Measurement of kynurenine pathway metabolites by tandem mass spectrometry

Highlights • A multiplex LC-MS/MS method was developed for the measurement of kynurenines.• Inta- and inter-assay imprecisions were less than 12% for all metabolites.• Total run time was 5 min.• The recoveries ranged between 94% and 105% and matrix effects were less than 6.7%.• The method has simple pretreatment, and extraction recovery and sensitivity within acceptable ranges.


Introduction
Tryptophan and its biological molecules have essential roles in metabolism. It is metabolized by four major biological pathways (serotonin, indole-3-pyruvate, kynurenine, and tryptamine). However, most of the tryptophan in the body is converted to kynurenine [1]. Tryptophan is converted to kynurenine by two enzymes (tryptophan 2,3-dioxygenase and indolamine 2,3-dioxygenase), followed by serial enzymatic reactions via riboflavin and pyridoxine as cofactors into kynurenic acid, anthranilic acid and their 3-hydroxy forms [2]. In addition to its structural participation in protein synthesis and being an essential amino acid, it has recently been popular for its role in psychiatric diseases [3]. methods [5]. High-performance liquid chromatography (HPLC) methods using UV detectors are among the preferred methods for clinical laboratories. However, it is a common problem that molecules with similar structures, mainly endogenous in the body, cause interference in analysis. Extended analysis time in UV detection is another significant problem [6]. The electrospray ionization technique (ESI) paired with liquid chromatography-mass spectrometry (LC-MS/MS) is a very sensitive and specific method that allows for multiplex analysis on many different sample types [7]. However, the tandem mass spectrometric methods reported to date are disadvantaged due to their low sensitivity, accuracy, extraction recovery and precision, and limited number of metabolites to be measured. In addition, they have time-consuming, costly pretreatment steps, extended analysis time, and require large amounts of samples [8][9][10][11][12]. The kynurenine pathway has been demonstrated to be involved in many diseases and disorders, including Alzheimer's disease, amyotrophic lateral sclerosis, Huntington's disease, AIDS dementia complex, malaria, cancer, depression, and schizophrenia [13]. Considering the comprehensive biological and pathological mechanisms of these metabolites, there is a need for reliable methods that allow for the measurement of these metabolites to be accurate, sensitive, and multiplex. Furthermore, it is possible to determine the levels of molecules in different metabolic pathways of the metabolism simultaneously with the sequential mass spectrometric method. Obtaining the levels of compounds in this pathway in a single chromatogram might give a broader overview of disease pathology [14,15]. Therefore, this study aimed to develop and validate a multiplex, sensitive, cost-effective, practical, accurate, and reproducible tandem mass spectrometric method for the measurement of kynurenine pathway metabolites.

LC gradient set up and MS methods
The separation of analytes was achieved using the C18 HPLC column (Phenomenex, 50 mm × 4.6 mm, part no: 00B-4041-E0), and the gradient elution of mobile phases A and B consisted of 100 % HPLC grade water with 0.1 % formic acid and 100 % acetonitrile with 0.1 % formic acid. The time-volume changes in the gradient program were as follows: 0-0.1 min., 25 % B; 1-2 min., 50 % B; 2-4 min., 100 % B; 4-5 min., 25 % B. Initially, isocratic elution was attempted. However, gradient elution was preferred as the metabolites could not be fully separated due to their physicochemical similarities. Different flow rates and run times from 1 min to 30 min were tried. The best peak shape and separation were achieved at a 0.8 mL / min flow rate and a run time of 5.0 min. The Shimadzu HPLC (Kyoto, Japan) system, coupled with a mass spectrometer, had a pump unit (LC-20 AD), an autosampler (SIL-20 AC HT), and an online degasser (DGU-20A3) unit. LC gradient for kynurenine pathway metabolites were shown in Fig. 1.
Kynurenine pathway metabolites were detected using an API 3200 (Applied Biosystems / MDS Sciex) tandem mass spectrometer set to ESI in positive multiple reaction monitoring (MRM) mode. The metabolites were identified by infusing pure standards. The Q1-Q3 ion transitions for tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, 3- The main method optimization parameters for the mass spectrometer were set as follows: ionspray voltage of 5000 V; ion source temperature of 350 • C; gas1 and gas2 at 50 psi; curtain gas at 20 psi; and collision gas at 6 psi. Declustering entrance, collision cell exit potential, collision energy, entrance potential and dwell time for each analyte are shown in Table 1. Data analysis was processed by SCIEX Analyst® 1.6.2 Software, and analyte concentrations were calculated using the internal standard method.

