A reliable LC-MS/MS method for the quantification of natural amino acids in mouse plasma: Method validation and application to a study on amino acid dynamics during hepatocellular carcinoma progression

Abstract

A simple and fast LC-MS/MS method was developed and validated for the quantification of 20 proteinogenic L-amino acids (AAs) in a small volume (5 μL) of mouse plasma. Chromatographic separation was achieved on an Intrada Amino Acid column within 13 min via gradient elution with an aqueous solution containing 100 mM ammonium formate and an organic mobile phase containing acetonitrile, water and formic acid (v: v: v = 95: 5: 0.3), at the flow rate of 0.6 mL/min. Individual AAs and corresponding stable-isotope-labeled AAs internal standards were analyzed by multiple reaction monitoring (MRM) in positive ion mode under optimized conditions. Method validation consisted of linearity, sensitivity, accuracy and precision, recovery, matrix effect, and stability, and the results demonstrated this LC-MS/MS method as a specific, accurate, and reliable assay. This LC-MS/MS method was thus utilized to compare the dynamics of individual plasma AAs between healthy and orthotopic hepatocellular carcinoma (HCC) xenograft mice housed under identical conditions. Our results revealed that, 5 weeks after HCC tumor progression, plasma L-arginine concentrations were significantly decreased in HCC mice while L-alanine and L-threonine levels were sharply increased. These findings support the utilities of this LC-MS/MS method and the promise of specific AAs as possible biomarkers for HCC.

1. Introduction

Liver cancer is the sixth most commonly diagnosed cancer and the fourth leading cause of cancer death worldwide in 2018 [1]. Accounting for 80% of total liver cancer cases, hepatocellular carcinoma (HCC) is the most common type of primary liver malignancy [1-3] that usually occurs with the progression of chronic liver diseases, particularly among patients with cirrhosis and chronic hepatitis B virus or hepatitis C virus infections [3-5]. Overall the ratio of mortality to incidence of liver cancer was revealed to be 0.95 [5], due to the lack of earlier diagnosis as well as treatment options at late stages [6, 7]. By contrast, sensitive and specific diagnosis of HCC at early stage opens more treatment options such as resection, liver transplantation and ablation which was able to improve overall survival rate of patients with chronic liver diseases [6]. Therefore, search for new biomarkers of HCC may not only improve our understanding of etiology of HCC but also allow early treatments to enhance clinical outcomes [8].

Liver plays an important role in the catabolism of amino acids (AAs) that contributes to a wide range of biochemical processes such as biosynthesis of proteins, nucleotides, and lipids essential for cell proliferation [9, 10]. Particular bio-fluids from patients with liver diseases [11-13] have been shown to exhibit significantly-changed AA concentrations. Recent studies using novel metabolomics technologies have also demonstrated the alterations of blood or urine metabolomic profiles, including specific groups of AAs or individual AAs in diagnosed HCC patients [14-17]. However, these clinical investigations are featured by a cross-sectional study design comparing samples collected from patients at diagnosis with samples from healthy subjects, which may not be indicative of HCC progression for individual patients. In addition, food and drink greatly influence the levels of AAs, especially those essential AAs such as threonine, tryptophan, and lysine, etc., whereas unified food and drink for enrolled human subjects is usually impractical. Therefore, studying AA dynamics during HCC progression among experimental animals under unified food, drink, and housing conditions could be more helpful to identify potential AA biomarkers for HCC and offer insights into understanding HCC metabolism and possible mechanisms.

The assessment of AAs in HCC and broad areas of cancer metabolism research rely on reliable, selective, and accurate analytical methods. There are indeed some challenges for targeted analyses of natural AAs because AAs are present intrinsically in complex biologic matrices at variable concentrations and AAs have low molecule weights and strong polarities. A number of detection methods, including ultraviolet (UV) [18-20], fluorescence (FL) [21-24], mass spectrometry (MS) [25-27], and electrochemical detection [28-30], have been employed for the determination of AAs in biological samples, following the separation by high-performance liquid chromatography (HPLC) or capillary electrophoresis (CE). Analysis of AAs by UV and FL detection generally requires a derivatization step, which improves the separation and sensitivity of detection but leads to tedious sample preparation and long analytical time [18, 22]. Efforts have been made recently for the analysis of underivatized AAs using tandem mass spectrometry (MS/MS) detection [25-27, 31-33] that often offer high selectivity and sensitivity, in addition to simple sample preparation. Nevertheless, accurate quantification of underivatized AAs still poses much difficulties to modern LC-MS/MS technology, largely due to the poor retention of natural AAs on reversed-phase (RP) columns, potential and variable effects of a wide variety of biological matrices as well as the lack of proper matrices free of analytes or samples with known concentrations of analytes [32, 34].

