Spurious serotonin dimer formation using electrokinetic injection in capillary electrophoresis from small volume biological samples

Abstract

One normally assumes that the analytical measurement process does not introduce spurious compounds. Capillary electrophoresis is a separation method frequently used for small-volume biological measurements. We demonstrate the potential for creating new peaks in a capillary electropherogram when using electrokinetic injections and illustrate the potential deleterious effects with biological samples involving serotonin and nitric oxide measurements. Specifically, when measuring the serotonin content from individual neurons using electrokinetic injections from 360 nL stainless steel vials, we detect a new peak that we identify as a serotonin dimer. We do not observe this peak when using hydrodynamic injections.

---

Introduction

The rapid separations, high separation efficiencies, and compatibility with small sample volumes make capillary electrophoresis (CE) a powerful analytical technique for identifying and quantifying chemical constituents of mass limited samples.^1–5 This is evidenced by the many applications that use CE, including inorganic ion determinations,^6,7 environmental analysis,^8,9 and food science.¹⁰ Indeed, the diminutive capillary dimensions and concomitant small volume requirements make CE a strong method for biological applications down to the subcellular level.

Many recent biological applications have focused on the gaseous neurotransmitter nitric oxide (NO). It has received an increasing amount of attention over the last decade owing to its importance in a variety of processes across a wide range of species. Since NO is a very reactive compound and is often localized in the central nervous system (CNS), there is potential for NO to chemically modify neurotransmitters in vivo. An unanswered question is whether nitration or other reactions between NO and indoles occurs in the CNS of animals. Nitration of an indole dipeptide has been reported in simple bacteria,¹¹ suggesting that this process is potentially important for many organisms. The resulting chemical changes in the affected cells could impact the neural function of the animal, as impairment of monoamine neurotransmitter systems can result in a variety of different brain disorders.¹² The marine molluscs Aplysia californica and Pleurobranchaea californica are often studied because of their large, identifiable neurons and simpler nervous system in comparison with mammalian counterparts and because they possess similar serotonin (5-hydroxytryptamine, 5-HT) neurotransmission processes relating to learning and memory.^13,14

Recent histochemical experiments have suggested that 5-HT and NO may be located in the same neurons of the marine mollusc Pleurobranchaea californica.^15,16 Neurons that appear to stain for both NO and 5-HT include the metacerebral cells and the G cells of the pedal ganglia. Since 5-HT is involved in arousal of feeding and locomotion in molluscs,^17,18 it is of interest to delineate the potential interactions of the neuromodulators NO and 5-HT and their effect on target cells.¹⁹† While earlier reports demonstrated the formation of biogenic amine derivatives in vitro,^20–22 evidence for these chemical derivatives produced in cellular media is elusive.

In this study, a custom-built CE-LIF system was used to characterize the indoles in several identified neurons. Specifically, by examining serotonergic neurons that either have been identified as NO producing or adjacent to NO producing cells, possible interactions between these two signalling systems can be probed. In one such cell, an unknown peak was detected and characterized as a 5-HT dimer, a major product in the reaction of 5-HT and NO. Is this peak actually present in the cell? We show that it can be a spurious peak caused by electrochemical reactions during the injection process. The effect of the antioxidant ascorbic acid on 5-HT dimerization is also considered. We demonstrate that, when employing electrokinetic injection in capillary electrophoresis, sampling bias can occur for many biologically relevant, easily oxidized species (such as 5-HT), and the bias often complicates biological analyses.

Experimental procedures

Reagents

The reaction buffer employed was 0.2 M acetate (0.96 g sodium acetate trihydrate and 0.17 mL concentrated glacial acetic acid in 50 mL ultrapure water [Millipore, Bedford, MA]), pH 5.0. The running buffer for electrophoretic separations was 50 mM borate (3.0 g boric acid, 9.2 g sodium borate decahydrate in 1.0 L ultrapure water), pH 8.8. The 5-HT used was 5-HT creatine sulfate complex. Artificial sea-water, which was used in the preparation of cellular samples, consisted of the following components: NaCl (420 mM), KCl (10 mM), CaCl₂ (10 mM), MgCl₂ (25 mM), MgSO₄ (25 mM), MOPS (10 mM), pH 7.5. All compounds were obtained from Sigma except sodium nitrite, which was obtained from Fluka.

Animals

Pleurobranchaea californica (Opistobranchia: Notaspidea) were obtained from the Pacific coast of California or were obtained from Sea-Life Supply (Sand City, CA), where they were kept in artificial sea-water at 15 °C.

