Highly Sensitive Quantification Method for Amine Submetabolome

Subscriber access provided by - Access paid by the | UCSB Libraries

Article

Highly Sensitive Quantification Method for Amine Submetabolome Based on AQC-labeled-LC-Tandem-MS and Multiple Statistical Data Mining: A Potential Cancer Screening Approach Qian Zhang, Huarong Xu, Ran Liu, Peng Gao, Xiao Yang, Pei Li, Xiaotong Wang, Yiwen Zhang, Kaishun Bi, and Qing Li Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b02372 • Publication Date (Web): 12 Sep 2018 Downloaded from http://pubs.acs.org on September 13, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Highly Sensitive Quantification Method for Amine Submetabolome Based on AQC-labeled-LC-Tandem-MS and Multiple Statistical Data Mining: A Potential Cancer Screening Approach Qian Zhang †,# , Huarong Xu

†,#

, Ran Liu†, Peng Gao ‡, Xiao Yang †, Pei Li †,

Xiaotong Wang †, Yiwen Zhang †, Kaishun Bi †, Qing Li*,†

†

School of Pharmacy, Shenyang Pharmaceutical University, 103 Wenhua Road,

Shenyang 110016, China ‡

Metabolomics Core Facility of RHLCCC, Northwestern University Feinberg School

of Medicine, Chicago, IL 60611, United States #

Qian Zhang and Huarong Xu contributed equally to the work.

* Corresponding author. Tel.: +86 24 43520589; fax: +86 24 43520589. E-mail address: [email protected]

ABSTRCT The relationship between amine submetabolome and cancer has been increasingly investigated. However, no study was performed to evaluate the current methods of amine submetabolomics comprehensively, or to use such quantification results to provide an applicable approach for cancer screening. In this study, a highly sensitive and practical workflow for quantifying amine submetabolome which was based on 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC)-labeled-HPLC-MS/MS analysis combined with multiple statistical data processing approach was established and optimized. Comparison and optimization of two analytical approaches, HILIC separation and pre-column derivatization, and three types of surrogate matrices of plasma were performed systematically. The detection sensitivities of AQC-labeled amines were increased by 50−1000 fold compared with the underivatization-HILIC method. Surrogate matrix was also used to verify the method after a large dilution factor was employed. In data analysis, the specific amino-index for each cancer sample was identified and validated by univariate receiver operating characteristic (ROC) curve analysis, partial least squares discrimination analysis (PLS-DA) and ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

multivariate ROC curve analysis. These amino-indexes were innovatively quantified by multiplying the raised markers and dividing the reduced markers. As a result, the numerical intervals of amino-indexes for healthy volunteers and cancer patients were provided, and their clinical value were also improved. Finally, the integrated workflow successfully differentiated the value of amino-index for plasma of lung, breast, colorectal and gastric cancer samples from controls, and among different types of cancer. Furthermore, it was also used to evaluate therapeutic effects. Taking together, the developed methodology, which was characterized by high sensitivity, high throughput and high practicality, is suitable for amine submetabolomics in studying cancer biomarkers and could also be applied in many other clinical and epidemiological researches. INTRODUCTION As an essential component of systems biology, metabolomics has been widely used for biomarker discovery by globally identifying and quantifying small molecule metabolites.1,2 However, profiling all metabolites in a certain biological system - the metabolome - is hampered by the great variety of physical-chemical properties and wide range of concentrations.3 One strategy to address this major challenge is to classify the metabolites into several subgroups based on their common functional moieties, then to perform in-depth analysis of the specific chemical group submetabolomes. In this approach, the whole metabolome was divided into four chemical group based submetabolomes approximately: amine, carboxyl, hydroxyl and carbonyl submetabolome.4-6 And with the customized sample preparation approach and optimized separation and detection technology, targeted submetabolome could be quantified accurately, resulting in a better defined and more robust data set for subsequent biological interpretation.7 Amine submetabolome includes amino acids and biogenic amines. It represents a significant class of biological compounds because they are modulated in a series of biological processes, such as cell growth, proliferation and differentiation.8-11 Reportedly, the dynamic changes of amine profile are closely connected with the onset and progression of cancer in which amines have already been considered as the ACS Paragon Plus Environment

