Introduction

Glutaminolysis consists of a series of biochemical reactions that catabolize glutamine (Gln) to glutamate (Glu), aspartate (Asp), pyruvate (Pyr), lactate (Lac), alanine (Ala), and citrate (Cit) [1, 2]. The degradation of Gln provides a new pillar for energy production by fueling the tricarboxylic acid (TCA) cycle and the malate-aspartate shuttle with its metabolites [3]. Moreover, glutaminolysis is found to be associated with redox balance, mTOR signaling pathway regulation, apoptosis, and autophagy, which can characterize physiological and pathological states in the biological system [4,5,6]. Hence, the disorder of glutaminolysis has been found to be closely related to various diseases [7,8,9]. Untargeted metabolomics studies reported that the levels of metabolites (Cit, Glu, etc.) in glutaminolytic pathway were alerted in the blood and urine samples of type 2 diabetic (T2D) patients [10, 11]. However, the untargeted approach is difficult to accurately characterize the meticulous changes in a specific metabolic pathway, especially involving isomers and low levels of metabolites. The study aims to develop a novel LC-MS/MS method for the simultaneous determination of these metabolites in glutaminolysis in human plasma. The discovered differential metabolites could be applied to diagnosis or targeted therapy for related metabolic diseases in further researches.

The main challenges in simultaneously determining the targeted metabolites (Table 1) in vivo are their high polarity, low molecular weight, and broad concentration range [12]. A number of analytical platforms have been established for the quantification of endogenous metabolites in human plasma, such as gas chromatography-mass spectrometry (GC-MS) [13] and liquid chromatography-mass spectrometry (LC-MS) [14], yet no method has directly accessed all 13 metabolites of interest.

Table 1 List of the structures, mass spectrometer parameters, linearity, and sensitivity for each analyte measured
Full size table

GC-MS is a powerful tool for the analysis of amino acids and organic acids involved in glutaminolysis [15]. However, derivatization, usually conducted in sample preparation of GC-MS, is tedious and time-consuming, even leads to the high complexity of chromatogram by the multi-peak and multi-origination phenomenon [16].

LC-MS technology is considered as a robust measurement of endogenous metabolites for the advantages of flexible separation mechanisms and specific monitoring mode [17]. So far, several LC-MS-based methods for a subset of the metabolites of interest have been reported [18,19,20,21]. For example, targeted profiling of 25 polar intracellular metabolites, including malic acid (Mal), fumaric acid (Fum), and Cit, was achieved on a porous graphitic carbon (PGC) column, whereas Lac and Pyr could not be eluted by any mobile phase investigated [19]. Ion pair chromatography (IPC) is another common technique for analyzing these polar molecules. Luo et al. [20] separated 29 negatively charged compounds on a common C18 column by adding tributylammonium acetate (TBAA) as the ion pair modifier. However, the irreversible contamination of MS and column by the ion-pairing reagent limited the application of this method. Additionally, pre-column derivatization was also used to enhance the LC retention and MS detection of endogenous metabolites, such as carboxylates related to TCA cycle [21]. However, the high labor intensity operation of derivatization may lead to inaccurate results.

Here, we present a simple, selective, and sensitive liquid chromatography-tandem mass spectrometry (LC-MS/MS) method to simultaneously determine 13 metabolites of glutaminolysis in human plasma. To our best knowledge, this is the first report of a comprehensive method that allows these metabolites to be measured in parallel in a single run. The main improvements compared with related state-of-the-art methods were easy sample pre-treatment, good isomer separation, and wide dynamic range. The method was fully validated and successfully applied to comparatively analyze the plasma samples from 38 healthy volunteers and 34 T2D patients. Its high sensitivity and specificity and comprehensive capacity made the present method a good choice to study glutaminolysis in pathological and physiological situations, particularly in energy metabolism–related researches.

Materials and methods

Chemicals and reagents

All standards (Table 1) with purity > 98% were purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC-grade methanol (MeOH) and acetonitrile (ACN) were purchased from Merck (Darmstadt, Germany). Deionized water was produced by a Milli-Q system from Millipore (Watford, UK). Formic acid, ammonium formate, ammonium acetate, ammonium bicarbonate, and ammonium hydroxide solution were obtained from Nanjing Chemical Reagent Co. Ltd. (Jiangsu, China).

