Advantages of HILIC Mobile Phases for LC–ESI–MS–MS Analysis of Neurotransmitters

Published on: 
, , ,

LCGC Europe

LCGC Europe, LCGC Europe-03-01-2013, Volume 26, Issue 3
Pages: 128–140

The aim of this work was to find the optimal conditions to achieve sufficient limits of detection (LOD) that would permit the detection of neurotransmitters by LC–MS–MS in biological samples. An optimized HILIC–ESI–MS–MS system for the analysis of the 12 selected compounds was proposed.

This article investigates the influence of liquid mobile phase composition on the response measured in electrospray ionization-mass spectrometry (ESI–MS) for 12 selected neurotransmitters. The aim was to find the optimal conditions to achieve sufficient limits of detection (LOD) that would permit their detection by liquid chromatography-tandem mass spectrometry (LC–MS–MS) in biological samples (brain extracts). The advantages of solvent-rich mobile phases typically used in hydrophilic interaction liquid chromatography (HILIC) are clear. The HILIC–ESI–MS–MS system optimized for the 12 selected compounds analysis presented significant advantages over other existing methods.

Neurotransmitters are important polar biological compounds. For example, catecholamines and indolamines are very important because anomalies in their concentrations lead to a number of diseases, including Parkinson's, Alzheimer's or Down's disease, depression, schizophrenia and epilepsy (1). Their physiological concentrations are very low and they are usually analysed in the presence of complex matrices. Mass spectrometry (MS) has become the preferred tool for their detection and quantification to reach very low limits of detection (LOD) (2–6). While some derivatization is required to analyse neurotransmitters with gas chromatography (GC) (7), their direct separation (without the need for a prior derivatization step) can be conducted with capillary electrophoresis (CE) (8) or liquid chromatography (LC). The latter can be achieved with different elution modes: ion-exchange (IEC), reversed-phase liquid chromatography (RPLC), ion-pairing liquid chromatography (IPLC) or hydrophilic interaction liquid chromatography (HILIC).

IEC (9) is of limited interest because it can only separate compounds under a given charge state (positive or negative), while neurotransmitters, their precursors and metabolites may be either neutral, cationic, anionic or zwitterionic in the same sample. IPLC, an alternative approach to RPLC can effectively increase the retention of ionizable polar compounds allowing satisfactory separation of both anionic, cationic and neutral neurotransmitters in the same run (10). However, the anionic ion-pairing agent added to the mobile phase prevents the MS detection under a negative ion (NI) mode and thus renders MS detection impossible for anionic analytes. RPLC (2,3,6,10-12) and HILIC (13–15) do not have any of these disadvantages. Both have proven to be capable to retain and separate neurotransmitters and compatible with a MS detection. RPLC separations can be made easier by using stationary phases possessing some polar character (such as mixed-mode phases like pentafluorophenyl phases or porous graphitic carbon) to get sufficient retention of the polar neurotransmitters.

Our group previously developed satisfying RPLC (10) and HILIC separations of catecholamine and indolamine neurotransmitters in varied mobile phase compositions (13), and we were keen to select those conditions that would provide the best MS response to allow for both robustness and low detection limits. For the biological analysis of neurotransmitters sample size is usually very limited, even if adequate sample preparation to concentrate samples prior to the analysis is performed. Thus the loest possible LOD with ESI–MS would be advantageous. However, it is well known that electrospray ionization (ESI) strongly depends on the eluent nature, thus detection issues cannot be considered independently of the chromatographic system.

Although it is generally accepted that the buffered solvent-rich HILIC mobile phases are most favourable for ESI–MS detection (16), only a few studies have investigated the effect of the chromatographic operating parameters on the MS response. Some research has simply shown how HILIC conditions compared favourably to RPLC conditions when using MS detection (17,18) but those examples were only provided for a couple of compounds. Fountain et al. (19) investigated the influence of HILIC mobile phase pH on MS response for 21 small test compounds, comprising acids, bases and organophosphonic acids.

While acetonitrile (ACN) and methanol (MeOH) provide similar eluotropic strength in RPLC, they behave very differently in the HILIC mode. In the latter case, methanol is actually a stronger eluent than acetonitrile for most compounds (20). The explanation usually given is that methanol would be too hydrophilic for the necessary two phases (mobile phase and aqueous layer covering the stationary phase) to appear. Thus so-called "true HILIC partitioning" would not be possible with methanol.

