The Application and Validation of HybridSPE-Precipitation Cartridge Technology for the Rapid Clean-up of Serum Matrices (from Phospholipids) for the Clinical Analysis of Serotonin, Dopamine and Melatonin
Phospholipids have been shown to cause matrix effects particularly in liquid chromatography–mass spectrometry (LC–MS) analysis of small molecules. This results in suppression of the analyte signal. This study provides a versatile validated method for the analysis of serotonin in serum along with dopamine and melatonin using LC–MS/MS. It utilises HybridSPE-Precipitation cartridges for the clean-up of serum samples. This technology involves a simple protein precipitation step together with a fast and robust SPE method that is designed to remove phospholipids. Serotonin and dopamine are major neurotransmitters in the brain which affect various functions both in the brain and in the rest of the body. Melatonin plays an important role in the regulation of circadian sleep–wake cycle. Good linear calibrations were obtained for the multiplex assay of analytes in serum samples (0.021–3.268 μmol L−1; R2 = 0.9983–0.9993). Acceptable intra- and inter-day repeatability was achieved for all analytes in serum. Excellent limits of detection (LOD) and limits of quantitation (LOQ) were achieved with LODs of 3.2–23.5 nmol L−1 and the LOQs of 15.4–70.5 nmol L−1 for these analytes in serum. The sample clean-up procedure that was developed provided efficient recovery and reproducibility while also decreasing preparation time and solvent use. A sample storage protocol was established, this was achieved by investigation of sample stability under different storage conditions. Evaluation of matrix effects was also carried out and the influence of ion suppression on analytical results reported. This clean-up protocol was then applied to the analysis of clinical serum samples.
Plasma and serum have a complex matrix that contain dissolved proteins and other trace materials that can often cause ion suppression when using liquid chromatography–mass spectrometry methods. Some of the main substances that can cause ion suppression are usually ionic substances or components that can result in charge transfer processes in the gas phase or in the electrospray ionisation process . Phospholipids have been shown to cause high amounts of matrix effects. Phospholipids are the main component of cell membranes and the most significant types found in plasma are glycerophosphocholines and phosphatidylcholine .
The method described here is an LC–MS/MS method developed for the analysis of serotonin along with dopamine and melatonin. Serotonin (5-Hydroxytryptamine, 5-HT) is a neurotransmitter in the central and peripheral nervous systems of the body. It is involved in regulating blood pressure, smooth muscle contraction, pain sensation, it controls many brain functions and has been associated with sleep disruption and aggression. It has also been implicated in several pathological conditions, such as hypertension, migraine, depression, schizophrenia, anxiety, anorexia nervosa, dementia, and carcinoid syndrome [3–9]. Abnormal concentrations in plasma have been shown to reflect the serotonergic function in the central nervous system . Serotonin levels, synthesis, uptake sites and receptor binding are all higher in the developing brain compared with adult values, and levels decline before puberty. This developmental process has been reported to be disrupted in autistic children .
Melatonin (N-acetyl-5-methoxytryptamine) is formed from an alternative metabolic pathway of serotonin in the human pineal gland at night under normal environmental conditions . Melatonin plays an important role in the regulation of the circadian sleep–wake cycle. The concentration of melatonin in blood is changeable with the circadian rhythm, low around 10 pg mL−1 in the daytime and high around 25–120 pg mL−1 at night .
Dopamine (4-(2-aminoethyl)benzene-1,2-diol) is a neurotransmitter with its own system of neurons. It has been shown that serotonin can influence the dopaminergic system . A dramatic loss of dopamine (DA) in the substantia nigra is constantly observed in the post mortem brains of patients with Parkinson’s disease . The dopaminergic system has been reported to affect cognition, motor function, brain-stimulation reward mechanisms, eating and drinking behaviours, sexual behaviour, neuroendocrine regulation and selective attention . The chemical structure of the target analytes is shown as inserts in Fig. 5b.
A variety of sample preparation techniques are used for the clean-up of complex matrices such as plasma and serum. These include solid phase extraction (SPE) [9, 10, 15], protein precipitation [16–18], liquid–liquid extraction [19–21] and online SPE [17, 22]. Many of these methods involve the use of relatively large amounts of solvents, long preparation times and incomplete removal of matrix components that may interfere with the analyte signal. The method described herein is a two-step preparation technique that utilises a protein precipitation step with a HybridSPE-Precipitation cartridge  for a rapid, cost effective and robust clean-up protocol. These marketed cartridges are specifically designed for the removal of phospholipids, a significant source of matrix interference in blood samples. Pucci et al.  monitored glycerophosphocholines and lysophosphatidylcholines which represent 70 and 10 % of total plasma phospholipids compared with reported a drastic decrease in the presence of these phospholipids when using the HybridSPE-Precipitation cartridges compared with simple protein precipitation leading to reduced matrix effects during analysis. Ardjomand-Woelkart et al.  utilised the 96-well HybridSPE-Precipitation technique for quantitation of CYP450 substrates/metabolites in rat plasma. Recoveries ranged from 69.25 ± 10.68 to 102.38 ± 7.22 %. Jiang et al.  saw a considerable reduction in matrix effects when using HybidSPE-precipitaion over protein precipitation for the determination of carboplatin in human plasma. A matrix factor of ~1 was seen for HybridSPE in comparison to 0.01 for protein precipitation. Recoveries were still quite low at <35 %.