LC-MS/MS analysis

A Finnigan™ TSQ Quantum Discovery MAXTM LC-MS/MS system (Thermo, USA) was composed of a Finnigan™ Surveyor LC pump, a FinniganTM Surveyor autosampler, and a triple-quadruple TSQ Quantum mass spectrometer. Xcalibur 3.0 software (Thermo, USA) was used to control the instruments and process the data. LC separations were achieved at 35 °C on a SeQuant™ ZIC®-pHILIC column (150 × 4.6 mm, 5 μm, Merck, Germany) with a SHIM-PACK GVP-ODS VanGuard column (2.0 I.D. × 5 mm, 5 μm, Shimadzu, Japan). In addition, ZORBAX SB-Aq column (100 × 2.1 mm, 1.8 μm, Agilent, USA) and XBridge® BEH Amide column (100 × 2.1 mm, 2.5 μm, Waters, USA) were used in our previous attempt. The mobile phase A was acetonitrile and B was water containing 5 mM ammonium acetate. The mobile phases were eluted at 0.5 mL/min following the gradient program: 0–8 min, 85% A → 50% A; 8–13 min, 50% A. A 7-min equilibrium was used before the next injection. The injection volume was 5 μL with partial loop injection, and the thermostatic autosampler tray temperature was kept at 4 °C constantly.

MS/MS detection was accomplished by an electrospray ionization (ESI) source in negative ion mode. The optimized MRM conditions for all analytes are listed in Table 1. Other operating parameters were set as follows: spray voltage 4.5 kV, capillary temperature 275 °C, sheath gas pressure 25 arbitrary units, auxiliary gas pressure 5 arbitrary units, collision gas (argon) pressure 1.5 mTorr. The scan width and scan time were 0.2 m/z and 0.1 s respectively. High-purity nitrogen served as both nebulizing and drying gas.

Stock and working solution preparation

The standards of Glu, iso-citrate (Iso-cit), Asp, Ala, and Gln were weighed accurately and dissolved in pure water with the concentration of 1.56, 1.56, 3.12, 15.6, and 31.2 mg/mL, respectively, and then stored at 4 °C until utilized. Stock solutions of Fum (1.56 mg/mL), succinic acid (Suc, 1.56 mg/mL), Mal (15.6 mg/mL), cis-aconitate (Cis-act, 15.6 mg/mL), α-ketoglutaric acid (Akg, 15.6 mg/mL), Pyr (15.6 mg/mL), Cit (31.2 mg/mL), Lac (62.4 mg/mL), and pregabalin (IS, 1 mg/mL) were prepared with methanol and stored in − 20 °C freezer. Combined working solutions were freshly made by serial dilutions of the stocks with MeOH/ACN (50/50, v/v) and stored at 4 °C.

Analyte-free matrix preparation

The analyte-free human plasma for calibration standards and quality control samples (QCs) was prepared by incubating with the activated carbon. Activated carbon was added to the plasma at the ratio of 100 mg/2 mL, then mixed and shaken with 1200 rpm for 6 h at room temperature. After centrifugation at 4 °C, 8000 rpm for 10 min, the supernatant was transferred to another 2.0-mL Eppendorf tube and centrifugated at 4 °C, 8000 rpm for 5 min again. The treated plasma was stored at − 80 °C until use.

Biological sample preparation

To eliminate the potential interferences and extract targeted metabolites from human plasma, 200 μL of MeOH/ACN (50/50, v/v) containing 10 μL IS (8 μg/mL) was added to 50 μL human plasma. The sample was vortexed for 3 min and centrifuged at 16,000 rpm, 4 °C for 10 min. The supernatant liquor was transferred to another tube and centrifuged at 4 °C, 16,000 rpm for 5 min again. Finally, an aliquot of 5 μL supernatant was injected into the LC-MS/MS system for analysis.

Method validation

The validation was conducted according to the FDA draft guidelines on validation of bioanalytical methods with additional experiments for endogenous analytes based on published methods [22]. Validation included lower limit of detection (LLOD), lower limit of quantification (LLOQ), linearity, accuracy, precision, carryover, stability, matrix effect, and extraction recovery.