The aim of this study was to develop a fast, accurate, and precise LC-MS/MS method for simultaneous, targeted analyses of underivatized AAs in biologic samples. Separation of AAs was accomplished by hydrophilic interaction chromatography, which offered better retention and peak symmetry for all analytes. To ensure accuracy and minimize matrix effects, uniformly [¹³C, ¹⁵N]-stable-isotope-labeled AAs were added into each sample as internal standards (ISs) for individual AAs prior to LC-MS/MS analysis. This method was fully validated for simultaneous analysis of AAs in a small volume (5 μL) of mouse plasma sample, and thus applied to the investigation of blood AA dynamics during orthotopic HCC progression in mice. Our studies revealed a significant decrease in plasma arginine concentrations and sharp increase in alanine and threonine levels as HCC progresses. These findings may be helpful for the exploration of possible AA biomarkers for HCC as well as investigations of new pathways and treatments for HCC.

2. Materials and Methods

2.1. Chemicals

Standard L-phenylalanine (Phe), L-tyrosine (Tyr). L-threonine (Thr), L-lysine (Lys), L-alanine (Ala), L-proline (Pro), L-isoleucine (Ile), L-glutamic acid (Glu), L-arginine (Arg), L-histine (His), L-cysteine (Cys) and L-tryptophan (Trp) were purchased from Thermo Fisher Scientific, Inc. (Waltham, MA, USA). L-glutamine (Gln), L-asparagine (Asn), L-aspartic acid (Asp), L-valine, glycine and L-serine (Ser) were bought from Tokyo Chemical Industry CO., LTD. (Portland, OR, USA). L-leucine (Leu), L-methionine (Met), and [¹³C, ¹⁵N]-stable-isotope-labeled, cell free AA mixture (Product No: 767964) were purchased from Sigma-Aldrich (St Louis, MO, USA), in which the concentration of each isotope-labeled AA was: Asp (60 mM), Thr (35 mM), Ser (35 mM), Glu (40 mM), Pro (20 mM), Gly (100 mM), Ala (100 mM), Val (40 mM), Met (10 mM), Ile (30 mM), Leu (45 mM), Tyr (10 mM), Phe (16 mM), His (5mM), Lys (15 mM), Arg (10 mM), Gln (20 mM), Asn (20 mM), Trp (20 mM), and Cys (20 mM). HPLC-grade acetonitrile (ACN), methanol (MeOH) and Optima water were purchased from Thermo Fisher Scientific. All other reagents and chemicals were of analytical grade.

2.2. LC and MS conditions

LC-MS/MS analysis was conducted on a Shimadzu Prominence Ultra-Fast Liquid Chromatography system consisting of binary pumps, an on-line degassing unit, an autosampler, and a column oven (Shimadzu Corporation, Kyoto, Japan), which is coupled with an AB Sciex 4000 QTRAP mass spectrometer consisting of an electrospray ionization (ESI) source (AB SCIEX, Framingham, MA, USA). Chromatographic separation was achieved on an Intrada Amino Acid column (50 × 3 mm, 3 μm; Imtakt, Portland, OR, USA) maintained at 35 °C, at a flow rate of 0.6 mL/min. The mobile phases consisted of Solution A (100 mM ammonium formate in water) and Solution B (acetonitrile: water: formic acid, v: v: v = 95: 5: 0.3). A gradient elution was optimized for the separation of individual AAs: 0-3.0 min, 92%-88% Solution B; 3.0-6.4 min, 88%-70% Solution B; 6.4-6.5 min, 70%-0% Solution B; 6.5-10 min, 0% Solution B; 10-10.1 min, 0-92% Solution B; 10.1-13 min, 92% Solution B, with a total run time of 13 min. The ion source was operated in positive mode under an optimal condition: curtain gas, 25 psi; nebulizer gas, 40 psi; auxiliary gas, 45 psi; ion spray voltage, 1500 V; and temperature, 600 °C. Optimal multiple-reaction monitoring (MRM) transitions were further identified for the analyses of individual AAs as well as corresponding isotope-labeled ISs (Table 1). Data acquisition and analysis were all performed with Analyst 1.6.3 software (AB SCIEX).

Table 1.

Optimized MS conditions for the analyses of individual AAs and corresponding isotope-labeled internal standards that are coeluted.

AA & abbreviation		DP (V)	CE (V)	MRM of AA	IS-AA (13C, 15N)	MRM of IS	Retention time (min)
L-Phenylalanine	Phe	42	20	166.1→120.1	IS-Phe	176.1→129.2	2.83-3.53
L-Tryptophan	Trp	45	25	205.1→146.0	IS-Trp	218.1→156.2	2.97-3.67
L-Leucine	Leu	38	36	132.1→43.1	IS-Leu	139.1→46.2	3.09-3.79
L-Isoleucine	Ile	45	25	132.1→69.1	IS-Ile	139.1→74.1	3.32-4.02
L-Methionine	Met	44	15	150.1→104.1	IS-Met	156.1→109.1	3.50-4.20
L-Proline	Pro	50	23	116.0→70.1	IS-Pro	122.0→75.2	3.57-4.27
L-Tyrosine	Tyr	44	19	182.1→136.0	IS-Tyr	192.1→145.0	3.79-4.49
L-Valine	Val	40	16	118.0→72.0	IS-Val	124.0→77.0	3.86-4.56
L-Cysteine	Cys	35	20	122.0→76.0	IS-Cys	126.0→79.1	4.17-4.87
L-Alanine	Ala	44	18	90.0→44.0	IS-Ala	93.8→47.2	5.19-5.89
L-Threonine	Thr	42	16	120.1→74.0	IS-Thr	125.1→78.0	5.24-5.94
L-Glutamic acid	Glu	40	22	148.0→84.0	IS-Glu	154.0→89.0	5.31-6.01
L-Aspartic acid	Asp	40	15	134.1→74.1	IS-Asp	139.1→77.1	5.60-6.30
L-Glycine	Gly	43	5	76.1→76.1	IS-Gly	78.8→78.8	5.72-6.42
L-Glutamine	Gln	40	25	147.1→84.0	IS-Gln	154.0→89.0	5.94-6.64
L-Serine	Ser	45	18	106.0→60.1	IS-Ser	109.9→63.1	5.95-6.65
L-Asparagine	Asn	40	22	133.1→74.0	IS-Asn	139.1→77.1	6.05-6.75
L-Histidine	His	45	19	156.1→110.0	IS-His	165.1→118.0	8.42-9.12
L-Lysine	Lys	40	25	147.1→84.0	IS-Lys	155.1→90.0	8.49-9.19
L-Arginine	Arg	50	33	175.1→70.1	IS-Arg	185.1→75.1	9.24-9.94