Cell isolation and preparation

For dissection, animals were anesthetized by injecting an isotonic solution of MgCl₂ corresponding to half the animal's body weight. Individual ganglia of the CNS were removed and manually desheathed prior to single cell isolations. Single neurons were identified by location and relative size prior to isolation by microdissection. Isolated cell bodies were transferred to stainless steel nanovials containing 0.4 µL of the appropriate solution and rapidly frozen on dry ice for subsequent capillary electrophoresis with laser-induced fluorescence analysis.

Electrophoresis system

The CE system was assembled in our laboratory and has been described previously.²³ Briefly, electrokinetic or hydrodynamic injection was employed to precisely inject solutions from stainless steel nanovials. After separation, the eluent stream was directed into a sheath-flow cell whose flow was generated hydrodynamically. Excitation of the core stream was achieved with a frequency-doubled argon ion laser at 257 nm focused to a spot approximately 1 mm below the capillary outlet, inducing fluorescence. The fluorescence was collected and focused onto a spectrograph that disperses light in a wavelength-resolved fashion onto a liquid nitrogen-cooled CCD for multidimensional readout.

Hydrodynamic injection

Hydrodynamic injection was accomplished by manually lowering the waste reservoir to a height 4.9 cm below the injection plane for 30 s while the capillary was positioned in the sampled solution. This volume was nearly equivalent to that injected under electrokinetic conditions (2.1 kV for 10 s). However, due to a local vacuum effect of the sheath flow cell, slightly more sample was injected hydrodynamically relative to that expected during comparable electrokinetic injections. This extra amount was reproducible (<10% deviation in signal intensity from the calculated average), and the hydrodynamic injection method still yielded a linear calibration curve (R² > 0.99) over a wide concentration range of analytes investigated.

Reconstitutive electrokinetic injection

Reconstitutive electrokinetic injection was performed by placing 400 nL of a concentrated solution into a nanovial and injecting ∼2 nL of solution. After the separation was complete (<15 min), the nanovial was dry, and 400 nL of pH 8.8 borate buffer was used to reconstitute the analytes from the nanovial walls. Electrokinetic injection was performed again and the process was repeated.

Reactions with NO

Saturated NO solutions were prepared by bubbling NO gas first through two columns of 5 M sodium hydroxide solution and then either directly into the solution of interest or into a water solution until saturation to allow for the combination of the saturated aqueous NO solution with the solution of interest. Alternatively, NO can form under acidic conditions from sodium nitrite. In this case, the addition of a standard sodium nitrite solution signaled the start of the reaction.

5-HT dimer, nitroso-5-HT and nitro-5-HT were prepared according to established procedures²⁴ and separated by HPLC with on-line electrospray ionization mass spectrometry (ESI-MS) for identification. A selective 5-HT dimer synthesis was performed in the following manner. A 1.5 mM L-ascorbic acid solution in 0.2 M pH 5.0 acetate buffer was created, and part of this solution was used to dissolve enough 5-HT to make the 5-HT concentration 0.5 mM. Finally, to start the reaction, part of this 5-HT/L-ascorbic acid solution was used to dissolve enough sodium nitrite to yield a final concentration of 0.5 mM sodium nitrite. The solution was maintained at 2–8 °C for approximately 1 week, and the reaction was followed with CE-LIF. After this time, all 5-HT was consumed and only 5-HT dimer was present by CE-LIF and ESI-MS (m/z = 351). ¹H NMR (500 MHz) confirms that the dimer was of similar structure to that previously reported.²⁴ The dimer is only stable for a short period of time, even in the presence of antioxidants, so it is critical for the analysis to be performed shortly following synthesis.

Results

5-HT dimer synthesis and characterization

The in vitro reactions of 5-HT and NO produced three major compounds (Scheme 1) whose prevalence depended upon the reaction conditions. On-line HPLC-ESI-MS confirmed the presence of all three compounds under previously reported conditions.²⁴ Under certain conditions (see Experimental), the symmetrical 5-HT dimer could be selectively synthesized, and the dimer matched the fluorescence properties and electrophoretic migration rate of the unknown compound. As demonstrated by standards (data not shown), the nitro and nitroso groups quenched the fluorescence of the aromatic indoleamine; therefore, the 5-HT dimer was used to gauge the progress of the reaction when employing CE-LIF. The fluorescence emission spectrum of the dimer was noticeably red shifted with respect to 5-HT and was further distinguished on the basis of electrophoretic migration time. These parameters were used for identification in cellular samples.

