego, OR). Postsource Decay. PSD analysis was carried out with a Voyager-DE-STR time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA) in a reflectron positive ion mode. The total accelaration voltage was 20 kV, and 100-ns delay time was used. Between 20 and 30 single scans were summed for each PSD window. RESULTS AND DISCUSSION The combination of microcolumn LC and MALDI-MS allowed the separation and mass identification of peptides present in a single neuron. Figure 2 shows a three-dimensional (3D) plot of the peptide signals obtained from a single VD1 neuron after (17) Dreisewerd, K.; Schuerenberg, M.; Karas, M.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1995, 141, 127-148.

Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1849

Figure 2. 3-D plot of neuropeptides in a single VD1 neuron, fractionated by microcolumn LC and analyzed by MALDI-MS. The microcolumn effluent mixed with DHB matrix solution was obtained every 60 s, producing fraction volume sizes of approximately 145 nL. After drying, the samples with spot sizes of approximately 1 mm in diameter were analyzed by MALDI-MS. Only spectra with clearly detectable analyte signals are shown. The synthetic peptide eluting in fractions 26 and 27 was added to the cell supernatant before injection into the microcolumn. The numbers at the intensity bars reflect the (different) absolute signal intensities in the mass spectra for different sets of fractions. Peptides derived for the VD1 precursors, i.e., R1, R2, and β, are indicated, as well as the SCPs derived from the SCP precursor. The peptides C, D, and E are the putative modified forms of R2 peptide. The novel putative peptides are labeled F, G, and I, respectively. For larger peptides, doubly charged molecular species (M + 2H)2+ in addition to the singly charged species are detected and are also indicated. Matrix: 2,5-DHB.

Figure 3. Direct MALDI-MS spectrum of a VD1 neuron. After dissection from the snail brain, the cell was directly placed on target in a 0.5-µL drop of DHB matrix. See Figure 2 and text for explanation of the peak labels.

separation by the reversed-phase packed microcolumn and detection by MALDI-MS using DHB as matrix. Only spectra from chromatographic fractions with clearly detectable signals are shown in the figure. The figure reveals a series of neuropeptides with molecular masses between 1000 and 7000 Da, which are separated into 17 fractions by the microcolumn LC run, and the synthetic peptide of 1376 Da added before injection into the LC 1850 Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

column. The synthetic peptide is used as an internal standard to assess the overall performance of the sample preparation, LC fractionation, and MS detection. Previous molecular biology studies have revealed that the VD1 neuron synthesizes pre-mRNA that is alternatively spliced to give two preprohormones, the VD1 preprohormones A and B, that differ in a single peptide domain, the so-called R-peptide domain.18

Figure 4. Postsource decay MALDI-MS spectra. (A) PSD analysis of native SCPb (pQNYLAFPRMamide) from a single VD1 neuron after microcolumn LC fractionation. (B) PSD analysis of synthetic SCPb. N- and C-terminal fragment ions of the a, b, and y series are indicated. Matrix: 4-HCCA.

The precursors are expected to be processed into at least four mature peptides, the δ peptide (MW 1159), the R1 (MW 2401) and R2 peptides (MW 2996), respectively, for the two different precursors, and the β peptide (MW 6375). Figure 2 reveals that the δ peptide eluted in an early fraction. The β peptide was found in fractions 33 and 34, whereas both the R1 and R2 peptides were found in fractions 24 and 25. In addition to the R2 peptide, three molecules with mass differences one (molecule C), two (D) and three (E) times 379 mass units higher than R2 peptide were detected in fractions 23-25. Preliminary studies indicate that these molecules, with molecular masses of 3374, 3753, and 4132 Da, probably constitute modified forms of the R2 peptide.19 The nature of the modification is still unknown and is currently under investigation. Neuropeptides derived from other distinct precursors were also observed. Figure 2 shows that the peptides with a protonated (18) Bogerd, J.; van Kesteren, R. W.; Heerikhuizen, H.; Geraerts, W. P. M.; Joosse, J. Cell. Mol. Neurobiol. B 1993, 13, 123-126. (19) Jime´nez, C. R. Ph.D. Thesis, Vrije Universiteit Amsterdam, 1997.