2.3. Standard and quality control (QC) solutions

AAs were weighed accurately into volumetric flasks using an analytical micro balance (Mettler-Toledo, Switzerland) and dissolved in 50% methanol (i.e., methanol: water, v: v = 1: 1) to produce individual AA stock solutions which were stored at −20 °C before use. The stock solutions were diluted serially with 50% methanol to generate working solutions. Standard solutions were comprised of 5-250 μM of Phe and Met, 6.25-312.5 μM of Tyr, His, Trp, Pro, Arg and Val, 12.5-625 of Leu, Ile, Ser, Asp, Asn and Thr, 25-1250 μM of Ala, Glu, Cys and Lys, 50–2500 μM of Gly and Gln (Supplementary Table 1). Quality control (QC) solutions for method validation were prepared independently by following the same procedures, which represented low (QC1), medium (QC2), and high (QC3) concentrations, specifically, 8, 40 and 200 μM of Phe and Met; 10, 50 and 250 μM of Tyr, His, Trp, Pro, Arg and Val; 20, 100 and 500 μM of Leu, Ile, Ser, Asp, Asn and Thr; 40, 200 and 1,000 μM of Ala, Glu, Cys and Lys, 80, 400 and 2,000 μM of Gly and Gln, respectively (Table 2). Furthermore, [¹³C, ¹⁵N]-labeled cell free AA mixture was diluted 20,000 times with solvent (acetonitrile: water: formic acid, v: v: v = 80: 20: 1) as IS working solution.

Table 2.

LOD, calibration range, recovery and matrix effect values of individual amino acids.

Analyte	Calibration range (μM)	Coefficient (r2)	LOD (μM)	Recovery									Matrix effect
				QC1			QC2			QC3			Mean (%)	SD (%)
				Conc. (μM)	Mean (%)	SD (%)	Conc. (μM)	Mean (%)	SD (%)	Conc. (μM)	Mean (%)	SD (%)
Phe	5-250	0.9984	0.1	8	64.9	16.8	40	83.4	11.3	200	97.0	1.9	95.3	2.0
Trp	6.25-312.5	0.9996	0.625	10	83.9	9.7	50	84.9	13.0	250	94.7	2.9	96.8	4.7
Leu	12.5-625	0.9986	3.125	20	91.3	10.0	100	89.0	6.4	500	96.7	1.3	99.1	9.3
Ile	12.5-625	0.9988	1.25	20	95.5	15.0	100	94.1	3.5	500	98.5	1.0	100.6	0.8
Met	5-250	0.9990	1.25	8	88.5	18.9	40	83.9	10.1	200	97.0	2.2	105.1	3.5
Pro	6.25-312.5	0.9988	0.25	10	67.0	11.9	50	82.4	8.4	250	96.9	0.8	90.3	0.5
Tyr	6.25-312.5	0.9996	0.625	10	96.4	7.4	50	86.7	7.7	250	98.9	1.6	97.8	1.6
Val	6.25-312.5	0.9994	0.625	10	99.3	24.8	50	94.9	13.7	250	96.8	2.8	82.9	8.5
Cys	25-1250	0.9988	2.5	40	92.5	2.9	200	99.7	2.6	1000	98.1	2.7	109.0	0.5
Ala	25-1250	0.9996	2.5	40	88.1	9.9	200	84.3	9.6	1000	97.0	2.4	101.5	3.2
Thr	12.5-625	0.9964	1.25	20	95.4	30.3	100	87.8	14.3	500	92.6	5.4	100.8	9.0
Glu	25-1250	0.9922	1.0	40	98.2	18.3	200	91.8	3.9	1000	93.3	3.5	99.7	0.5
Asp	12.5-625	0.9942	3.125	20	95.6	3.7	100	100.1	3.2	500	99.6	3.7	100.6	2.3
Gly	50-2500	0.9894	25	80	74.7	15.0	400	103.9	12.0	2000	98.9	9.5	100.0	11.2
Gln	50-2500	0.9982	1.0	80	83.4	14.4	400	86.2	9.0	2000	99.0	3.8	113.4	2.5
Ser	12.5-625	0.9988	3.125	20	94.9	15.9	100	82.9	7.8	500	95.9	3.3	118.8	8.0
Asn	12.5-625	0.9980	3.125	20	95.7	18.1	100	94.6	2.5	500	100.4	1.5	101.8	2.1
His	6.25-312.5	0.9988	0.25	10	82.4	11.8	50	87.8	8.5	250	97.6	2.5	100.8	4.1
Lys	25-1250	0.9994	1.0	40	76.9	16.3	200	84.9	10.4	1000	97.7	2.7	103.9	0.9
Arg	6.25-312.5	0.9988	0.25	10	74.3	33.2	50	80.8	12.3	250	97.1	2.4	101.7	4.0