Consequently, for most HILIC applications, methanol was found to be of little use because it provided only limited retention. However, for some particular applications, methanol was preferred to acetonitrile [see for instance (21)]. It appears that methanol is most useful in the HILIC mode when hydrophilic partitioning is not the sole mechanism, but when extra interactions occur between analytes and stationary phase (22). It sems that simple partitioning is rarely the case in HILIC chromatography. Secondary retention mechanisms such as electrostatic attraction and repulsion (23-25) and hydrogen bonding with the stationary phase (26) have also been reported and play a significant part in the retention and separation process (27). The possibility of using methanol in the HILIC mode is therefore highly dependent both on the nature of the analytes and the stationary phase chemistry. For example, lactic acid was well retained in methanol mobile phases on basic stationary phases such as amino and imidazole phases, probably as a result of electrostatic forces, while other stationary phase chemistries proved unsuccessful (28).

When adequate retention is obtained with methanol, it can beneficial to replace acetonitrile (totally or partly) for solubility issues (29), economic reasons (30,31), or because peak shapes are improved (32). Another reason for using methanol, which is particularly relevant to the current study, is that ionization efficiency in the electrospray interface to MS is improved, which provides better sensitivity (30,33).

The selected analytes for this study were the catecholamines: dopamine (DA), adrenaline (A), noradrenaline (NA); the indolamine: serotonine (S), along with some of their precursors tyrosine (Tyr), 3,4-dihydroxy-phenylalanine (DOPA), tryptophan (Trp); and metabolites: homovanillic acid (HVA), 3-methoxytyramine (3MT), 3,4-dihydroxy-phenylacetic acid (DOPAC), and 5-hydroxyindole-3-acetic acid (5HIAA). Thus three different classes were present: acids, bases and amino acids. Their chemical structures are presented in Figure 1.

Figure 1: Chemical structure of the analysed compounds.

We then set about evaluating the influence of different chromatographic parameters on the mass spectrometric signal intensity of neurotransmitters. The tested chromatographic parameters were: the concentration of the buffer salt, the organic solvent nature (acetonitrile and methanol) and the percentage. The latter ranged from 10% to 90%, thus encompassing both RPLC and HILIC mobile phase compositions. These effects were studied by monitoring ion response using flow injection analysis (FIA)–ESI–MS. Finally, we measured the detection limits obtained for these compounds in a brain extract sample with an optimized HILIC–ESI–MS–MS system.


Instrumentation: A chromatographic system composed of a binary pump and autosampler series 200 with 10 µL loop was used (PerkinElmer, Toronto, Canada). Detection was realized with a API 3000 mass spectrometer (Applied Biosystems, Foster City, California, USA) with triple-quadrupole and TurboIon spray as the ion source (Applied Biosystems, Foster City, California, USA). The mass spectrometer was operated in both positive ionization (PI) and negative ionization (NI) modes. The optimized MS parameters were as follows:

Positive ionization: ion spray voltage 5800 V, nebulizer gas was compressed air at a flow rate of 1.2 L/min, curtain gas was nitrogen at a flow rate of 0.9 L/min, collision exit potential 10 V;

Negative ionization: ion spray voltage -4500 V, nebulizer gas was compressed air at a flow rate of 1.2 L/min, curtain gas was nitrogen at a flow rate of 0.9 L/min, collision exit potential -15 V.

The source temperature was 300 °C for both PI and NI. The values for the focusing potential (FP), the declustering potential (DP), the entrance potential (EP) and the collision energy (CE) are different for each selected transition and are presented in Table 1. Data handling was realized with Analyst 1.4.1. software (Applied Biosystem MDS Sciex, Foster City, California, USA).

Table 1: Optimized parameters for the mass spectrometric detection.

Flow injection analysis (FIA) at 1 mL/min flow (split 1/3) was used for this study. FIA is used as a quick and easy way to study ion formation. In this configuration, no chromatographic column is employed and the LC system is simply used as an introduction apparatus. Each analyte was injected individually, and the analysis was repeated four times for each analyte in each of the 72 mobile phase conditions tested.

For the LC–MS–MS, the column used was a silica column, Pursuit Si (Varian, Les Ulis, France) 150 mm × 2 mm, 3 µm. Mobile phase flow-rate was 0.2 mL/min (no split required).