Sample preparation
Serum metabolite concentrations were quantified using a modified method [16]. Briefly, 300 µL of the sample, 100 µL of donepezil (internal standard, 100 ng / ml), and 1000 µL of acetonitrile with 0.1 % (v/v%) formic acid were vortexed for 30 s. The mixture was then centrifuged at 2000 × g for 10 min, and the supernatants were evaporated with nitrogen gas at 37 • C. The residues were dissolved with a 200 µL acetonitrile:water (25:75; v/v) solution, including 0.1 % formic acid, and 30 µL of the mixture was injected into the system.

Method validation
The tandem mass spectrometric method was primarily validated according to the Clinical and Laboratory Standards Institute (CLSI) C62-A: Liquid Chromatography-Mass Spectrometry Methods protocol [17]. Method validation parameters, such as linearity, precision, stability, recovery, carry-over, matrix effect, selectivity, and specificity, have been determined.

Preparation of standard and working solutions
The calibration standards and working solutions were prepared with the help of a surrogate matrix [18]. Since the kynurenine pathway metabolites are endogenous compounds, a surrogate matrix was adopted to calibrate and prepare working solutions [19]. Different surrogate matrices, such as phosphate-buffered saline (PBS) solution (0.01 M phosphate buffer, 0.0027 M potassium chloride, 0.137 M sodium chloride, pH 7.4 at 25 • C) including 1 % bovine serum albumin (BSA), water, acetonitrile, and methanol were evaluated as candidate surrogate matrices. The PBS solution containing 1 % BSA was preferred as the surrogate matrix for validation studies because it had equivalent protein content and ionic strength to human plasma [20]. All solutions used for calibration and validation were freshly prepared by spiking analytes into the PBS solution, including 1 % BSA, and stored at + 4 • C. Standards at three different concentration levels (low, medium and high) were prepared by spiking into the PBS solution including 1 % BSA for each metabolite to demonstrate the equivalence between the selected surrogate matrix (1 % BSA solution) and real matrix (human serum). Therefore, a blank matrix was prepared using serum samples from at least six different blood donors for each metabolite, and baseline metabolite concentrations were measured. The percent of equivalence was calculated with the following formula [21]: Response serum spike − Response serum blank Response surrogate spike Response serum spike was the peak area of the analyte-spiked serum pool. Response serum blank was the peak area of the analyte in the serum pool, and the surrogate response spike was the peak area of the analyte-spiked surrogate matrix. All metabolites' equivalency (%) ranged from 90 to 104 %. Consequently, the PBS solution, including 1 % BSA and human serum, had a similar matrix effect. Stock solutions of tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, and 3-hydroxykynurenine were prepared at final concentrations of 25, 10, 10, 5, and 4 mg/mL, respectively. The calibration standards were prepared at concentration ranges of 48.8-25,000, 0.98-500, 1.2-5000, 1.2-5000, and 0.98-250 ng/mL for tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, and 3-hydroxykynurenine via serial dilution of the stock solutions with the surrogate matrix. A 100 ng/ mL internal standard (donepezil) solution was prepared in methanol. Prepared solutions were stored at 2 to 8 • C until analysis.

Statistical analysis
SPSS statistical software package version 21.0, EP Evaluator Release 8 version (Data Innovations, South Burlington, VT), and Excel (2010) were used for the statistical evaluation, and a p-value less than 0.05 was accepted as statistically significant.

Linearity
The linearity study followed the CLSI EP06-A protocol [17]. Calibration standards for tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, and 3-hydroxykynurenine were prepared as described above. Calibration plots were set up by plotting the analyte/internal standard peak area ratios against the nominal metabolite concentrations. The linearity results were assessed using linear regression analysis. The correlation coefficients of the tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, and 3-hydroxykynurenine calibration curves were found to be 0.9994, 0.9996, 0.9984, 0.9968 and 0.9976, respectively. The developed method was found to be linear across the entire range of the prepared calibration standards, respectively.