2.4. Plasma sample preparation

Mouse plasma sample (5 μL) was mixed with 600 μL of IS working solution that offers better coverage of variable concentrations of individual AAs in the biological samples as well as the isotope-labeled internal standards. The mixture was vortexed for 5 min, and centrifuged at 16,100 g for 10 min at 4 °C. The supernatant was transferred to a new vial, and 5 μL was injected for LC-MS/MS analysis.

2.5. Method validation

As the 20 AAs are all endogenous compounds in mouse plasma, method validation was conducted by using working solutions and stable isotope ISs, similar as described [25, 35]. In particular, selectivity was assessed by comparing the blank solvent, neat solution and pooled mouse plasma sample extract. The lower limit of detection (LOD) of each AA was determined from a signal-to-noise ratio of 3.

The calibration range for each analyte (Table 2) was selected to cover the expected concentrations in the biological samples in which the lower limit of quantification (LLOQ) of the method was established to give a signal-to-noise over 10. Calibration standards were thus prepared with standard solutions containing individual analytes at 6 optimal concentration levels (Supplementary Table 1) and processed identically as plasma sample preparation. Specifically, 5 pL of standard solution were mixed with 600 μL IS working solution to build individual AA calibration curves, where each AA had 6 calibration concentration levels besides the blank sample. Calibration curves were established by a weighted (1/x²) linear regression of the peak area ratio (analyte/IS) on analyte concentration since the intercept was not significantly different from zero while it provided sufficient accuracy. Analyte concentrations were quantitated by interpolation of corresponding response ratios on the calibration curves.

Intra- and inter-day accuracy and precision were established by analyzing QC (low, medium, and high) and LLOQ samples that were processed in the same way as standard and plasma samples. Specifically, the intra-day precision and accuracy were determined by analyzing QC and LLOQ samples in six replicates on one day. The inter-day precision and accuracy were determined with the four concentration levels of samples on three successive days. Precision was evaluated as the percentage relative standard deviation (RSD), and accuracy was defined as the deviation from nominal value (relative error, RE).

The extraction recovery was determined by comparing peak areas of corresponding analytes between mouse plasma sample spiked with standard solution (A_spiked) and the post-preparation plasma sample spiked with standard solution (A_post-spiked) at three QC levels (low, medium, and high). Due to the presence of endogenous AAs in mouse plasma, recovery calculation was corrected with the corresponding peak area in the original pooled plasma (A_matrices). Specifically, recovery was calculated by using the equation, Recovery (%) = (A_spiked –- A_matrices) / (A_post-SPiked-A_matrices) × 100. Likewise, the matrix effect was determined by comparing the slopes obtained in solvent calibration (n = 3) and the slopes of the standard addition in mouse plasma (n = 6) because of the presence of AA analytes in mouse plasma and the lack of blank matrix. In particular, matrix effect was calculated by using the equation, Matrix effect (%) = Mean slope in mouse plasma / Mean slope in solvent × 100.

The stability of the analytes was assessed with QC samples spiked in plasma samples at three levels under a number of conditions, 2 h at room temperature, auto-sampler at 15 °C for 12 h, and three freeze-thaw cycles. RSD and RE values (Supplementary Table 2) were calculated by comparing with freshly-prepared samples that were processed and analyzed under the same experimental conditions.

2.6. Plasma AA dynamics during tumor progression in orthotopic HCC mouse models

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of California, Davis. Fifteen 4-week-old male athymic nude mice (Jackson Laboratory, Bar Harbor, ME) were divided into two groups, 5 mice as healthy controls and 10 mice for the production of orthotopic HCC models to be verified by live animal imaging (see below). Briefly, luciferase/GFP-expressing Huh7 cells suspended in PBS were mixed with Matrigel (v: v = 1: 1) to a final concentration of 1×10⁸ cells/ml. Male athymic nude mice were anesthetized and an incision (~1 cm) along the linea alba in the midline of the abdominal muscle layer was made. Then 20 μL of Huh7 cells in Matrigel suspension (2×10⁶ cells) were injected into the left lobe of mouse liver. One week after inoculation, seven mice were confirmed to outgrow HCC tumors by bioluminescent imaging (Fig. 2A) following intraperitoneal injection of D-luciferin (150 mg/kg) (BioVision, Inc. Milpitas, CA), as we described recently [7], which were enrolled for this study and further imaged once per week to monitor tumor progression.

Fig. 2.