Calibration curves and QCs

Working solution was spiked with analyte-free plasma to yield a calibration curve. The 6-point calibration curves were constructed by plotting the peak area ratios of all of the metabolites to the IS against the concentrations of the analytes. Based on the statistical analysis of the standard curves, the linear regressions were fitted with y = ax + b, and the regression coefficient (r) was calculated by weighting factor 1 ∕ x2. The LLOD was calculated at a signal-to-noise ratio of least 3:1. The LLOQ was defined as the lowest concentration on the calibration curve, with an analyte signal-to-noise ratio of 10. The QCs at the concentration of LLOQ, low-quality control (LQC), middle-quality control (MQC), and high-quality control (HQC) were assessed as calibration standards by spiking working solutions at an appropriate concentration.

Accuracy and precision

The inter-day and intra-day accuracy and precision of the method were evaluated by analyzing four concentration levels (LLOQ, LQC, MQC, and HQC), and each level contained five replicates on the same day and three validation days. The accuracy (%) at each concentration was calculated by the percentage ratio of measured concentrations of QC samples to the nominal value according to calibration curves. The precision (%) was expressed by the relative standard deviation (RSD). The acceptance criteria for accuracy are 85–115%, while for precisions are no more than 15% deviation from the nominal concentration, except for LLOQ which is acceptable in the range of 80–120% and within 20% deviation, respectively.

Carryover

Carryover was tested by sequentially injecting a blank standard solution MeOH/ACN (50/50, v/v) after monitoring an upper limit of quantification (ULOQ) sample. The signal response in the blank standard solution at the retention times of the metabolites and IS should be less than 20% of the mean response of the LLOQ sample.

Stability

Stability of each analyte was tested by comparison of the QC concentrations under different conditions with the freshly prepared sample concentrations. Short-term storage (6 h at room temperature) stability, long-term storage (30 days at − 80 °C) stability, autosampler storage (4 °C for 12 h) stability, and three freeze-thaw cycles stability of QC samples were evaluated by analysis of samples at each concentration level of QCs (LQC, MQC, and HQC). Final QC samples were considered stable if the accuracy was within 85–115% of nominal concentrations and the RSD within 15%.

Matrix effect

Matrix effect was used to evaluate the enhancement and suppression of the ionization and was investigated at three QC levels and obtained by comparing the mean peak area ratio (analyte/IS) of the analytes in six replicates of post-extraction samples with the corresponding neat solution. The matrix effect was accepted if the ratio in the range of 85–115%.

Extraction recovery

Recovery experiment was carried out at three levels of QCs (LQC, MQC, and HQC). Moreover, the extraction recovery was calculated by the following equation: ER% = AS-R ∕ AR-R × 5 ∕ 4 × 100%, where AS-R is the peak area ratio of each analyte extracted from spiked samples (250-μL aliquots) to IS added in post-extraction samples and AR-R is the mean peak ratio of same amount analytes to IS spiked into the post-extraction samples (200-μL aliquots) which represent 100% recovery. Five replicates were included at each QC level.

Robustness

The robustness of a method is a measure of its capacity to remain unaffected by deliberate variations in method parameters, and it indicates reliability during normal usage. A Shimadzu Nexera UFLC system coupled to an MS-8040 triple quadrupole mass spectrometer (Shimadzu Co., Tokyo, Japan) equipped with ESI source was applied to assess robustness. Three critical parameters including flow rate (± 0.2 mL/min), mobile phase additive concentration (± 0.2 mM), and column temperature (± 2 °C) were selected to be examined using MQC samples. The design of robustness experiment based on the settings is given in Table 2. The nominal value is the retention time (RT) in condition1 and RT variation (%) was calculated by the ratio of RT in different conditions to the nominal value. In addition, another new ZIC-pHILIC column was investigated to evaluate method robustness.

Table 2 Settings of different conditions for robustness experiment
Full size table

Participants and sample collection

T2D patients (n = 34) and healthy volunteers (n = 38) were recruited at the Second Affiliated Hospital of Harbin Medical University (Harbin, China) between March 2017 and June 2017. The detailed clinical information is shown in Table 3. The morning fasting antecubital venous blood plasma samples were collected and stored at − 80 °C until use. All participants voluntarily gave written informed consent and joined this study. The study protocol was approved by the Medical Ethics Committee of the Second Affiliated Hospital of Harbin Medical University and was conducted according to the Helsinki Declaration of 1964, as revised in 1975.