Chemicals and Reagents: HPLC-grade acetonitrile and methanol were purchased from J.T. Baker (Noisy-le-Sec, France). Ammonium acetate, ammonium formate, acetic acid, formic acid and 3,4-dihydroxybenzylamine (DHBA) were purchased from Fluka (St.-Quentin-Fallavier, France). Adrenaline (A), noradrenalin (NA), dopamine (DA), serotonin (S), 3,4-dihydroxy-phenylalanine (DOPA), 3-methoxytyramine (3MT), tryptophan (Trp), homovanillic acid (HVA), tyrosine (Tyr), 5-hydroxyindole-3-acetic acid (5HIAA), and 3,4 dihydroxyphenyl acetic acid (DOPAC) were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). The perchloric acid used was produced by VWR Prolabo (Darmstadt, Germany). Deionized (18 MO) water, purified using an Elgastat UHQ II system (Elga, Antony, France), was used for preparation of analyte and mobile phase solution.

Procedure and Preparation of Samples: Stock standard solutions of each catecholamine, indolamine,and metabolite prepared at a concentration of 1000 mg/L were obtained by dissolving the weighed amount of each compound with 0.2 M perchloric acid. All stock solutions were stored at -80 °C. The injected solutions were prepared by dilution of the stock solutions to have an injection solvent as close as possible to the mobile phase. The concentrations for the injected solutions were 5 µg/mL.

A selection of 72 mobile phase compositions were prepared for a study of the organic solvent and buffer influence on the electrospray response. For organic solvent effect, the mobile phase used was ammonium formate buffer at pH 3 with either methanol or acetonitrile, in proportions varying from 10% to 90%. Salt concentration effect was studied with ammonium formate buffer, with pH adjusted at 3. The concentration of salt in the aqueous phase varied from 10 mM to 150 mM.

For the determination of the LODs, mixtures of the 12 analytes of decreasing concentration were injected in LC–MS–MS until the signal-to-noise ratio (S/N) of each compound was equal to 3.

For the brain extract preparation, the sheep encephalon was dissected out of the skull and was separated in different regions that were weighed and then immersed in cold 0.2 mol/L perchloric acid at a ratio of 5 mL/g tissue. The brain tissue was homogenized by sonication. The tissue homogenate was centrifuged at 20,000 g for 1 h at 4 °C. The supernatant was used as the brain extract and stored at -80 °C. Just before analysis, the brain extract was filtered through a 0.45 µm syringe filter. An aliquot (100 µL) of the filtrate was then diluted in 900 µL of a mixture composed of 50 µL ammonium formate buffer and 850 µL of acetonitrile, to obtain a final solvent composition as close as possible to that of the HILIC mobile phase.

Data Analysis: Marvin 4.1.11. software (ChemAxon, Budapest, Hungary) was used to calculate the analytes pKa and log D values. Principal component analyses (PCA) were performed with XLStat 2009.2.03 software (Addinsoft, New York, NY, USA). PCA calculations were performed on MS response for all 12 compounds after auto-scaling (all columns were centred and reduced) to give all variables the same significance.

Results and Discussion

Our previous studies concerning the optimization of HILIC mode analysis have shown, along with others, that the most important parameters to achieve a separation are: organic solvent nature and percentage, salt concentration and pH (13). Thus in this paper we studied the influence of these parameters on the electrospray ionization of catecholamines and indolamines.


Solute Ionization: As appears in Table 1, all solutes were observed in the positive ionization mode, apart from HVA and DOPAC, which could only be observed in the negative ionization mode. The MS response among the 12 solutes varied to a large extent, as illustrated in Figure 2. Regardless of the salt concentration and solvent proportion, 3MT, A and S generally provided the most significant response, while HVA and DOPAC observed in the negative ionization mode generally provided the least significant response. Besides, it appears that the six amine compounds, which certainly have the strongest proton affinities, generally provided more significant MS response (in the positive ionization mode) than amino acids and acids.

Figure 2: Influence of the solvent nature on the electrospray ionization. Mobile phase: Methanol (blue or dark blue) or acetonitrile (orange or red) with (a) 90% ammonium formate 10 mM, pH 3 or (b) 10% ammonium formate 100 mM, pH 3. All compounds observed in the positive ionization mode, except homovanillic acid (HVA) and 3,4-Dihydroxyphenylacetic acid (DOPAC) observed in the negative ionization mode.

Influence of the Organic Solvent Nature and Percentage: The influence of two organic solvents (methanol and acetonitrile) on the selected analytes peak intensity was tested. The mobile phases were composed of 10% to 90% organic solvent, mixed with ammonium formate buffer pH 3. The salt concentration in the buffer solution varied from 10 mM to 150 mM, and thus global salt concentration varied between 1 mM and 135 mM.