Plasma AA dynamics during tumor progression in the orthotopic HCC Huh7 xenograft mouse models. (A) Bioluminescent signals of HCC xenograft tumors over time following the inoculation. (B) Comparison of plasma AA dynamics between healthy control and HCC mice. Concentration of each AA was normalized to that of the first time point before implantation of HCC cells. Values are mean ± SD (n = 5 for Control group; n = 7 for HCC Progression group). * P < 0.05 and *** P < 0.001 (2-way ANOVA with Bonferroni post-tests).

To investigate plasma AA dynamics during HCC progression, about 20 μL blood was collected from each mouse caudal venous into heparinized centrifuge tubes in the morning (9:00-10:00 am) once a week. Blood sample was also collected from each mouse in both HCC Progression and Control Group before cell implantation as reference for basal levels. Plasma samples were separated immediately by centrifugation at 3,300 g for 5 min and then stored at −80 °C until analysis. Aliquot of 5 μL plasma was used for extraction and analysis by the validated LC-MS/MS method.

2.7. Statistical analysis

Differences in plasma AA levels between HCC Progression Group and Control Group were analyzed by two-way ANOVA with Bonferroni post-tests (Prism, GraphPad Software Inc., San Diego, CA, USA). P < 0.05 was considered to be statistically significant.

3. Results and Discussion

3.1. Method development and optimization

A major challenge for simultaneous determination of the 20 endogenous AAs selected in this study is the diversity of their zwitterionic characters and hydrophilic properties. This makes such AAs to retain weakly on a RP column and thus it is not easy to achieve simultaneous analysis. Furthermore, isobaric and isomeric AAs pose another major challenge to MS/MS analysis and thus a baseline separation on column is desired. Therefore, a set of strategies was utilized; and the chromatographic and mass spectrometric conditions were optimized in this study to address the challenges.

ESI was employed for the ionization of all analytes, and stable-isotope-labeled ISs were utilized to enhance ion-transfer efficiency and detection sensitivity. The precursor and product ions of individual AA analytes and corresponding ¹³C, ¹⁵N-labeled ISs (Table 1) were identified from the full scan MS analysis and MRM mode with stepped collision energy. Declustering potential (DP) and collision energy (CE) were then optimized manually for individual MRM transitions (Table 1) by injecting the standard compounds via infusion pump. Other mass spectrometric parameters, including the ionspray voltage, temperature of the heater gas, flow rates of nebulizer gas, curtain gas and auxiliary gas were optimized separately using both a standard mixture solution and extracted plasma sample.

To establish an optimal LC condition, different stationary phases spanning a traditional RP column and two novel hydrophilic interaction liquid chromatography (HILIC) columns were evaluated. With high polarity, the analytes were weakly retained on an Eclipse Plus C₁₈ column (2.1 × 50 mm, 3.5 μm, Agilent, USA). To address this issue, a Luna NH₂ column (50 × 3 mm, 3 μm, Phenomenex, USA) was further assessed as it showed good retention of some highly polar compounds under various conditions [36]. However, the targeted analytes showed relatively broad peaks and poor separation on this column. Finally, the Intrada Amino Acid column [31, 33] was chosen to separate target AAs, which indeed led to satisfactory retention and peak symmetry (Fig. 1 and Table 1). To improve the sensitivity and chromatographic performances of individual AAs, the compositions of the mobile phases including the types and concentrations of organic solvent and additives were further optimized. Only acetonitrile was utilized as organic solvent component after the elimination of tetrahydrofuran (THF) used in the previous study [31] that could cause swelling in polyetheretherketone (PEEK) materials. The addition of formic acid led to greater signal responses for most AAs and give better peak shape of Asp and Glu. While the increase of ammonium formate concentration improved the elution capacity, ammonium formate at higher concentrations decreased the responses for most compounds. Through an optimal gradient elution with 100 mM ammonium formate in water as Solution A and 0.3% formic acid in acetonitrile/water (v: v = 95:5) as Solution B, satisfactory peak intensity, elution time, resolution and peak shape were obtained for each AA, as shown in Fig. 1. Baseline separation was almost achieved for the isomers Leu (3.26 min) and Ile (3.51 min). In addition, Ile could be analyzed by using an MRM transition (132→69) different from that for Leu (132→43), with a cross-talk less than 5%. The basic AAs such as His, Lys and Arg, and the acidic AAs such as Glu and Asp were eluted with relatively narrow peak widths and good peak shapes (Fig. 1). Overall chromatographic performance was good for the retention and separation of 20 natural AAs with retention time shift and related derivatives less than 0.2 min, as well as a total run time of 13 min, whereas without the requirement for derivatization [27, 37].

Fig. 1.

Typical LC-MS/MS chromatograms of amino acids in mouse plasma samples. Blue: endogenous amino acid analytes in mouse plasma; Purple: isotope-labeled amino acids added to plasma samples as IS.

3.2. Method validation

A variety of parameters including selectivity, sensitivity, linearity of the calibration curve, accuracy and precision, recovery, matrix effects, and stability were critically evaluated. The chromatograms of AAs in blank solvent showed no interferences by either neat solution or pooled matrix, indicating a good selectivity. LOD of each AA analyte was determined as signal-to-noise ratio (Table 2). A relatively higher absolute LOD value identified for Gly was probably due to its small molecular weight and the moderate sensitivity of pseudo-MRM method applied to Gly (Table 2).