Table 3 Clinical characteristics of 34 T2D patients and 38 healthy volunteers
Full size table

Statistical analysis

All statistical analyses were performed using SPSS statistics 22.0 software (IBM SPSS Inc., USA). Differences in clinical and laboratory continuous variables were compared using Student’s t test or Mann-Whitney test, depending on data normality. Categorical variables were compared using χ2 test. p < 0.05 was considered significant.

Results and discussion

Optimization of chromatographic conditions

In this study, we chose 13 metabolites included since they are degradation products of glutaminolysis. Targeted metabolites were compounds with highly polar and weakly retained on a reversed-phase column. Therefore, the optimization of LC condition was primarily focused on evaluating the performance of different columns with stationary phases spanning a reversed-phase (RP) column and two hydrophilic interaction liquid chromatography (HILIC) columns. It was found that the Agilent ZORBAX SB-Aq column failed to provide effective retention for most targeted analytes, not to mention separating them well. The Waters XBridge® BEH Amide column as an alternative mode provided good separations for organic acids (except Cit and Iso-cit) but poor peak shapes for amino acids. Utilization of HILIC has grown in popularity for retaining and separating polar compounds [23]. The preliminary retention and separation of all 13 analytes were obtained on a Merck ZIC®-pHILIC column that carried densely bonded, zwitterionic functional groups with a charge balance of 1:1. Notably, the separation (resolution, R = 1.952) of isomers, Cit and Iso-cit, was also achieved on this column, probably due to their different solubility in acetonitrile (see Electronic Supplementary Material (ESM) Fig. S1).

To elute all analytes with high efficiency, combinations of different kinds of mobile phase additives, ionic strength, pH, and gradient were investigated. Based on the key chromatographic parameters, such as resolution, symmetry factor, and peak width, five aqueous solvents used as solvent B were evaluated, including 0.1% formic acid; 0.1% ammonium hydroxide; 5 mM, 10 mM, and 15 mM ammonium acetate; 5 mM ammonium formate; and 5 mM ammonium bicarbonate in water. Among these additives, 5 mM ammonium acetate gave all analytes the best chromatographic behavior. To examine the influence of pH, tests were conducted on a series of pH values from 3.5 to 9.5. It was found that the MS responses of most targeted metabolites enhanced as the alkalinity of the solvent B increased (see ESM Fig. S2), whereas the isomers Cit and Iso-cit could not be chromatographically separated when pH ≥ 7.5 (ESM Fig. S3). To keep a balance between the sensitivity and specificity, the pH value at 6.8 in solvent B was selected. Eventually, the flow gradient was optimized to achieve the elution of 13 metabolites and IS in 10 min and the total analysis time of a single run of 20 min.

Optimization of tandem mass spectrometry conditions

MS/MS operation parameters were carefully optimized to obtained high selectivity and sensitivity. It was observed that four amino acids (Ala, Asp, Glu, and Gln) could be ionized both in positive and negative ESI modes, while other analytes could only be detected in negative mode due to their acidic nature. Therefore, to avoid switching between the two modes in a single run, the negative mode was employed for all metabolites as well as IS. By flow injecting a neat standard solution which contains 1 μg/mL of the individual analyte, [M-H]− was found the richest relative abundance as the precursor ion in the full-scan spectra. Then, the product ion of each analyte was obtained by optimizing collision energies, spray voltage, capillary temperature, sheath gas pressure, and auxiliary gas pressure.

Noticeably, when monitoring the fragment ion at m/z 111.0 which was used by reported work [24], the difference in MS response between Cit and Iso-cit in human plasma was so great (~ 30 times) that the peak of Iso-cit was almost merged by the peak of Cit. After the optimization, we found a new fragment ion at m/z 172.9 providing similar MS responses to two isomers in the real sample (Fig. 1).