In Figure 2, ionization is compared for 10% [Figure 2(a)] and 90% [Figure 2(b)] organic solvent mixed with 10 mM and 100 mM ammonium formate buffer, respectively. Therefore the overall concentration of buffer in each mobile phase is close (9 mM and 10 mM, respectively). It is important to note that the scales are different between Figure 2(a) and 2(b), indicating that the MS response with large proportions of solvent was 2 to 50 times larger than with low proportions of solvent. This was not unexpected. At 10% organic solvent, methanol and acetonitrile provided a very similar response for most analytes. There were two exceptions: i/ DHBA was significantly better ionized with acetonitrile, and ii/ 5HIAA, the sole acidic analyte to be observed in the positive ionization mode, provided a significant response with methanol while it was only rarely detected with acetonitrile. At 90% solvent, the differences between the two solvents were more noticeable, with methanol providing the most significant response in most cases. Only 3MT and S had a significantly better response with acetonitrile. Solvent evaporation rate from the droplet surface in ESI is a function of vapour pressure, thus higher volatility of methanol probably favoured ESI ionization.

It is well known that aqueous buffer-solvent mixtures generally provide better ESI–MS response than purely aqueous buffers, because of the relatively low vapour pressure of water. Thus, better sensitivity results are generally obtained when surface tension is decreased through the addition of a volatile organic solvent. However, a certain amount of water must remain to provide adequate conduction in the droplets. Figure 3 shows that signal intensity varies differently with the percentage of the two organic solvents selected for this study. The figures were intentionally plotted with identical scales of the y-axis to show the relative variations observed with acetonitrile and methanol. In addition, standard deviations were calculated, based on four measurements repeated for each solute, in each mobile phase condition, but the error bars are not visible on the figure as they are smaller than the data points.

Figure 3: : Influence of the solvent proportion on the electrospray ionization, with varied concentrations of salt in the buffer. (a) 3,4-Dihydroxybenzylamine (DHBA) in acetonitrile, (b) DHBA in methanol, (c) HVA in acetonitrile and (d) HVA in methanol. (a) and (b) in the positive ionization mode; (c) and (d) in the negative ionization mode. Mobile phase: solvent = ammonium formate buffer, pH 3.

Increasing the percentage of solvent improved the MS response for all 12 compounds. However, the extent of this improvement depended on the analyte, solvent nature and buffer salt concentration. The curves obtained for DHBA and HVA are provided as a representative example (Figure 3). We must point out that the increase of solvent proportion is concomitant to the decrease of the overall buffer salt concentration in the mobile phase, but we will show that salt concentration variation is not sufficient to account for the variations observed.

With acetonitrile and low buffer salt concentrations, it was puzzling to observe that the trend was not monotonic. Most compounds in the PI mode exhibited a curve similar to the one presented in Figure 3(a), where the MS response first increased, reached a maximum at 30-40% organic solvent, decreased down to a minimum at 50-60% organic solvent, and then increased again. These variations were only seen for small concentrations of buffer salt, while the curves obtained with larger salt concentrations were constantly increasing. The two compounds observed in the negative ionization mode did not seem to exhibit such discrepancies, as can be observed with HVA [Figure 3(c)].

With methanol [Figures 3(b) and 3(d)], the situation was different: increasing the methanol proportion essentially caused an increase in the MS response, but while the curves were not perfectly smooth, the up-and-down variations described previously were not observed.

Based on these figures, solvent-rich HILIC mobile phases seem to be advantageous compared to water-rich RPLC mobile phases. However, the advantage seems much more significant when methanol is the chosen co-solvent than when acetonitrile is used. The latter may provide a very similar response in RPLC and HILIC conditions, if the RPLC proportion of solvent can be adjusted around the first optimum value (30-40%). In addition, the composition of the HILIC mobile phase is very significant because the signal increases very sharply between 70% and 90% solvent.

Nevertheless, at this point, the influence of solvent proportion on MS response still remains questionable because salt concentration must be considered too. The concentration of salt was maintained constant in the buffer component of the mobile phase, while increasing solvent proportion causes a decrease in total salt concentration. This point will be further discussed in the following section.