LOD (Limit of detection) and LLOQ (Lower limit of quantitation)
There are various approaches to determining the LLOQ value. The first method is based on determining the lowest analyte concentration that can be quantified with satisfactory precision and accuracy. According to FDA guidelines, data from at least five replicates of the analyte-spiked sample in at least three independent runs should be used. Calculated imprecision values should be ≤ ±20 %, and a deviation in accuracy should be ≤ ±20 % [22]. Therefore, 20 samples were studied for five consecutive days to determine the LLOQ values, and intra-and inter-assay imprecisions and accuracy values were calculated. The accuracy study was performed by analyzing 20 replicates of a pre-treated low-level analyte-spiked PBS solution containing 1 % BSA, and the bias % values were calculated using the following formula: Bias %=.

Intra-and inter-assay precision
This study was performed according to the CLSI EP05-A3 protocol [17]. Precision studies were carried out at low, medium, and high concentration levels determined for each metabolite throughout the calibration curve (Table 2). Inter-day imprecision was calculated based on 20 measurement results by analyzing four replicates per day for five consecutive days for each concentration level. Intra-day imprecision was calculated based on measuring 40 samples within a day. Then, imprecision values were determined using the following formula: The precision study results are presented in Table 2.

Stability
The stability study followed (Table 3) the CLSI EP25-A protocol [17]. The effects of the freeze-thaw process and long-term storage (at − 20 • C for 45 days) were evaluated. A long-term stability study was performed by analyzing the metabolite concentrations in the blank matrix obtained from 30 different serum samples on the collection day and then measuring the metabolite concentrations after storage at − 20 • C on the 15th, 30th, and 45th days. The freeze-thaw stability was evaluated during 4 freeze-thaw cycles following the analyses of analyte concentrations on the collection day. The bias% values were calculated compared to the measured metabolite levels on the collection day (expected value) via the following formula: All metabolites were found to be stable during 45 days and four freeze-thaw processes.

Recovery and matrix effects
The recovery study followed the CLSI EP34 protocol [17]. The recovery study used three different concentration levels of each analyte (Table 4). The results were calculated using the following formula: Re-covery%=[(C2 − C0)/C1] × 100.
C2 was the measured analyte concentration after spiking, C0 was the initial analyte concentration of the solution, and C1 was the added known concentration of standard.
The matrix effect is generally evaluated with one of two standard methods: post-column infusion or post-extraction spiking. In this study, the second method was preferred, as reported by Chambers et al. [24]. In this context, the low, medium, and high-level analyte response in neat solutions (such as water, mobile phase mixture, acetonitrile, and methanol) was compared with the analyte response in the spiked matrix at equal levels after pretreatment. Firstly, three different concentration levels were obtained by spiking low, medium, and high (Table 4) level analytes into a mixture of water and methanol (50:50, v/v%) to prepare the neat solution. Secondly, a surrogate matrix PBS solution containing 1 % BSA was pretreated, and after the residues were dissolved with a mobile phase mixture, equal concentration levels to the neat solution were prepared. The matrix effect values were calculated using the following formula: ME% = (Mean post-extracted peak area / Mean unextracted peak area) × 100.
The results of the recovery and matrix effect studies are shown in Table 4.
Results of the matrix effect study. Matrix effects were less than 6.7 % for all metabolites.

Carry-over
The carry-over study analyzed low-and high-level samples in the order stated in CLSI EP10-A3 [17]. Low concentration levels were fixed in PBS, including BSA for tryptophan, kynurenic acid, kynurenine, 3- Table 2 Inta-and inter-assay precision study results of kynurenine pathway metabolites.
The calculated results from the tandem mass analysis were recorded onto the EP Evaluator Release 8 program (Data Innovations, South Burlington, VT) and carry-over results were determined for each analyte. The EP Evaluator program determined the acceptability criteria based on CLSI protocol EP10-A3 guidelines.
The carry-over values are shown in Table 5.

Application of the method for patient samples
The validated method was applied to clinical samples obtained from patients with chronic kidney disease (CKD) who had been undergoing hemodialysis for at least three months in the dialysis unit of Selcuk University Faculty of Medicine Hospital. Two milliliters of blood were drawn into collection tubes, and then centrifuged at 2000 g for 10 min. Patient samples were kept at − 80 • C until the day of analysis. This study was approved by Selcuk University's local ethics committee (Number: 2020/582, Date: 30/12/2020). The chromatograms of kynurenine pathway metabolites obtained from a patient serum sample before and after dialysis are shown in Figs. 2 and 3, respectively.
The kynurenine pathway metabolite levels and routine laboratory parameters measured in post-and pre-dialysis serum samples are presented in Table 6.