Scheme 1 Indole reaction with nitric oxide.

5-HT dimer in cellular samples

A single cell assay of an A-cluster neuron from the cerebropleural ganglion was examined by CE with electrokinetic injection for indole contents, and a large quantity of an unknown indole was detected (Fig. 1). This unknown peak had fluorescence properties and a migration time matching that of the 5-HT dimer, suggesting that 5-HT and NO might be co-localized and react in this cell. However, reproducing this result using hydrodynamic injection (see below) with samples from additional animals failed to confirm this result. Furthermore, due to their positive serotonergic and/or NADPH-diaphorase immunostaining,^15,16 we also examined metacerebral cells, pedal G cells, and the As cluster cells. Even when identified serotonergic cells were lysed under acidic conditions or bathed in the presence of the NO donor SIN-1, 5-HT dimer was not observed. The large sample size of serotonergic single cells examined to date on this CE-LIF system (N > 15) suggested that the dimer initially found in the one cell sample was an outlier. Previous work has both demonstrated that 5-HT dimer could be produced electrochemically and investigated the electrochemical mechanisms of its formation in great detail.^25–28 It appears that the electrokinetic injection process itself may be responsible for the potential false positive detection of 5-HT dimer.

Fig. 1 The electropherogram of an A cluster neuron reveals the presence of 5-HT and an unknown with indole-like fluorescence properties.

Hydrodynamic injection alleviates electrochemical 5-HT dimer formation

Is the occasional detection of the 5-HT dimer an artifact? Using 0.5 mM 5-HT in 50 mM borate buffer, little if any 5-HT dimer was detected from an initial injection. However, when 0.5 mM 5-HT was sampled under high salt conditions (a 0.2 M pH 5.0 acetate buffer solution of lower ionic strength than that of sea-water but greater than that of CE running buffer), as are normally found in biological tissues, the first electrokinetic injection often resulted in the detection of the 5-HT dimer (Fig. 2). This indicated that 5-HT dimer can be electrochemically formed from the electrokinetic injection of nanoliter-volume samples in CE. Furthermore, studies employing reconstitutive electrokinetic injection resulted in an increase in the 5-HT dimer peak forming along with several other unidentified peaks. Concurrently, 5-HT signal intensity subsequently decreased.

Fig. 2 Electrokinetic (light trace) versus hydrodynamic (dark trace) injection of a 0.5 mM 5-HT solution in 0.2 M pH 5.0 acetate buffer (no reconstitution). Inset: magnification of the above showing that 5-HT dimer is formed electrochemically in the injection vial via electrokinetic injection but not via hydrodynamic injection.

As a result, a hydrodynamic injection method was developed to eliminate the unwanted electrochemistry. To test the effectiveness of this method, a concentrated 5-HT sample was prepared (in 0.2 M pH 5.0 acetate buffer solution) and was repeatedly sampled with both injection methods (Fig. 3). Even after three nanovial injections using the hydrodynamic method, no 5-HT dimer was observed, and the 5-HT signal intensity only decreased 50% from the original signal intensity. In comparison, after three nanovial electrokinetic injections of a fresh aliquot of the identical starting solution, a strong dimer signal was observed, and the 5-HT signal decreased 90% from the initial injection. Additional work demonstrated that air oxidation of 5-HT was not responsible for the dimer formation but could result in a mild decrease in 5-HT signal intensity (data not shown).

Fig. 3 Repeated injections of a concentrated 5-HT solution (1 mM) in 0.2 M pH 5.0 acetate buffer with 50 mM pH 8.8 borate buffer, and the effect of the injection method. (A) The third electrokinetic injection repetition results in a 90% decrease in initial 5-HT signal and the formation of 5-HT dimer. (B) The third hydrodynamic injection repetition results in only a 50% decrease in initial 5-HT signal with no detectable dimer formation.