This is the first reported use of HybridSPE-Precipitation cartridges for the rapid and sensitive analysis of serotonin, dopamine and melatonin in serum samples.
Materials and Methods
Chemicals used in this study included formic acid purchased from Merck (Darmstadt, Germany), ammonium acetate and acetic acid were purchased from Sigma-Aldrich (Dublin). Hydrochloric acid and all HPLC grade solvents (methanol, water, and acetonitrile) were purchased from Labscan (Dublin, Ireland). Nitrogen gas was purchased from Irish Oxygen (Ireland). All standard materials including the Supelco HybridSPE-Precipitation cartridges were purchased from Sigma-Aldrich (Dublin, Ireland) and the deuterated compounds (serotonin-d4, dopamine-d4 and melatonin-d4) were purchased from CDN Isotopes (Quebec, Canada). Serum samples were obtained from the Irish Blood Transfusion Service (IBTS).
Standards (serotonin, dopamine, and melatonin) in solvent (water) were prepared from a stock mixture of all compounds (57, 65 and 43 μmol L−1 for serotonin, dopamine and melatonin) by serial dilutions. Deuterated internal standards (serotonin-d4 and melatonin-d4) were subsequently added with final internal standard concentrations of 1.11 and 0.85 μmol L−1, respectively.
Standards in serum matrix were prepared by spiking standards from the stock solution into 100 μL of 0.1 % bovine albumin in phosphate buffer solution (pH 7), vortexing and proceeding with sample extraction.
Sample preparation involved addition of internal standards (20 μL) into each serum sample (100 μL), the samples were vortexed followed by the addition of acetonitrile (300 μL) with 1 % formic acid, vortexed again and then centrifuged for 10 min (3,000×g). The supernatants (~400 μL each) were put through HybridSPE-Precipitation cartridges, which were then eluted with methanol containing 0.1 mM ammonium acetate (600 μL; Fig. 2-5), and then the samples were dried under nitrogen and reconstituted in water (100 μL). An aliquot (5 μL) was injected into the LC–MS/MS.
Study of the Stability of Serotonin in Serum
A whole blood sample (2 mL) was collected by venepuncture and separated into two 1 mL aliquots, designated sample #1 and sample #2. Sample #1 was left at room temperature for 6 h, then placed in a fridge (4 °C) overnight, then centrifuged at 3,000 rpm for 15 min and stored at 4 °C for 4 days. Sample #2 was centrifuged immediately and stored at −80 °C for 4 days. Aliquots of each sample were prepared as in “Materials and Methods” and analysed on each of the 4 days to monitor the stability of serotonin. The purpose of this stability study was to compare the IBTS storage procedure which is followed by the sample #1 procedure in comparison to freezing samples at −80 °C, following sample collection.
Liquid Chromatography–Multiple Tandem Mass Spectrometry (LC–MS/MS)
Optimised LC–MS/MS parameters
Mass [M + H]+
Collision energy (%)
Ion Suppression Post-Column Infusion Assessment
Post-column infusion experiments were conducted to determine ion suppression/enhancement effects using the approach described by Choi et al. . The setup used in these experiments involved the infusion of a standard during the analysis of a blank matrix solution. The standard mixture (2.8, 2.2 and 3.3 μmol L−1for serotonin, melatonin and dopamine, respectively), monitored by MRM, was introduced at 5 μL min−1 into the LC eluent. A blank matrix (LC grade water) was injected (5 μL) using the autosampler into the LC system. The response of the standard mixture was monitored continuously to produce a profile of the matrix effect. Any drop in the constant baseline signal (‘suppression of signal’) indicates presence of interfering matrix components [28–30]. The process was repeated with samples of serum. Five different serum samples were tested to access any possibility for sample-specific ion suppression effects.
Results and Discussion
The presence of phospholipids in a sample matrix can cause significant matrix effects when analysing small molecules in blood samples. Phospholipids are often extracted with analytes of interest and are difficult to resolve when using mass spectrometry. The HybridSPE-Precipitation cartridge facilitates protein precipitation together with solid phase extraction (SPE) technology for the targeted removal of phospholipids (i.e., through binding of phospholipids to the cartridge stationary phase) from plasma and serum samples. The unbound compounds of interest are then collected at the samples load step. A further elution step may be applied to increase the recoveries of the target compounds. No wash step to remove matrix interferences is required in this procedure.