Fig. 1
figure 1

MRM chromatograms and possible fragmentations of Cit and Iso-cit in human plasma. Selected reaction monitoring transitions are am/z 191.0–111.0 and bm/z 191.0–172.9

Full size image

Method validation

The calibration curve of each analyte within the linear range exhibited good linearity (correlation coefficient, r ≥ 0.9911). The LLOQ, representative regression equations of the calibration curves, and the correlation coefficients relating to the 13 metabolites are detailed in Table 1. The intra-day and inter-day precisions and accuracies at four concentration levels are summarized in Table 4. For low, middle, and high QC levels, the accuracy was between 89.24 and 108.6% and the precision of each analyte was less than 12.96%. With regard to the LLOQ, the precisions and accuracies for all metabolites were within 18.38% and between 92.03 and 113.4%, respectively. The robustness result (Fig. 2) of RT variation was between 95.2 and 105.0% in each condition, and the resolution of the isomer was at least 1.615 (see ESM Fig. S5), which confirmed the robustness of the method. Moreover, the new column exhibited similar chromatographic retention behaviors to the former. These results all indicated that the method was reliable, accurate, and reproducible.

Table 4 Intra- and inter-day precisions and accuracies of analytes
Full size table
Fig. 2
figure 2

Retention time variation of targeted metabolites in each condition

Full size image

By sequentially injecting the blank solution MeOH/ACN (50/50, v/v) after the ULOQ sample, no obvious MS response was observed in the chromatographic profiles at the RT of each analyte (see ESM Fig. S4), which indicated no carryover from residues in samples. The stability tests were designed to cover the anticipated conditions of handling of the real samples. As shown in ESM Table S1, the variations within 15.30% demonstrated that the analytes in plasma samples were stable when stored under the conditions, including short-term, long-term, autosampler, and three freeze-thaw cycles. Any possible matrix effect was also studied and no significant suppression or enhancement was observed (see ESM Table S2). Recovery of QC samples ranged from 85.25 to 97.65% (see ESM Table S3), indicating the reproducibility and efficiency of sample preparation processes.

Method application

The validated method was applied to analyze the plasma samples of T2D patients and healthy controls. The typical extracted-ion chromatogram (EIC) of each analyte in human plasma is shown in Fig. 3. The quantitative results are summarized in Table 5. Student’s t test revealed that 7 of the 13 metabolites in glutaminolysis showed significant differences in plasma levels between T2D and control samples, of which p values were less than 0.05. Among the 7 differential metabolites, the concentrations of Gln and Glu were upregulated in plasma samples of T2D patients, confirming the results with the previous report [10]. Our novel finding is that the effect of T2D on glutamine metabolism is further extended to the whole TCA cycle, resulting in lower levels of Fum and Suc and higher levels of Cis-act and Iso-cit (Fig. 4).

Fig. 3
figure 3

The typical extracted-ion chromatogram (EIC) of each analyte obtained under optimal LC-ESI-MS/MS conditions in human plasma

Full size image
Table 5 Results of LC-MS/MS analysis of plasma samples from healthy volunteers and patients with T2D
Full size table
Fig. 4
figure 4

A diagram for the significantly difference in glutaminolysis pathway of T2D patients compared with healthy controls. Metabolites with significant differences are represented by green boxes

Full size image

In TCA cycle, aconitase is an essential enzyme catalyzing the stereo-specific isomerization of Cit to Iso-cit via Cis-act [25]. Prior efforts have found that aconitase might be a potential therapeutic target for diabetes by normalization of aconitase activity in the lung [26]. However, the published LC-MS-based methods could not completely separate Cit and Iso-cit [27, 28], causing the role of aconitase to be ignored. In this work, abnormalities in concentrations of not only 11 polar metabolites but also a pair of isomers were distinguished and observed by the proposed method, showing its great value in exploring biomarkers and metabolic mechanisms of diseases involving changes in glutaminolysis.

Conclusions

In this study, a novel LC-MS/MS method using a pHILIC column was developed and validated for the determination of 13 metabolites in glutaminolysis in human plasma. The key parameters of separation and detection were systematically investigated and optimized. Compared with previously reports, key advantages of this method included are easy sample pre-treatment (required only 1 precipitation step), fast chromatographic separation (10 min), and wide dynamic range (0.05–800 μg/mL). This method also successfully achieved the separation and quantification for the isomers of Cit and Iso-cit. Validation results confirmed the robustness, reliability, and accuracy of this method. Moreover, the developed method was successfully applied to study the alters of glutaminolysis in T2D patients and healthy controls, demonstrating its capability in biomarker discovery and mechanism exploration.