Quantitative analysis of serotonin secreted by human embryonic stem cells-derived serotonergic neurons via pH-mediated online stacking-CE-ESI-MRM

Abstract

A CE-ESI-MRM-based assay was developed for targeted analysis of serotonin released by human embryonic stem cells-derived serotonergic neurons in a chemically defined environment. A discontinuous electrolyte system was optimized for pH-mediated online stacking of serotonin. Combining with a liquid–liquid extraction procedure, LOD of serotonin in the Krebs’-Ringer’s solution by CE-ESI-MS/MS on a 3D ion trap MS was 0.15 ng/mL. The quantitative results confirmed the serotonergic identity of the in vitro developed neurons and the capacity of these neurons to release serotonin in response to stimulus.

Serotonin, also called 5-hydroxytryptamine (5-HT), is an important biogenic amine neurotransmitter involved in the regulation of a wide variety of behaviors and physiological functions such as mood, sleeping, circadian rhythmicity, and neuroendocrine functions [1]. It is mainly synthesized from L-tryptophan in the serotonergic neurons by the enzymes L-tryptophan-5-monooxygenase and aromatic L-amino acid decarboxylase, and primarily stored in vesicles and released into synaptic cleft by exocytosis [1]. To study the human central serotonergic neurons that mainly located in the Raphe Nucleus of the hindbrain, Lu et al. have developed an in vitro model system to induce human embryonic stem cells (hESCs) to differentiate into an enriched population of serotonergic neurons [2]. In order to confirm the in vitro functionality of the hESCs-derived serotonergic neurons, the secretion of serotonin from these cultured neurons must be measured and quantified. Hence, we developed a CE-ESI-multiple reaction monitoring (MRM) assay to quantify the serotonin released from the neurons under normal and depolarizing conditions.

Analytical techniques, including ELISA [3], reversed phase LC [3–9] or CE [9–17] coupled with UV [11, 13, 15, 16] or fluorescence detection [12, 14], electrochemical detection [10, 18], or MS detection 3–9, [17] have been applied for determination of serotonin in different types of biological matrices such as serum [18], plasma [3, 8, 17], whole blood [11], urine [9, 13, 15, 16], brain tissue [4, 10, 14], plant tissues [7]. We choose CE-ESI-MS/MS as the analytical platform for quantification of the low nanograms amount of serotonin released by mature hESCs-derived serotonergic neurons because CE provides rapid and efficient separation and ESIMS/MS detection provides the required specificity without chemical derivatization. One potential pitfall of CE analysis is the small sample loading volume (<1% of the capillary column volume, typically 10~20 nanoliters) that usually limits the detection sensitivity. To avoid this constraint, several on-column stacking methods, such as field-amplified sample stacking, pH-mediated stacking or dynamic pH junction, transient isotachophoresis, chromatographic stacking or sweeping, have been applied by researchers to increase the sample loading amount without sacrificing the separation efficiency [19,20]. Among these techniques, pH-mediated stacking is easy to perform, suitable for weak ionic species such as serotonin, and compatible with ESI-MS detection. Abersold and coworkers first demonstrated the online concentration effect of pH-mediated stacking by loading zwitter-ionic peptides dissolved in alkaline solution into a capillary column prefilled with acidic BGE [21]. This strategy of employing discontinuous pH electrolyte system was further extended to enhance CE detection of a class of molecules with pH-dependent electrophoretic mobility [22], such as amino acids, peptides, proteins, and nucleosides. In this work, we employed the pH-mediated stacking method for improved CE-ESIMS/MS analysis of serotonin and accurately quantified the trace-level amount of serotonin secreted by the serotonergic neurons.

Human ESCs (H9, passages 25–40, originally from Wi-Cell) was induced to differentiate into serotonergic progenitors after around 3 weeks of culturing according to a protocol described by Lu et al. [2]. When the neurons matured around the fourth weeks, the cell culture medium was removed and the neurons were sequentially treated with Krebs’-Ringer’s solution and depolarizing Krebs’-Ringer’s solution. The serotonergic neurons were first washed with Krebs’-Ringer’s solution containing 130 mM NaCl, 3 mM KCl, 2 mM CaCl₂, 0.8 mM MgSO₄, 10 mM glucose, and 20 mM HEPES (pH 7.4), and then washed with depolarizing Krebs’-Ringer’s solution with increased KCl concentration of 50 mM. Around 1.6 mL solution was collected after each washing step for subsequent serotonin measurement. The CE-ESI-MRM experiments were carried out with an Agilent HP 3D-CE system coupled with an Agilent 1100 MSD Trap mass spectrometer (Agilent Technologies, Inc. Santa Clara, CA), and the CE-ESI-MS interface was originally developed by Maxwell et al. [23, 24]. More details of the CE-MS setup and parameters are included in supplemental information.