Influence of the Salt Type and Concentration: When ionic compounds are analysed using HILIC, the presence of salt in the mobile phase is necessary for the analyte elution and good peak shape (13,15,24). The HILIC mobile phases usually contain high organic solvent percentages (typically between 60% and 95%). The number of salts that are soluble under these conditions are therefore limited to: ammonium acetate, ammonium formate, bicarbonate salt, triethylammonium phosphate and sodium perchlorate. The last two salts are not volatile and are thus incompatible with MS detection. For this study only the influence of ammonium formate concentration was explored. Previous exploratory studies have shown that the mass spectrometric signal intensity was not significantly affected by the salt nature when replacing ammonium formate with ammonium acetate.

For the study of the salt concentration influence on the ionization yield, the mobile phases described above, composed of 10% to 90% acetonitrile or methanol and ammonium formate solutions with salt concentrations of between 10 mM and 150 mM buffered at pH 3 with a 1 M formic acid solution, have been used. This corresponds to an overall salt concentration varied between 1 mM and 135 mM.

A significant ion suppression effect was observed for all compounds with the salt concentration increase, whatever the nature and proportion of organic solvent. As illustrated in Figure 4, where NA is shown as a representative example, the decreasing rate is more significant up to 10 mM (total salt concentration in the mobile phase), while the curve slope was lower with further concentration increase. Indeed, in ESI, buffers and salts cause a reduction in the vapour pressure and consequently a reduced signal.

Figure 4: Influence of the salt concentration on the ionization of noradrenaline (NA). Mobile phase: (a) acetonitrile (b) methanol with aqueous solution of ammonium formate pH 3.

When comparing Figure 4(a) and 4(b) it appears that ion suppression was much more significant with methanol than with acetonitrile, as the largest values measured with methanol are larger than with acetonitrile.

Figure 4 was plotted using the data acquired with different solvent proportions. In some cases, two or three points represent measurements obtained for identical total salt concentrations but different proportions of solvent. For instance, 5 mM total salt concentration can be observed both with acetonitrile - 10 mM buffer 50:50 (v/v) and with acetonitrile - 50 mM buffer 90:10 (v/v). Another example is 20 mM total salt concentration, which can be observed both with acetonitrile - 50 mM buffer 60:40 (v/v) and with acetonitrile - 100 mM buffer 80:20 (v/v). Consequently, the points are scattered around the tendency curve, particularly for total salt concentration in the 1-20 mM range. This clearly indicates that total salt concentration is not the sole parameter influencing ionization efficiency. Whenever two points with identical total salt concentration but different solvent proportions can be compared, the conditions with the highest percentage of solvent generally provide a more intense MS response. This conclusion therefore supports of the above comments on the effect of solvent proportion.

We have previously shown (13) that for catecholamine separation the HILIC chromatographic columns could be divided into columns that gave satisfactory separations and peak efficiency at low salt concentration (<10 mM total salt concentration) and columns requiring high salt concentrations (>20 mM) to obtain satisfactory separations and peak shape. Good peak shape and large column efficiency values are as important as good MS response to achieve low limits of detection. It appears that the signal intensity loss is no longer significant above 10 mM. Thus, for those columns requiring a high salt concentration, it is advisable to work with the best chromatographic conditions, that will also allow a satisfying detection sensitivity, even if the salt concentration is high.

Low salt concentration would not favour robust conditions of analysis because slight variations of concentration (as a result of incorrect buffer preparation or inaccurate proportions of solvents) cause large variations in the MS response. In addition, in gradient elution mode, when the salt is present only in the aqueous component of the mobile phase, detection would be significantly affected by the mobile phase composition at the moment of entering the ESI source. This would be even more critical when methanol is used. As a result, low salt concentrations (below 10 mM total salt concentration) and methanol should only be favoured for those compounds that do not ionize sufficiently to allow for larger LODs.

Principal Component Analysis: For an overview of the influence of the various factors studied on the MS signal intensity, we used principal component analysis (PCA). PCA is a standard tool in modern data analysis. It provides a way of identifying statistical patterns in data, and expressing the data to highlight their similarities and differences. In this study, the 12 compounds represented by 72 values of MS response measured in 72 mobile phase compositions, were treated as variables. In other words, each of the 72 mobile phase compositions is defined by the 12 MS response values measured for the 12 compounds. Thus, this was initially a 12-dimension problem. Since patterns in data can hardly be found in such a high dimension, where no graphical representation is possible, the primary aim of PCA is to project all data points onto a space of reduced dimension, most practically on a plane. This plane, defined by two axes named principal components, must be chosen to retain as much of the initial variance as possible. The plots issued from a PCA reveal simplified structures and patterns. The score plots allow a direct comparison of the similarity of the mobile phase compositions while the loadings provide a means to identify which compounds are mostly influenced by a change in mobile phase composition.