Ismaiel et al.  have investigated the effect of several biological extract injections on the elution time of glycerophosphocholines to determine the effect of repeated injections on analytical performance. They commented that “monitoring phospholipids may provide a means to ensure the avoidance of matrix effects in each individual sample and may provide a more practical tool for avoiding matrix effects than commonly used post extraction addition and post-column infusion”. In another study, Ismaiel et al.  confirmed that ion suppression and enhancement coincided with the elution profiles of phospholipids in the analysis of chlorpheniramine in plasma.
Previously reported analytical methods describing the sample preparation technique along with the recoveries and sensitivity of each method
LOD 0.005 nmol L−1
LOQ 0.013 nmol L−1
Eriksson et al. 
SPE (Thermo Hypersil Keystone C18)
LOD 1.2 and 2.0 nmol L−1 for blood and urine
LOQ 6.4 nmol L−1
Stephanson et al. 
Hydroxyindole acetic acida
On-line SPE (HySphere resin GP cartridges 10 × 2 mm)
LOD <0.10 μmol L−1
LOQ 0.13 μmol L−1
de Jong et al. 
On-line SPE Phenomenex strong cation exchange (SCX), 4.0 mm × 3.0 mm,
LOD 1.5 nmol L−1
LOQ 5 nmol L−1
98 % mean (range 89.5–115.5 nmol L−1)
Monaghan et al. 
SPE Oasis MCX 10 mL (30 mg),
LOD 150 nmol L−1
Peterson et al. 
Hydroxyindole acetic acida
LOD 5 nmol L−1
LOQ 15 nmol L−1
Miller et al. 
Serotonin, hydroxyindole acetic acida, homovanillic acidb
SPE-SAX Oasis cartridges
LOD 0.05 ng mL−1
LOQ 0.1 ng mL−1
Saracino et al. 
LC-18 and Carbograph
LOD 0.013 nmol L−1
Lagana et al. 
Determination of Serotonin Using LC–MS/MS
Multiple reaction monitoring (MRM) studies were carried out to investigate the fragmentation of analyte precursor ions into their characteristic product ions and to establish the optimum collision energies. The most abundant MRM ions were chosen for quantitation (Table 1) with other optimised MRM ion combinations used for unequivocal identification of the analytes.
Optimisation of Chromatography
This method was originally developed for eight compounds, dopamine, 5-hydroxytryptophan (5-HTP), serotonin (5HT), tryptophan, 5-hydroxytryptophol (HTOL), 5-hydroxyindole acetic acid (HIAA), N-acetyl-hydroxytryptamine (N-acetyl-5HT) and melatonin. A gradient method was developed to resolve these compounds from each other. This was particularly important as HTOL was seen to interfere with serotonin quantitation giving a co-eluting peak due to similar MRM transitions. Another requirement of the method was the need to ensure dopamine did not co-elute with the instrument void volume (~1.5 min) as many matrix components and sample salts elute in this chromatographic region and caused ion suppression of the dopamine signal.
HybridSPE-Precipitation for Sample Preparation
A post-column infusion experiment was carried out using the established method to investigate for matrix effects. The serum samples (See Online Resource 1) show large areas of ion suppression and enhancement, this is due to the matrix components that exit the column during the high organic wash stage of the gradient (9–13 min). By having a high aqueous mobile phase composition at the beginning of the gradient, the dopamine peak was separated from the area of ion suppression at 2 min but signal response of dopamine was still compromised by an ion suppression effect of ~20 % (Online resource 1b). Serotonin was affected to a lesser degree by ion suppression of ~5 %.
A dilution study was carried out on the serum sample to investigate the extent of ion suppression on the signal response of serotonin; a similar study was described by Villagrasa et al. . Online resource 2 shows the effect of dilution on the signal response. On dilution, the response of the sample should match the theoretical response, but a slight increase was observed. This illustrates the impact of ion suppression on the serotonin signal response: as the matrix is diluted the components responsible for ion suppression are also diluted and, therefore, contribute less to signal suppression, resulting in a higher sample response. Although this proves that serotonin is affected by ion suppression the extent of ion suppression in this dilution study is relatively low at 5–8 %.