Discussion
Tryptophan is an essential amino acid that is important for living organisms. Less than 1 % of the available tryptophan is used for protein synthesis [25]. Tryptophan's main metabolic route is the kynurenine pathway, through which approximately 90-95 % of tryptophan degrades into nicotinamide adenine dinucleotide (NAD + ). The degradation of tryptophan by the kynurenine pathway causes the generation of various intermediates with neuroactive, immunoregulatory, oxidant, or antioxidant effects, such as kynurenine, kynurenic acid, 3-hydroxykynurenine, and 3-hydroxyanthranilic acid before NAD + [26]. Since kynurenine pathway metabolites have physiologically important and sometimes opposing effects, their levels are tightly regulated. Imbalances in the kynurenine pathway metabolite levels can cause various pathologies, such as Alzheimer's, Huntington's, Parkinson's, schizophrenia, depression, bipolar disorder, rheumatoid arthritis, and cancer [27][28][29][30]. Therefore, the importance of reliable, accurate, rapid, and multiplex measurement methods for kynurenines has increased. However, quantifying tryptophan and metabolite levels is challenging due to their low stability, low blood concentrations, and presence of interferences in biological matrices. Since each of the kynurenine pathway metabolites has different physiologically characteristic functions, it is vital that most of the metabolites are able to be quantified by a simultaneous measurement method. However, an important aspect of the problem is that the physiological concentrations of metabolites vary over a wide range; for example, tryptophan is found in micromolar concentrations, while other kynurenine pathway metabolites are found in nanomolar concentrations. Another critical problem is the need for selective and specific methods because these metabolites have similar physical properties and structures [6,31]. While HPLC-based techniques such as HPLC-UV [32], HPLC-FD [33], and HPLC-ECD [34,35] have been commonly used in the measurement of kynurenine pathway metabolite levels, these methods have some essential disadvantages. For example, HPLC-UV methods are disadvantageous due to their low sensitivity, specificity, and selectivity. The risk of interference from matrix components is higher in HPLC-UV methods. The relatively long analysis time is another problem with these methods. Although HPLC-FD methods have sufficient sensitivity, they have disadvantages due to complex and time-consuming pretreatment steps that require derivatization. Although HPLC-ECD methods also have high sensitivity, the main disadvantage is the low reproducibility caused by electrode clogging and loss of selectivity [6,31]. However, LC-MS/MS allows for the simultaneous measurement of kynurenine pathway metabolites in a single experiment. This method has advantages such as good separation ability, high selectivity, specificity, and sensitivity; it requires a small amount of sample; and has a short run time and practical pretreatment steps. Various LC-MS/MS methods are reported for measuring various kynurenine pathway metabolites [2,[7][8][9][10][11][12].
Möller et al. revealed a tandem mass method for simultaneously measuring six different kynurenine pathway metabolites, including tryptophan, kynurenine, kynurenic acid, anthranilic acid, 3-hydroxyanthranilic acid, and quinolinic acid. Intra-and inter-assay imprecision values of the method ranged from 1.0 % to 9.9 %, and accuracy values ranged from 75 % to 107 %. The mean recovery values were between 96 % and 101 %, and the LLOQ values of tryptophan, kynurenine, kynurenic acid, anthranilic acid, 3-hydroxyanthranilic acid, and quinolinic acid were 0.0352 μM, 0.035 μM, 0.035 μM, 0.047 μM, 0.084 μM, and 0.096 μM, respectively. The method had advantages such as sufficient sensitivity, good recovery, and accuracy; however, its disadvantages were the required high amount of sample (800 µL), time-consuming pretreatment steps including solid phase extraction (SPE), and long run time (12 min) [8]. Fuertig et al. reported a tandem mass method based on protein precipitation for the multiplex analysis of tryptophan, kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, quinolinic acid, picolinic acid, kynurenic acid, xanthurenic acid, serotonin, dopamine, and neopterin levels. Intra-and inter-assay imprecision of this method for all analytes ranged from 1 % to 23 %, while accuracy values ranged from 42 % to 116 %. The LOQ values ranged between 0.25 and 2.5 nM. The significant advantages of the method were excellent accuracy, precision (except for 3-hydroxyantranilic acid), short run time (5.5 min), and sufficient sensitivity; however, it had time-consuming processing steps requiring a 30 min incubation at − 20 • C and evaporation of samples [9]. Huang et al. reported an LC-MS/MS method based on protein precipitation with trifluoroacetic acid to quantify plasma tryptophan and kynurenine levels. The intra-and inter-assay imprecisions of the method were less than 7.1 %, and the accuracy values were between 89 % and 108 %. The LOQ values of tryptophan and kynurenine were 1250 ng/mL and 62.5 ng/mL, respectively. The analysis time of each sample was 4.5 min [10]. Similarly, Miller et al. reported a tandem mass spectrometric method for quantifying tryptophan and kynurenine levels based on precipitation with trifluoroacetic acid. The extraction recovery of the method was between 89 % and 95 %, and the intra-and inter-assay imprecision values were less than 7 %. The run time was approximately 4.1 min. This method had a short analysis time, simple sample preparation procedures, good precision, extraction recovery, and accuracy; however, its low sensitivity and lack of other metabolites were Table 5 The carry-over study results of kynurenine pathway metabolites.