Quantification of the low parts per billion level of serotonin released by the hESCs-derived serotonergic neurons is challenging. This biogenic amine does not bare charge in basic solution, but is positively charged in acidic BGE and migrates toward the cathode under electric field. Therefore, pH-mediated online stacking can be applied to increase the sample loading volume. This process is schematically illustrated in Fig. 1. First, the capillary column is prefilled with acidic BGE and a long plug of analyte dissolved in basic solution is injected from the inlet (Fig. 1A). Once the electric field is applied, protons from the inlet buffer reservoir sweep from the rear end of the sample plug and gradually neutralize the hydroxyl ions in the sample plug. The serotonin molecules at the rear end of the sample band then become positively charged and start to move toward the cathode (Fig. 1B). During the stacking process, the charged serotonin molecules at the rear end further focus into a narrower band and migrate toward the outlet. However, at the same time, some uncharged serotonin molecules at the front end of the sample plug could diffuse into the acidic BGE and then accelerate toward the outlet (Fig. 1C). In order to prevent the diffusion induced band broadening during the focusing process, a short plug of basic solution serving as a spacer between the sample plug and acidic BGE can be injected into the capillary column before sample loading (Fig. 1D). With the addition of the short spacer, the analytes can be focused into a sharp band and migrate toward the cathode end after completion of the pH-mediated stacking process (Fig. 1E).

Figure 1.

Schematic illustration of pH-mediated online stacking process of serotonin.

(A) A long plug of sample dissolved in basic solution is injected into a capillary column prefilled with acidic BGE. (B) When electric field is applied, protons from the inlet buffer reservoir migrate into the sample plug from the anodic side. Protonated serotonin [BH]⁺ at the anodic side migrate toward the cathode, and the sample plug starts to focus into a narrower band. (C) Serotonin molecules at the front end of the sample boundary could diffuse into the acidic BGE and become protonated. (D) A short plug of basic solution, serving as spacer between the sample plug and acidic BGE, is injected into the capillary column before sample loading. In this way, uncharged serotonin molecules are prevented from diffusing into acidic BGE before focusing process completed. (E) With the addition of a spacer, the focusing process completes and the concentrated sample move toward the outlet as a narrow band.

Injection lengths of the sample plug and spacer plug for the pH-mediated stacking conditions were optimized for CE-ESI-MS/MS analysis of serotonin. A 50% MeOH and 0.2% formic acid was used as the separation BGE and modifier solution. The electropherograms of the transition channel m/z 177 → m/z 160 were overlaid to compare the results of applying different sample loading methods (Fig. 2). When serotonin standard was dissolved in the same buffer as BGE, only a very small volume (50 mbar for 20 s) could be loaded onto the column for the generation of a Gaussian peak (Fig. 2A). When the sample was dissolved in 50% MeOH with 1% NH₃, the sample concentration could be further reduced because a longer sample plug (50 mbar for 100 s) could be injected (Fig. 2B). However, a small shoulder appeared at the front end of the major peak due to sample diffusing into the acidic BGE during the stacking process as illustrated in Fig. 1C. To address this problem, a short plug of 1% NH₃ was injected as spacer between the acidic BGE and sample plug. With 50 mbar and 15 s injection of 1% NH₃ solution, the peak shoulder was alleviated (Fig. 2C); by increasing the pressure injection time of 1% NH₃ to 25 s, the basic spacer could almost eliminate formation of the small peak shoulder (Fig. 2D). To further improve the detection sensitivity, even longer injection lengths of the spacer and sample plug (30 s for the spacer, 130 s and 160 s for the sample) were attempted (Fig. 2E, F). It was found that 50 mbar pressure injection of 1% NH₃ for 30 s followed by 50 mbar pressure injection of sample dissolved in 50% MeOH and 1% NH₃ for 130 s generated optimal sensitivity as well as peak shape (Fig. 2E). Since only the portion of column filled with BGE was used for separation, the sample loading amount was not further increased in case effective separation length of the column was not long enough. This optimized pH-mediated online stacking method improved the detection sensitivity of serotonin by approximately 4.5-fold and was adopted for the absolute quantification of serotonin released by the serotonergic neurons. To further improve the detection sensitivity, ultrasonication-assisted liquid–liquid extraction (see supplemental information) was employed for desalting and preconcentration prior to CE-ESI-MS/MS analysis.