The PCA was performed using the data from the 72 experiments performed with all 12 compounds, with varying proportions of acetonitrile or methanol (10% to 90%) and varying salt concentration in the ammonium formate buffer (10 mM to 150 mM). All initial data were normalized, that is to say they were centred (the mean was subtracted) and reduced (they were divided by the standard deviation). This produces a data set whose mean is zero and standard deviation is one for every variable, thus gives all variables the same significance. This was particularly important to be able to compare the mobile phase conditions in terms of their influence on all solute ionization because the MS responses were initially on different scales, as shown in Figure 2.

Figures 5(a) and (b) present the score plot and loading plot obtained under the 72 mobile phase conditions investigated. PC1 and PC2 together account for nearly 88% of the variance, thus the PC1-PC2 plots are largely amenable to interpretation, and there is no need for attention to further components. PC1 explains more than 80% of the variance, while PC2 only carries about 6% of the variance.

Figure 5: Principal component analysis based on the ionization information of all the compounds and with 72 tested mobile phases. (a) PC1–PC2 score plot and (b) PC1–PC2 loading plot. Red squares represent experiments performed with acetonitrile; blue diamonds are experiments with methanol. See text for details on the experimental conditions. The AX CY points indicate that the experiment was performed with X% acetonitrile and Y% mM buffer salt concentration. The MX CY points indicate that the experiment was performed with X% methanol and Y mM buffer salt concentration.

First of all, it is important to notice that all compounds-variables point to the right-hand side of the loading plot, indicating that the mobile phase compositions situated on the right of the score plot provide more intense MS responses for the 12 compounds than the mobile phase compositions on the left. Looking at the points on the score plot, it appears, not surprisingly, that solvent proportion increases from left to right, while salt concentration decreases from left to right. This indicates that the MS responses for all compounds are affected by solvent proportion and salt concentration in a similar manner. The majority of the points are grouped on the left-hand side of the score plot, indicating that most of the compositions we have tested provide nearly equivalent MS ionization.

Additionally, it appears that the nature of the solvent is also significant because it favours the ionization of certain species over others. Indeed, in the score plot all data points related to experiments in methanol are plotted above the PC1 axis, while most data points related to experiments in acetonitrile are plotted below the PC1 axis. Only the acetonitrile data points obtained with 30% and 40% acetonitrile and 10 mM buffer are situated in the upper part of the figure. This particular scatter of the methanol and acetonitrile points can be related to the position of the compounds-variables in the loading plot. The electrospray ionization of the compounds which are pointing in the upper right quadrant (5HIAA, A, NA, DHBA and DOPA) is favoured by methanol, while the ionization of the compounds pointing in the bottom right quadrant (Tyr, DOPAC, HVA, 3MT and DA) is favoured by acetonitrile. However, the influence of solvent nature is much less significant than solvent proportion and salt concentration as the variance explained by the PC2 axis is much lower than the variance explained by the PC1 axis.

The interpretation of this figure must not be mistaken: Methanol does provide a stronger MS response to most compounds when used in large proportions and particularly with small concentrations of buffer. However, acetonitrile provides a stronger signal increase for some compounds when its proportion is increased. For example, increasing the proportion of acetonitrile from 10% to 90% with the 10 mM buffer causes a 30-fold signal increase for DOPAC, while the same change with methanol causes only a 15-fold signal increase. Similar observations can be made for other compounds in the bottom right quadrant. The alternative is true for the compounds in the upper right quadrant, for which methanol is more advantageous. For example, increasing the proportion of acetonitrile from 10% to 90% with the 10 mM buffer causes a two-fold signal increase for 5HIAA, while the same change with methanol causes a 20-fold signal increase.

In conclusion, if we classify the influence of the different parameters tested on the ionization yield, we can unambiguously say that the salt concentration and organic solvent percentage are the factors that most influence the ionization, while solvent nature plays a less significant part.

Figure 6: LC–MS–MS (specific extracted ion current, XIC) analysis of neurotransmitters under selected HILIC conditions. (a) Positive ionization mode; standard concentration: 100 ng/L for each injected solute: (b) and (c) negative ionization mode; DOPAC concentration: 1000 ng/L and HVA concentration: 5000 ng/L. See chromatographic conditions: Pursuit Silica Column; mobile phase: mixture of acetonitrile and aqueous solution of ammonium formate 150 mM (85:15, v/v).