Analytes
Low-low results (ng/ ml) simultaneously determining tryptophan, kynurenine, kynurenic acid, xanthurenic acid, and 5-hydroxytryptamine in human plasma. The mean recovery values ranged between 82 % and 107 %, and intra-and interassay imprecision values were less than 13.2 %. Simple pretreatment steps including protein precipitation and evaporation; short run time (3.5 min); good precision; and extraction recovery are the main advantages of this method [16]. The number of methods for quantifying different kynurenine pathway metabolites may still increase. We have developed a tandem mass spectrometric method for the multiplex analysis of serum tryptophan, kynurenine, kynurenic acid, 3-hydroxykynurenine, and 3-hydroxyanthranilic acid levels. The pretreatment steps were based on protein precipitation and evaporation. LLOQ values were determined as 48.8, 1.96, 2.4, 2.4, 1.96 ng/mL for tryptophan, kynurenic acid, kynurenine, 3-hydroxyanthranilic acid, and 3-hydroxykynurenine respectively. The intra-and inter-day imprecisions were less than 12 %, and the accuracy values ranged between 88 % and 112 %. The recovery values ranged between 94 % and 105 %, and the matrix effect values were less than 6.7 % for all analytes. This method was advantageous regarding sufficient sensitivity, simple sample processing, short run time (5 min), good extraction recovery, accuracy, and precision. The second important aspect of our study was the clinical application of a validated method for measuring kynurenine pathway metabolite levels in pre-and post-dialysis serum samples from patients with CKD who had undergone hemodialysis. There are few studies investigating kynurenine pathway metabolite levels in patients with CKD. Debnath et al. reported mean serum tryptophan, kynurenine, 3-hydroxykynurenine, kynurenic acid, and quinolinic acid levels in patients with CKD

Table 6
Pre-and post-hemodialysis levels of kynurenine pathway metabolites, creatinine, urea, potassium and eGFR levels in patients with chronic renal failure. found the mean serum tryptophan and kynurenine levels as 5295 and 844.5 ng/mL in hemodialysis patients, respectively [40]. The median serum values of tryptophan, kynurenine, kynurenic acid, 3-hydroxykynurenine, and 3-hydroxyanthranilic acid levels were determined as 10,530, 1,100, 218, 17.6, and 25.4 ng/mL in pre-dialysis blood samples respectively in our study. The levels of these metabolites in post-dialysis blood samples were 4,560, 664, 135, 7.4, and 12.8 ng/mL respectively. Our results were consistent with previous studies; however our findings showed that serum metabolite levels decreased approximately twofold post-dialysis compared to pre-dialysis samples.

Conclusions
A validated LC-MS/MS method was developed to quantify concentrations of metabolites in the kynurenine pathway. It was then applied to clinical samples obtained from hemodialysis patients. The primary advantages of the method were its short run time, simple pretreatment steps, cost-effectiveness, good extraction recovery, high precision, sufficient sensitivity, and ability to provide multiplex analysis with enough selectivity. The metabolite levels measured in hemodialysis patients were consistent with those of previous studies, meaning that the developed method can be reliably used in clinical samples.