Figure 2.

Extracted ion electropherograms of serotonin standard with different sample and basic spacer injection lengths.

CE conditions: 65 cm, 50 μm id; bare fused silica capillary; BGE, 50% MeOH and 0.2% formic acid; voltage, 30 kV; modifier solution, 50% MeOH and 0.2% formic acid infused at 300 nL/min. MS conditions: nebulizer gas disabled; dry gas, 300°C, 1.0 L/min; MRM transition channel, m/z 177 → m/z 160; collision energy, 0.76 V. Sample loading methods: (A) 1.5 μM serotonin dissolved in 50% MeOH and 0.2% formic acid, 50 mbar pressure applied for 15 s; (B) 1 μM serotonin dissolved in 50% MeOH and 1% NH₃, 50 mbar pressure applied for 100 s; (C) 1% NH₃ in H₂O was injected at 50 mbar pressure for 15 s, followed by 50 mbar pressure injection of 1 μM serotonin dissolved in 50% MeOH and 1% NH₃ for 100 s; (D) 1% NH₃ in H₂O was injected at 50 mbar pressure for 25 s, followed by 50 mbar pressure injection of 1 μM serotonin dissolved in 50% MeOH and 1% NH₃ for 100 s; (E) 1% NH₃ in H₂O was injected at 50 mbar pressure for 30 s, followed by 50 mbar pressure injection of 1 μM serotonin dissolved in 50% MeOH and 1% NH₃ for 130 s; (F) 1% NH₃ in H₂O was injected at 50 mbar pressure for 30 s, followed by 50 mbar pressure injection of 1 μM serotonin dissolved in 50% MeOH and 1% NH₃ for 160 s.

Absolute quantification of serotonin secreted by the serotonergic neurons were achieved by constructing a calibration curve with a series of concentrations of serotonin standard and constant concentration of d₄-serotonin (internal standard) spiked into Krebs’-Ringer’s solution (see more details in the Supporting Information Fig. 2, 3). The limit of detection of serotonin, calculated as 3×SD_blank/slope of calibration curve, was determined as 0.15 ng/mL using this pH-mediated stacking CE-ESI-MRM strategy combining with liquid–liquid extraction. Figure 3 compares the concentrations of serotonin secreted by three groups of serotonergic neurons in K-R solution and depolarizing K-R solution. In response to the stimulus from a high potassium ion concentration environment, the serotonergic neurons were depolarized and the serotonin release was increased by an average of 1.8-fold, which confirms the functionality of the in vitro differentiated hESCs-derived serotonergic neurons. As the biological control experiment, serotonin in the releasate from the dorsal hindbrain derived-neurons under both depolarizing (Supporting Information Fig. 3E) and normal conditions (Supporting Information Fig. 3F) was not detectable.

Figure 3.

Concentration of serotonin secreted by the three groups of hESCs-derived serotonergic neurons in Krebs’-Ringer’s solution (red) and depolarizing Krebs’-Ringer’s solution (blue). Error is estimated from standard deviation of three technical measurements.

In conclusion, this study aims to establish a rapid and effective CE-ESI-MRM assay for quantitative analysis of the low ppb amount of serotonin released by hESCs-derived serotonergic neurons. The matrix in the neuron washing solution was removed by liquid–liquid extraction. pH-mediated online stacking further improved the overall detection sensitivity of serotonin by CE-ESI-MRM. This method was successfully applied to determination of serotonin released by the serotonergic neurons under normal and depolarizing conditions and the results confirmed the identity and capacity of the in vitro cultured neurons. It was observed that high concentration of potassium ions in the extracellular environment boosts the secretion of the neurotransmitter serotonin from the hESCs-derived serotonergic neurons in response to the stimulus.

Abbreviations:

hESC

human embryonic stem cell

MRM

multiple reaction monitoring