Based on these observations, it is clear that HILIC mobile phases should provide improved sensitivity for the MS detection of the 12 tested neurotransmitters than RPLC mobile phases.

LOD Measurement: Although generally better signal intensity was obtained with methanol as organic solvent, we found that we could not obtain adequate retention and separation of the 12 neurotransmitters with this solvent. We therefore developed an acetonitrile-based separation.

Figure 6 depicts the chromatographic separation obtained on an optimized HILIC system. In accordance with the above remarks, we have favoured a large proportion of acetonitrile to ensure good MS response, but the total salt concentration was 22.5 mM, therefore above 10 mM, which was shown to be a critical limit for robustness of the method.

The LODs obtained when this chromatographic system was coupled to a triple quadrupole mass spectrometer are presented in Table 2. When analysing real samples, the presence of the matrix caused varied effects: for some compounds, the LODs are improved (particularly for NA or DOPA), while for others the matrix caused signal suppression and thus higher LOD values (Table 2).

Table 2: Limit of detection (LOD) of the optimized hydrophilic interaction chromatography system.

On one hand, the LODs we obtained in the HILIC mode are better than those presented by Hasegawa et al. (34) in reversed-phase chromatography. Using an ODS L-column (150 mm × 4.6 mm, 5 µm) and a mobile phase composed of water and methanol, each containing 1% formic acid (in gradient elution mode), they obtained LODs of about 1-1.5 µg/mL for NA, A, DA and DOPA. These values are at least 200 times larger than what we present here. However, their MS analysis was performed using a different mass spectrometer from ours: an electrospray ionization time-of-flight mass spectrometer. Thus it is difficult to know if the improvement in LOD values is related to mobile phase composition or to the mass spectrometer.

On the other hand, Bourcier et al. (3), using the same triple quadrupole mass spectrometer as the one used here, obtained similar LODs (of about 10-20 ng/mL) for the analysis of 28 neurotransmitters (their mixture includes all our test compounds with the exception of DHBA). The chromatographic conditions they used were RP-HPLC with a gradient elution of 0.1% formic acid in water and acetonitrile. The peculiar variations in the intensity of the MS response using acetonitrile mobile phase may thus have been advantageous in this case.


The study of the influence of various chromatographic parameters on the signal intensity in MS has shown that optimum conditions for ionization of neurotransmitters exist: the best MS sensitivity was obtained with mobile phases with low salt concentrations and large proportions of solvent. However, such mobile phase conditions are not desirable as MS detection is not very robust with low salt concentrations. In this case a compromise must be found between the optimum conditions of separation and the MS signal.

The HILIC system presented has the significant advantage of being compatible to both positive and negative ionization modes to provide the possibility of detecting the acidic compounds that responded only in the negative mode.


The authors wish to thank Professor Tobias Hevor from the University of Orléans (Orléans, France) for providing a sheep brain extract.


(1) K. Yamamoto and O. Hornykiewicz, Progr. Neuro-Psychopharmacol. Biol. Psychiatry28(5), 913 (2004).

(2) A. Törnkvist, P.J.R. Sjöberg, K.E. Markides and J. Bergquist, J. Chromatogr. B801(2), 323 (2004).

(3) S. Bourcier, J.F. Benoist, F. Clerc, O. Rigal and M. Taghi, Rapid Commun. Mass Spectrom. 20(9), 1405 (2006).

(4) V. Carrera, E. Sabater, E. Vilanova and M.A. Sogorb, J. Chromatogr. B 847(2), 88 (2007).

(5) M.M. Kushnir, F.M. Urry, E.L. Frank, W.L. Roberts and B. Shushan, Clin. Chem. 48(2), 323 (2002).

(6) W. Li, D.T. Rossi and S.T. Fountain, J. Pharm. Biomed. Anal. 24(2), 325 (2000).

(7) U.E.G. Bock and P.G. Waser, J. Chromatogr. A 213(3), 413 (1981).

(8) H. Sirén and U. Karjalainen, J. Chromatogr. A 853(1–2), 527 (1999).

(9) B.M. De Borba and J.S. Rohrer, J. Chromatogr. A 1155(1), 22 (2007).

(10) R.-I. Chirita, A.-L. Finaru and C. Elfakir, J. Chromatogr. B 879(9–10), 633 (2011).