In vitro reaction between 5-HT and NO

After an injection methodology was established that did not result in spurious 5-HT dimer formation, the reaction between 5-HT and NO could be more thoroughly investigated with respect to the 5-HT dimer. The dimer increased concomitantly with an increase in reaction time or NO concentration (data not shown). Since the 5-HT dimer could indicate a cellular co-localization of 5-HT and NO (reaction of hydroxyl and superoxide radicals in aqueous 5-HT solution fails to produce appreciable amounts of 5-HT dimer²⁵), it is important to consider other chemicals that might colocalize with 5-HT and NO and their effect on the reaction. One such example is a biologically relevant antioxidant that might protect cells from higher order nitrogen oxides, which are otherwise toxic to living cells. Ascorbic acid was chosen because of its high concentration (∼1.5 mM) in the serotonergic and NADPH-diaphorase positive metacerebral cells (MCCs) of Aplysia.²⁹ We examined the effect of ascorbic acid on 5-HT/NO reactions under acidic conditions to simulate the vesicular environment, where 5-HT is sequestered in high concentrations in neurons; these areas normally possess a pH of approximately 5. Fig. 4 shows the effect of ascorbic acid on the proportion of dimer signal intensity versus that of 5-HT. A reduction in dimerization rate occurred with an increasing amount of antioxidant present in the original solution. Furthermore, at a high ascorbic acid concentration (25 mM), no dimerization occurred. By the end of day four in the absence of ascorbic acid, both 5-HT and its dimer completely decomposed in solution.

Fig. 4 Effect of ascorbic acid on the formation of 5-HT dimer. Initial concentration of 5-HT was 0.5 mM, sodium nitrite was 1.0 mM, and the reactions proceeded at 4 °C in 0.2 M acetate buffer, pH 5.0.

Discussion

Based on histochemistry, it appears that 5-HT and NO are colocalized in several neurons in the CNS. Do these messengers chemically interact? While we do not detect products confirming their reaction, amounts of dimer might form that are below the detection limits. Unfortunately, as no powder standard is available for the 5-HT dimer, the exact detection limits are difficult to quantify. As there is up to 2 mM NO₂⁻ present in these marine mollusc cells,³⁰ and similar amounts of ascorbic acid,²⁹ the results in Fig. 4 suggest that dimer formation is still possible in the presence of equimolar ascorbic acid in serotonergic vesicles. Alternatively, the three compounds could be differentially packaged within a cell, requiring subcellular analysis of individual 5-HT vesicles for an accurate understanding of these processes. However, since most of the cell samples were lysed in a pH 8.8 buffer (preventing nitrite back-conversion to NO), little extraneous reaction should occur between 5-HT and NO after isolation and lysing, even though NO is freely diffusible under these conditions. Even when lysed in an acidic buffer (which would promote NO regeneration from nitrite ion) or when intact cells were bathed in artificial sea-water in the presence of the NO donor SIN-1, no dimer was detected. This suggests that it is unlikely for 5-HT dimer to form in the presence of these particular cell samples, possibly owing to the abundance of cellular antioxidants. The cell sample found to contain 5-HT dimer likely was the result of an unusual absence of such antioxidants, an abundance of oxidants, the particular state of the individual animal, or an artifact caused by the electrokinetic injection.

5-HT dimer could form from 5-HT in vitro using electrophoretic injection under salty conditions, and this effect was successfully avoided by employing hydrodynamic injection. This is especially significant for cellular samples, given that cellular environments typically require high osmolarities. While lysing of the cell normally occurs in a lower ionic strength buffer, the environment of the cell, coupled with its contents, makes the final nanovial contents relatively salty.

These results indicate that care must be taken to ensure that spurious peaks are not unintentionally fabricated by the process of injection. The aberrant presence of 5-HT dimer in the cell sample warrants such caution. In fact, some sampling artifacts have already been associated with CE,³¹ including the depletion of 5-HT via (presumed) electrochemical or electromigratory consumption and the formation of unknown degradation products (see Fig. 5C–5E in ref. 31), as well as the significant differences observed when quantifying a standard 5-HT solution as a function of capillary-electrode distance.

While altered neuronal signaling could occur upon the reaction of 5-HT and NO in a single cell or synapse, detection of its products requires sensitive and careful analysis. This issue is relevant to other biomolecules as well, since other easily oxidized species such as dopamine could suffer similar fates with electrokinetic injection. As future analytical advances allow measurement of ever-smaller quantities, observation of the native chemical conversion of biomolecules, in correlation with its resultant physiological effects, will become a reality.

Acknowledgements

We would like to acknowledge Sarah Sheeley, Jennifer Jakubowski, Dimuthu Jayawickrama, Leonid Moroz and Won-Suk Kim for their assistance throughout the work, as well as the SCS Mass Spectrometry Center. The financial assistance of the USA National Institutes of Health and the National Science Foundation through grants NS31609 and CHE0400768 are gratefully acknowledged.