mass of 1041.1 and 1122.4 Da correspond to the small cardioactive peptides (SCP) a (SGYLAFPRMamide) and b (pQNYLAFPRMamide), respectively. Furthermore, three more novel molecules (peptides) of masses of 6025 (peptide F), 6455 (peptide G), and 3970 Da (peptide I) were also found. Their identities are currently under investigation. The same set of neuropeptides detected in the microcolumn HPLC fractions is also found from direct singlecell MALDI-MS analysis on a VD1 neuron (Figure 3). The combination of both techniques can, therefore, allow the rapid screening of the entire peptide profile of a single cell and the chromatographic separation of the peptides into nearly pure microfractions for subsequent analysis or further treatment. The ability to obtain sequence information from a native peptide in a single cell by mass spectrometric means was demonstrated by PSD-MALDI-MS analysis. To characterize the peptide, another single VD1 neuron was processed as described in the Experimental Section. Instead of DHB, 4-HCCA was used as the matrix in this case because of the higher tendency of 4-HCCA to induce intense metastable fragmentation of the analyte.20 Since Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1851

the amino-terminal glutamine of SCPb has been shown to be modified to pyroglutamate in VD1, it would not be amenable to Edman sequencing. Therefore, SCPb was chosen to demonstrate the ability to sequence this peptide in the VD1 neuron by PSD analysis. The fraction that contained the peptide with a mass corresponding to that of SCPb was used for PSD analysis. The a, b, and y ion series were clearly detected (Figure 4A). The fragmentation patterns for the SCPb molecule present in VD1 are comparable to those for synthetic SCPb (Figure 4B). Furthermore, the resolutions for the parent and the fragment ions between those of native and synthetic peptides are very similar. In conclusion, we show for the first time that peptides contained in single neurons can be fractionated by microcolumn LC and detected by MALDI-MS analysis. The major advantages of such an approach for MS single-cell peptide profiling are as follow: (1) The fidelity of the assignment of the peptide identities, based on peptide elution positions from the LC column and their measured masses, is higher than that obtained with the measurement of the peptide molecular masses alone. (2) In the case of the detection of novel peptides or for further confirmation of a peptide species, PSD analysis can be performed. In the analysis of complex mixtures the previous microcolumn LC fractionation step brings about “cleaner” mass spectra, eliminating the stringency for very high precursor ion selection capabilities of the PSD analysis system to a great extent. To characterize larger peptides that cannot be identified by PSD (or tandem MS), on-target enzymatic digestion of the fractionated molecule may be performed. The smaller product peptides can then be further examined by PSD. (3) The microcolumn LC fractionation also opens up the field for down-scaling of peptide chemical techniques to the single-cell level. The peptides recovered can (in principle) (20) Karas, M.; Bahr, U.; Strupat, K.; Hillenkamp, F.; Tsarbopoulos, A.; Pramanik, B. N. Anal. Chem. 1995, 67, 675-679.

1852 Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

be used for other biochemical, immunochemical, and/or functional studies. (4) Furthermore, by the microcolumn LC step, the initial salt content of the sample is eliminated. In the present study, this advantageous feature may not be apparent because single neuron preparation contains low levels of salts. However, for other biological samples with a higher initial salt content, this can lead to an improved crystallization of the analyte/matrix mixture and enhancement in sensitivity and performance. Overall, the results show that microcolumn separations in combination with mass identification and structural elucidation by MALDI-MS and PSD hold many promising applications in the area of microanalysis of (complex) biological samples. Recently, an alternative approach employing ESI coupled on-line with capillary electrophoresis and using a Fourier transform ion cyclotron resonance mass spectrometer as mass analyzer have been applied to analyze proteins from single erythrocyte.21,22 ACKNOWLEDGMENT We acknowledge financial support from the Research Institute Neurosciences, Vrije Universiteit Amsterdam, to S.H., from the National Institutes of Health (GM 39515) to J.W.J., an apparatus grant from The Netherlands Research Organization (NWO) to K.W.L. and W.P.M.G., and the North Atlantic Treaty Organization for a grant (961152) to K.W.L. and J.W.J. We thank Mr. Ronald Baden for his technical assistance and Luigi Sanna for drawing some of the figures. Received for review July 30, 1997. Accepted January 16, 1998. AC9708295 (21) Hofstadler, A.; Severs, J. C.; Smith, R. D.; Swanek, F. D.; Ewing, A. G. Rapid Commun. Mass Spectrom. 1996, 10, 919-922. (22) Valaskovic, G. A.; Kelleher, N. L.; McLafferty, F. W. Science 1996, 273, 1199-1202.