(11) P. Manini, R. Andreolia, S. Cavazzinia, E. Bergamaschia, A. Mutti and W.M.A. Niessen, J. Chromatogr. B 744(2), 423 (2000).

(12) A.T. Wood and M.R. Hall, J. Chromatogr. B 744(1), 221 (2000).

(13) R.-I. Chirita, C. West, A.-L. Finaru and C. Elfakir, J. Chromatogr. A 1217(18), 3091 (2010).

(14) Q. Gu, X. Shi, P. Yin, P. Gao, X. Lu and G. Xu, Anal. Chim. Acta 609(2), 192 (2008).

(15) X. Zhang, A. Rauch, H. Lee, H. Xiao, G. Rainer and N.K. Logothetis, Rapid Commun. Mass Spectrom. 21(1–2), 3621 (2007).

(16) H.P. Nguyen and K.A. Schug, J. Sep. Sci. 31(9), 1465 (2008).

(17) E.S. Grumbach, D.M. Diehl and U.D. Neue, J. Sep. Sci. 31(9), 1511 (2008).

(18) E.S. Grumbach, D.M. Wagrowski-Diehl, J.R. Mazzeo, B. Alden and P.C. Iraneta, LCGC North America 22(10), 1010 (2004).

(19) K.J. Fountain, J. Xu, D.M. Diehl and D. Morrison, J. Sep. Sci. 33 (6–7), 740 (2010).

(20) S. Vikingsson, R. Kronstrand and M. Josefsson, J. Chromatogr. A 1187(1–2), 46 (2008).

(21) H.S. Kim, D. Siluk and I.W. Wainer, J. Chromatogr. A 1216(16), 3526 (2009).

(22) L. Discenza, C. D'Arienzo, T. Olah and M. Jemal, J. Chromatogr. B 878(19), 1583–1589 (2010).

(23) R. Chirita, P. Chaimbault, A. Finaru and C. Elfakir, LCGC en Français 1, 8 (2008).

(24) Y. Guo and S. Gaiki, J. Chromatogr. A 1074(1–2), 71 (2005).

(25) A.J. Alpert, Anal. Chem. 80(1), 62 (2008).

(26) W. Bicker, J.Y. Wu, M. Lammerhofer and W. Lindner, J. Sep. Sci. 31(16–17), 2971 (2008).

(27) R.-I. Chirita, C. West, S. Zubrzycki, A.-L. Finaru and C. Elfakir, J. Chromatogr. A 1218(35), 5939 (2011).

(28) M. Storton, J. Exarchakis, T. Waters, Z. Hao, B. Parker and M. Knapp, J. Sep. Sci. 33(6–7), 982 (2010).

(29) M.S. Ali, M. Ghori, S. Rafiuddin and A.R. Khatri, J. Pharm. Biomed. Anal. 43(1), 158 (2007).

(30) C. Hubert, S. Houari, F. Lecomte, V. Houbart, C. De Bleye, M. Fillet, G. Piel, E. Rozet and P. Hubert, J. Chromatogr. A 1217(19), 3275 (2010).

(31) H. Pligoropoulou, A. Vonaparti and I. Panderi, J. Chromatogr. B 893–894, 114–120 (2012).

(32) M. Schwalbe-Herrmann, J. Willmann and D. Leibfritz, J. Chromatogr. A 1217, 5179–5183 (2010).

(33) Z. Xiong, Y. Zhang, F. Qin, T. Qin, S. Yang and F. Li, J. Chromatogr. B 878(23), 2111 (2010).

(34) T. Hasegawa, K. Wada, E. Hiyama and T. Masujima, Anal. Bioanal. Chem. 385(5), 814 (2006).

Raluca-Ioana Chirita-Tampu is an assistant professor at the University of Bacau (Bacau, Romania). This work was part of her role during her post-graduation position at ICOA, University of Orléans (Orléans, France).

Caroline West is an associate professor in analytical chemistry at the University of Orleans. Her scientific interests lie in the fundamentals of chromatographic selectivity in SFC and HPLC.

Laetitia Fougere is a chemist engineer at the University of Orléans. Her scientific interests concern chromatographic development coupled to mass spectrometry. She applies these methodologies mainly to characterize active molecules from plant extracts.

Claire Elfakir is a professor in analytical chemistry. She is head of the analytical research group in the Institute of Organic and Analytical Chemistry at the University of Orléans. Her main interests concern strategy in separation science and bioactive molecules in complex media.