Percentage recovery for compounds in serum after HybridSPE-Precipitation cartridge clean-up
Spiked concentration (nmol L−1)
Measured concentration (nmol L−1)
% RSD (n = 5)
Retention time stability of an analyte is an important parameter to investigate during validation, as shifting retention times can lead to inaccuracies in identification as well as variations in the signal response of the analyte. The high selectivity of MRMs in LC–MS/MS allows for a higher degree of confidence in using retention time as identification marker. The EC council directive 2002/657/EC (implementing the 96/23)  suggests that retention time should not vary by more than 5 %. For this method, retention times were recorded over 3 days with three repeat injections per day of seven standards in the range of 0.086–3.286 μmol L−1 spiked into blank serum. Retention time shifts of ≤2.6, 3.6 and 1.2 % for serotonin, dopamine and melatonin, respectively, were observed. Analytes with an internal standard showed almost no difference in relative retention time as the deuterated internal references matched the retention characteristics of the analytes.
The three most abundant MRM transitions were chosen for each compound. Each precursor/product ion profile allows positive identification of the analytes and the Q1/Q3 ion pairs corresponds to 5.5 identification points according to the EC directive 2002/657/EC (implementing the 96/23) . The ion ratios for the MRM transitions of each target compound were monitored over the range of 0.043–3.29 μmol L−1. The ion ratios and their standard deviations over this range are provided in Table 1. The ion ratios seen here were observed to be common for all calibrations in matrix and in samples, indicating that they can be used for analyte confirmation.
Results for linear range, trueness and precision for serotonin, melatonin and dopamine in serum
Linear range/calibration curve
Intra + interday precision
Equation of the line
Linear range (μmol L−1)
LOD (nmol L−1)
LOQ (nmol L−1)
Theoretical concentration (nmol L−1)
Measured concentration (nmol L−1)
RSD % n = 5
Concentration (nmol L−1)
Intraday RSD % n = 6
Interday RSD % n = 9
0.0014x + 0.0184
985.82x − 13245
0.0011x + 0.0332
Limits of Detection and Quantitation
The LOD and LOQs for serotonin in serum samples are given in Table 4. The LOD and LOQ were calculated for each compound in urine. The S/N was established from the standard deviation of the Y-intercept in six low range calibrations which is then divided by the slope of the average calibration for the six calibrations. This value was then multiplied by 3.3 for the LOD and 10 for the LOQ.
Trueness and Precision
The trueness of this method was expressed as percentage relative error of three different concentrations spiked into serum (water and 1 % bovine albumin in phosphate buffer pH 7). Medium and high concentration gave excellent recoveries for all compounds with percentage relative error (% RE) values ranging from 0.3 to 8.4 % (see Table 4). However, analytes showed reduced accuracy at low concentrations with % RE values at 284 and 215 nmol L−1 of 13.6 and 12.3 % for serotonin and melatonin, respectively, and 20.7 % for dopamine (at 131 nmol L−1). This loss in recovery is possible due to ion suppression which corresponds to the signal suppression observed for the post-column infusion experiment, i.e. 5 % for serotonin and melatonin and 20 % for dopamine.
In serum, the % RSD values for intra-day and inter-day precision were excellent for both serotonin and melatonin (Table 4) with both compounds showing good values even at concentrations close to their LODs: ~4 and 3 % for serotonin and LOD ~7 and 8 % for melatonin for intra-day and inter-day studies. Dopamine gave higher % RSD at the LOD level (65 nmol L−1) of 22.0 and 18.1 %, while all other higher concentrations provided good levels of precision ranging between 4.5 and 10.5 %.
Application of Method
Results for serotonin and dopamine detected in clinical samples
Serum (nmol L−1)
This study outlines a validated method for serum for the analysis of serotonin, dopamine, and melatonin in serum. The method shows excellent results for accuracy and precision at concentrations expected in samples. It has been demonstrated that matrix effects are present with serum samples. This results in a reduction in analyte signal response at low concentration levels, particularly with the analysis of dopamine as it elutes close to an area of high ion suppression. Serotonin and melatonin are affected to a lesser extent by ion suppression and the additional presence of an internal standard for both compounds helps correct for these effects. Additionally, the study provided insight into the optimum storage conditions for the afore-mentioned biological samples. Finally, this method presents a simple, fast and reproducible sample preparation protocol using the HybridSPE-Precipitation cartridge. While dopamine showed issues of poor recovery at low concentrations, all parameters showed excellent results at levels present in serum samples. This shows that effective removal of phospholipids results in a sufficiently clean sample and that the HybridSPE-Precipitation cartridge can be applied to routine testing.
We gratefully acknowledge funding from the Irish Research Council for Science, Engineering and Technology (IRCSET) funding M. Moriarty. The Council of Directors, Technological Sector Research-Strand III 2006 Grant Scheme, awarded to Dr. A. Furey is also acknowledged for funding the formation of the ‘Team Elucidate’ research group. The Higher Education Authority (Programme for Research in Third-Level Institutions, Cycle 4 (PRTLI IV) National Collaboration Programme on Environment and Climate Changes: Impacts and Responses is also acknowledged.