References

1. J. N. Stuart and J. V. Sweedler, Anal. Bioanal. Chem., 2003, 375, 28–29.
2. F. N. Lamari, M. Militsopoulou, T. N. Mitropoulou, A. Hjerpe and N. K. Karamanos, Biomed. Chromatogr., 2002, 16, 95–102.
3. H. Watzig and S. Gunter, Clin. Chem. Lab. Med., 2003, 41, 724–738.
4. S. C. Beale, Anal. Chem., 1998, 70, 279R.
5. H. J. Issaq, Electrophoresis, 2001, 22, 3629–3638.
6. A. R. Timerbaev and O. A. Shpigun, Electrophoresis, 2000, 21, 4179.
7. J. S. Fritz, J. Chromatogr. A, 2000, 884, 261.
8. W. C. Brumley, LCGC, 1995, 13, 556.
9. D. Martínez, M. J. Cugat, F. Borrull and M. Calull, J. Chromatogr. A, 2000, 902, 65.
10. T. Watanabe and S. Terabe, J. Chromatogr. A, 2000, 880, 311.
11. J. A. Kers, M. J. Wach, S. B. Krasnoff, J. Widom, K. D. Cameron, R. A. Bukhalid, D. M. Gibson, B. R. Crane and R. Loria, Nature, 2004, 429, 79.
12. J. P. Kiss, Brain Res. Bull., 2000, 52, 459–466.
13. C. H. Bailey, M. Chen, F. Keller and E. R. Kandel, Science, 1992, 256, 645–649.
14. F. Noel, C. Koumenis, M. Nunez-Regueiro, U. Raju, J. H. Byrne and A. Eskin, Proc. Natl. Acad. Sci. USA, 1994, 91, 150–154.
15. L. C. Sudlow, J. Jing, L. L. Moroz and R. Gillette, J. Comp. Neurol., 1998, 395, 466–480.
16. L. L. Moroz and R. Gillette, J. Comp. Neurol., 1996, 367, 607–622.
17. S. C. Rosen, K. R. Weiss, R. S. Goldstein and I. Kupfermann, J. Neurosci., 1989, 9, 1562–1578.
18. D. R. McPherson and J. E. Blankenship, J. Neurophysiol., 1991, 66, 1366–1379.
19. P. Fossier, B. Blanchard, C. Ducrocq, C. Leprince, L. Tauc and G. Baux, Neuroscience, 1999, 93, 597–603.
20. M.-L. de la Breteche, C. Servy, M. Lenfant and C. Ducrocq, Tetrahedron Lett., 1994, 35, 7231–7232.
21. F. Shintani, T. Kinoshita, S. Kanba, T. Ishikawa, E. Suzuki, N. Sasakawa, R. Kato, M. Asai and T. Nakaki, J. Biol. Chem., 1996, 271, 13561–13565.
22. M. d'Ischia and C. Costantini, Bioorg. Med. Chem., 1995, 3, 923–927.
23. R. R. Fuller, L. L. Moroz, R. Gillette and J. V. Sweedler, Neuron, 1998, 20, 173–181.
24. B. Blanchard, M. Dendane, J.-F. Gallard, C. Houee-Levin, A. Karim, D. Payen, J.-M. Launay and C. Ducrocq, Nitric Oxide Biol. Chem., 1997, 1, 442–452.
25. M. Z. Wrona and G. Dryhurst, Chem. Res. Toxicol., 1998, 11, 639–650.
26. M. Z. Wrona and G. Dryhurst, J. Org. Chem., 1987, 52, 2817–2825.
27. M. Z. Wrona and G. Dryhurst, J. Org. Chem., 1989, 54, 2718–2721.
28. M. Z. Wrona and G. Dryhurst, Bioorg. Chem., 1990, 18, 291–317.
29. W. S. Kim, R. L. Dahlgren, L. L. Moroz and J. V. Sweedler, Anal. Chem., 2002, 74, 5614–5620.
30. L. Cruz, L. L. Moroz, R. Gillette and J. V. Sweedler, J. Neurochem., 1997, 69, 110–115.
31. R. R. Fuller and J. V. Sweedler, Anal. Chem., 1999, 71, 4014–4022.

---

Footnote

† One such study involved the bath incubation of Aplysia californica buccal ganglia cells with 5-HT and an NO donor.¹⁹ Concomitantly, both neuromodulators decreased acetylcholine release to a lesser extent than bath application of either 5-HT or NO alone.¹⁹ Thus, a chemical interaction between NO and 5-HT in vivo may alter neuronal activity.

---