The calibration curves were established for individual AAs by using standard solutions at six different concentration levels, among which each was mixed with the same quantity of uniformly ¹³C, ¹⁵N-labeled AA internal standards and processed in the same way as mouse plasma samples. Diverse ranges of concentrations were identified for different AAs to better cover the variable levels of AAs expected in mouse plasma samples. The peak area ratio of each analyte over corresponding IS was plotted against AA concentration and preceded for linear regression. The results showed that every AA analyte exhibited an excellent linear regression coefficient (r² > 0.992) within its calibration range (Table 2).

The extraction recovery values of individual AAs from mouse plasma samples are summarized in Table 2. At the medium and high QC levels, the extraction recovery values ranged from 80.8% to 103.9% for these analytes, and all RSD values were less than 15%, indicating good recovery. Nevertheless, the recovery values of a few AAs such as Lys at the low QC level showed a higher RSD greater than 15% (Table 2). As the recovery values of individual analytes were calculated by subtracting the peak areas of corresponding endogenous AAs in the mouse plasma sample, such higher RSD values are likely attributable to the fact that the low QC concentrations of these AAs added to the matrix accounted for only about 10-20% of corresponding endogenous AAs within the mouse plasma. By contrast, because the concentrations of some endogenous AAs were relatively lower or even too low to be determined (e.g. Cys) in the mouse plasma, such AAs (e.g., Cys) showed much better recovery values at low QC level (Table 2).

The matrix effect was determined by comparing the slopes obtained in solvent calibration and the slopes of the standard addition in mouse plasma samples. The results showed that matrix effects ranging from 82.9% to 118.8 % for individual AA analytes (Table 2). Indeed, the inclusion of isotope-labelled AA internal standards was able to diminish matrix effects, supporting the importance of using appropriate internal standards to achieve accurate quantification of endogenous compounds in biologic matrices.

The performance data for this method in simultaneous analyses of 20 AAs at four different concentration levels, specifically the intra-day and inter-day accuracies as well as precisions, are presented in Table 3. The intra- and inter-day precision values of all analytical AAs were less than 10.1%, and the accuracy values were between −7.6 to 9.4%. These results indicate that this LC-MS/MS method provides an accurate, reliable and reproducible measurement of target AAs.

Table 3.

Precision and accuracy for the analysis of each amino acid (n = 6).

AA	LLOQ					QC1					QC2					QC3
	conc. (μM)	Experimental conc.				conc. (μM)	Experimental conc.				conc. μM	Experimental conc.				conc. μM	Experimental conc.
		Intraday		Interday			Intraday		Interday			Intraday		Interday			Intraday		Interday
		Mean (μM)	RSD (%)	Mean (μM)	RSD (%)		Mean (μM)	RSD (%)	Mean (μM)	RSD (%)		Mean (μM)	RSD (%)	Mean (μM)	RSD (%)		Mean (μM)	RSD	Mean (μM)	RSD (%)
Phe	5	4.66 ± 0.14	3.0	4.74 ± 0.17	3.6	8	7.84 ± 0.11	1.4	7.92 ± 0.17	2.1	40	40.68 ± 0.38	0.9	40.67 ± 0.35	0.9	200	198.8 ± 0.8	0.4	198.9 ± 0.8	0.4
Trp	6.25	6.45 ± 0.16	2.4	6.47 ± 0.14	2.2	10	10.69 ± 0.59	5.5	10.72 ± 0.53	4.9	50	51.45 ± 0.47	0.9	51.53 ± 0.83	1.6	250	252.8 ± 1.6	0.6	253.1 ± 2.2	0.9
Leu	12.5	12.08 ± 1.04	8.6	12.20 ± 0.92	7.5	20	20.00 ± 0.71	3.5	20.28 ± 0.91	4.5	100	102.8 ± 1.7	1.7	102.3 ± 2.0	2.0	500	518.7 ± 8.7	1.7	513.2 ±11.3	2.2
Ile	12.5	12.85 ± 0.38	3.0	12.72 ± 0.40	3.1	20	20.47 ± 0.27	1.3	20.46 ± 0.37	1.8	100	100.8 ± 0.8	0.8	101.0 ± 1.0	1.0	500	498.8 ± 8.1	1.6	500.7 ± 7.3	1.5
Met	5	5.12 ± 0.18	3.5	5.13 ± 0.16	3.1	8	8.17 ±0.25	3.0	8.18 ± 0.29	3.5	40	40.87 ± 1.09	2.7	40.61 ± 1.13	2.8	200	204.7 ± 6.3	3.1	205.1 ± 5.8	2.8
Pro	6.25	6.18 ± 0.23	3.7	6.15 ± 0.19	3.2	10	10.32 ± 0.15	1.4	10.25 ±0.17	1.7	50	51.85 ±0.80	1.5	51.80 ± 0.75	1.4	250	249.0 ± 3.6	1.5	249.5 ± 3.1	1.2
Tyr	6.25	6.46 ± 0.27	4.1	6.41 ± 0.25	3.9	10	10.00 ± 0.19	1.9	10.09 ± 0.29	2.9	50	50.20 ± 1.19	2.4	50.42 ± 1.03	2.1	250	250.8 ± 4.6	1.8	251.9 ± 4.2	1.7
Val	6.25	6.33 ± 0.34	5.3	6.25 ± 0.29	4.7	10	10.38 ± 0.27	2.6	10.30 ± 0.26	2.5	50	51.45 ± 0.52	1.0	51.25 ± 0.54	1.0	250	250.8 ± 2.2	0.9	250.4 ± 2.2	0.9
Cys	25	25.83 ± 1.16	4.5	26.00 ± 1.12	4.3	40	43.77 ± 1.39	3.2	42.83 ± 2.00	4.7	200	206.0 ± 11.0	5.3	205.1 ± 8.8	4.3	1000	974.8 ± 38.2	3.9	978.6 ± 34.7	3.5
Ala	25	25.87 ± 1.59	6.2	25.94 ± 1.58	6.1	40	40.98 ± 2.80	6.8	41.31 ± 2.34	5.7	200	202.7 ± 4.3	2.1	202.3 ± 3.9	1.9	1000	1014± 17	1.7	1012 ±15	1.5
Thr	12.5	12.28 ± 0.92	7.5	12.65 ± 0.94	7.4	20	19.93 ± 0.97	4.9	20.41 ± 1.09	5.4	100	102.5 ±5.1	5.0	103.2 ± 4.9	4.7	500	519.8 ± 18.9	3.6	521.9 ± 24.6	4.7
Glu	25	23.80 ± 1.82	7.6	23.10 ± 1.74	7.6	40	38.20 ± 2.71	7.1	38.06 ± 2.25	5.9	200	199.5 ± 10.3	5.2	199.6 ± 9.9	4.9	1000	975.5 ± 51.2	5.3	995.3 ± 54.4	5.5
Asp	12.5	12.57 ± 1.27	10.1	12.86 ± 1.16	9.0	20	19.85 ± 1.13	5.7	20.06 ± 1.27	6.3	100	106.6 ± 2.1	2.0	104.4 ± 4.5	4.3	500	502.5 ± 26.1	5.2	502.0 ± 24.5	4.9
Gly	50	51.40 ± 3.38	6.6	51.12 ± 3.40	6.6	80	80.55 ± 6.82	8.5	79.89 ± 6.28	7.9	400	416.2 ± 17.2	4.1	415.3 ± 17.5	4.2	2000	2010 ± 166	8.2	2006 ± 140	7.0
Gln	50	51.10 ± 2.16	4.2	50.98 ± 1.86	3.7	80	81.92 ± 2.01	2.5	82.14 ± 1.76	2.1	400	410.0 ± 3.2	0.8	412.8 ± 7.3	1.8	2000	1967 ± 50	2.5	1984 ± 48	2.4
Ser	12.5	12.6 ± 0.43	3.4	12.30 ± 0.58	4.7	20	20.25 ± 1.86	9.2	20.06 ± 1.51	7.5	100	100.9 ± 1.7	1.7	101.5 ± 1.8	1.7	500	506.2 ± 9.8	1.9	508.8 ± 10.8	2.1
Asn	12.5	12.97 ± 0.63	4.8	12.84 ± 0.65	5.1	20	20.27 ± 0.90	4.4	20.31 ± 0.75	3.7	100	99.47 ± 0.98	1.0	99.96 ± 1.78	1.8	500	493.3 ± 13.3	2.7	496.4 ± 12.7	2.6
His	6.25	6.16 ± 0.22	3.5	6.18 ± 0.22	3.6	10	9.91 ± 0.40	4.0	9.90 ± 0.34	3.5	50	50.23 ± 0.93	1.9	50.41 ± 1.17	2.3	250	248.0 ± 2.7	1.1	249.2 ± 3.2	1.3
Lys	25	25.67 ± 0.39	1.5	25.61 ± 0.48	1.9	40	41.27 ± 1.02	2.5	41.24 ± 0.84	2.0	200	205.7 ± 2.3	1.1	205.9 ± 2.2	1.1	1000	1010 ± 21	2.1	1008 ± 17	1.7
Arg	6.25	6.66 ± 0.14	2.2	6.48 ± 0.29	4.5	10	10.62 ± 0.45	4.3	10.47 ± 0.45	4.3	50	52.00 ± 1.06	2.0	50.63 ± 4.40	8.7	250	250.8 ± 2.4	2.1	251.4 ± 4.5	1.8

QC samples spiked with mouse plasma were subjected to further studies on a short-term (room temperature for 2 h), freeze-thaw stability (three times), and autosampler stability (processed samples at 15 °C for 12 h in autosampler). All stability experiments were carried out at low, medium and high concentrations levels in three replicates. The data are summarized in Supplementary Table 2, demonstrating that all AAs were stable under the experimental conditions.

3.3. Application to the investigation of plasma AA dynamics during orthotopic HCC xenograft tumor progression in mouse models

The validated method was thus applied to a comprehensive study on the dynamics of mouse plasma AAs while orthotropic HCC xenograft tumors were progressing (HCC Progression group), which were verified by live animal imaging (Fig 2A) as we reported recently [7]. A cohort of healthy mice (Control group) was included for comparison. While the concentrations of unconjugated Cys in mouse plasma were surprisingly lower than LLOQ, all other 19 AAs were readily determined (Fig. 1). The absolute AA concentrations in individual samples are provided in Supplementary Table 3, and the normalized plasma AA concentration versus time courses are shown in Fig. 2B to better illustrate the changes in AA dynamics. Interestingly, our data demonstrated that, with the increase of HCC severity over time, plasma Arg levels were decreased remarkably, whereas Ala and Thr levels were increased significantly (Fig. 2A and 2B), especially when compared to the control healthy mice.

Our finding on the decrease of Arg in HCC mouse blood is consistent with previous studies, including those with HCC patients [11, 38], chemical-induced HCC rat models [38], and other types of cancer patients [39]. Arginine, a conditionally essential AA, is an important precursor for the production of biologic proteins, polyamines, creatinine and nitric oxide, which can be acquired from diet and synthesized de novo in mammals involving mainly some cationic AA transporters such as SLC7A1/7A2 and a rate limiting enzyme, argininosuccinate synthetase (ASS1), respectively [40, 41]. Given the importance of Arg in cancer metabolism, Arg deprivation therapy may be used for the control of tumor progression which is actually dependent on tumoral ASS1 levels [42]. Interestingly, ASS1 was revealed to be considerably downregulated in HCC tumors [43, 44]. Therefore, HCC tumor progression likely relies on extracellular Arg supply, providing a reasonable explanation for the lower plasma Arg concentrations in HCC xenograft mice found in current study (Fig. 2). Furthermore, the high-affinity Arg transporter SLC7A1/CAT-1 was shown to be upregulated in HCC specimens [45], offering additional support for the decrease of plasma Arg levels with HCC progression. In addition, Arg levels in mouse liver tissues were found to be elevated sharply when SLC7A1/CAT-1 was downregulated via the inhibition of liver-specific miR-122 [46]. Therefore, a lower blood Arg level in orthotopic HCC xenograft mice identified in this study not only suggests its utility as biomarker for HCC progression but also supports arginine deprivation therapy against HCC [43, 47], which both warrant more extensive preclinical and well-controlled clinical studies.

The significantly higher levels of Ala and Thr found in well-monitored orthotopic HCC xenograft mice in this study is rather surprisingly since previous studies reported a decrease of Ala and Thr as well as essentially all other AAs in HCC patients [11, 38], lack of change in Ala and decrease of Thr in chemical-induced HCC rats [38], and an either unchanged, higher or lower Ala and Thr level in other types of cancer patients [39, 48-50]. Differences in results between these studies may be due to variable stages of cancer patients examined or degrees of tumor severity, as well as possible species differences in AA sources and catabolism. Furthermore, previous animal study involved chemical-induced rats [38], while our study employed xenograft tumor mouse models derived from human HCC cells that were closely monitored over time through live animal imaging. In addition, food and drink as well as housing conditions were more unified for experimental animals in current study, which is uncommon for clinical studies. Indeed, the nonessential AA Ala is not only synthesized from pyruvate and other AAs in animal body [51] but also acquired from the diet. On the other hand, the essential AA Thr must be obtained from the diet by animals, and dietary Thr can be easily converted to other AAs [52]. Therefore, caution is advised when interpreting and comparing results from experimental animals and humans.

4. Conclusion

A rapid, selective and accurate LC-MS/MS method was developed for simultaneous, targeted quantification of natural AAs, which involved satisfactory separation with an Intrada Amino Acid column. The results demonstrated that this method is simple and sensitive in monitoring AAs in mouse plasma samples. This validated method was successfully applied to the study on plasma AA dynamics, and our results revealed a significantly lower Arg concentration and higher Ala and Thr levels in HCC mouse models when compared with healthy cohorts. These findings may offer insights into molecular mechanisms behind HCC progression that warrants future investigations.

Highlights.

• A simple and fast LC-MS/MS method was developed for simultaneous quantification of 20 natural amino acids in mouse plasma samples.

• Chromatographic separation was achieved on an Intrada Amino Acid column.

• Method validation demonstrated that this LC-MS/MS method is a specific, accurate, and reliable assay with minimal matrix effects and excellent extraction yields.

• Validated LC-MS/MS method was applied to a study on plasma amino acid dynamics in orthotopic hepatocellular carcinoma (HCC) mice in vivo, which revealed a significant decrease in L-arginine concentrations as well as sharp increase of L-alanine and L-threonine levels during HCC progression.

Abbreviations

AA

amino acids

HCC

hepatocellular carcinoma

IS

internal standard

TCA

tricarboxylic acid

Phe

L-phenylalanine

Tyr

L-tyrosine

Thr

L-threonine

Lys

L-lysine

Ala

L-alanine

Pro

L-proline

Ile

L-isoleucine

Glu

L-glutamic acid

Arg

L-arginine

His

L-histine

Cys

L-cystein

Trp

L-tryptophan

Gln

L-glutamine

Asn

L-asparagine

Asp

L-aspartic acid

Ala

L-valine

Gly

glycine

Ser

L-serine

Leu

L-leucine

Met

L-methionine