Page 2 of 20

Page 3 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


potential biomarkers for early diagnosis and prognosis.12-17 Hence, a more sensitive and practical amine submetabolome technique is desperately needed to discover more meaningful biomarkers and serve as an adjuvant diagnosis tool for cancer screening. There are a variety of methods for the analysis of amine compounds, from which liquid chromatography tandem mass spectrometry (MS) technique stands out by its high sensitivity, high selectivity and high throughput.18-23 However, hydrophilic amines are unable to be retained on common used reversed-phase stationary phases. Alternative strategies have been carried out, such as the usage of hydrophilic interaction liquid chromatography (HILIC), the usage of derivatization reaction to increase the hydrophobicity of analytes and further analysis by reversed phase liquid chromatography (RPLC), or even the addition of ion-pairing agents into the mobile phase. Since the usage of ion-pairing agents may lead to ion-suppression problem and system contamination, the first two methods, HILIC and pre-column derivatization, which are also dominant applications, were optimized and compared systematically in this study to provide a more proper approach for amine submetabolome and make a reference for the future analysis. Due to the endogenous nature of amines, it is hard to obtain the absolute blank or analyte-free biological matrix. Therefore, alternative strategies for reference samples, such as authentic analytes in surrogate matrix and surrogate analytes in authentic matrix have to be employed.24,25 However, each method has its disadvantages and may lead to compromising quantitation results. For example, surrogate and authentic analytes are not always identical in extraction recovery, chromatographic retention, and signal intensity. Also, preparing stable isotope labeled surrogate standards for every analyte is uneconomical and unpractical. In the surrogate matrix, the extraction recovery and matrix effects of analytes may be different. Therefore, it is important to develop an appropriate approach to minimize the error and improve the practicality according to the nature of the analytes and method. Many of the recent literatures on metabolomics discovered potential biomakers without going further for comprehensive evaluation and validation,18,19,22 which slowed down the clinical application of biomarkers to some extent. This illustrates a ACS Paragon Plus Environment


Page 4 of 20

critical need to develop a systematic data processing workflow to discover and validate the biomarkers and to provide a guidance for further research. So we proposed an integrated strategy for identifying, validating and quantifying biomarkers and providing reference intervals for healthy people and cancer patients to take a step forward in clinical application. In our study, a sensitive and practical workflow of quantitative amine submetabolome based on AQC-labeled-HPLC-MS/MS method and multiple statistical data mining stretegies was presented. Meanwhile, the AQC-labeled regulation of amine containing metabolites was summarized to make a reference for further studies. Three types of surrogate matrices were compared to overcome the difficulty of no absolute blank matrix and also guarantee the reliability of method validation. Then the well validated method was applied to discover potential amino-indexes in plasma for distinguishing lung, breast, colorectal and gastric cancer from healthy controls and distinguishing various types of cancer from each other. Finally, an amino-index was verified by the plasma of lung cancer patients before and after treatment. Results indicated that the developed method was suitable for cancer biomarker study and could be further expected to apply for cancer early diagnosis. EXPERIMENTAL SECTION Chemicals and Reagents. Amine

standards

(N-acetylspermine

for

hydrochloride,

epinephrine

and

norepinephrine for bitartrate) (Table S1), bovine serum albumin (BSA) and activated charcoal were from Sigma (St. Louis, MO, USA) and AccQTag Ultra reagent was purchased from Waters Corporation (Milford, MA, U.S.A.). Methanol and acetonitrile of HPLC grade were from Fisher Scientific (Fair Lawn, NJ, USA). Distilled water was from Wahaha Co. Ltd (Hangzhou, China). HPLC-grade formic acid and ammonium formate were from Shandong Yuwang Industrial Co. Ltd. (Yucheng, China). Samples. The human plasma for study dataset was collected from 78 cancer patients, including 28 with lung cancer, 18 with breast cancer, 21 with colorectal cancer, 11 with gastric ACS Paragon Plus Environment

Page 5 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


cancer and 38 healthy volunteers who were recruited in the General Hospital of Shenyang Military Region. The demographic and clinical characteristics of participants were summaried as Table S2. Plasma for test dataset was also collected from 8 lung cancer patients before and after treatment by Traditional Chinese Medicine (TCM), Kangai injection (patient 1, 3, 4, 8) or Shenqi Fuzheng injection (patient 2, 5, 6, 7). Kangai injection consists of five active ingredients extracted from Ginseng (ginsenosides Rg1, Re, Rf, Rb1) and Astragalus (astragaloside) while Shenqi Fuzheng

injection

comprised

of

Codonopsis

pilosula

and

Astragalus

membranaceus.26-29 And these 8 patients were all diagnosed as recovered according to the reduction rate (%) of “Lung Cancer Symptom Scale”30-32 (Table S3) after 4 therapeutic cycles. Cancer was diagnosed through routine clinical and laboratory results and verified by biopsy. And all plasma samples were collected after overnight fast. The study was carried out in conformity with the Declaration of Helsinki and approved by ethics committees of General Hospital of Shenyang Military Region (Ethics ID number K (2016) 38). All subjects were acquainted with the goal of the research and signed written informed consent. Preparation of Standard Solutions. Stock solutions of forty-nine analytes and two IS were separately prepared in 0.1M HCl or methanol: water (20:80, v/v). And then the series of standard solutions were further obtained by diluting the stock solutions with methanol. All the solutions were stored in a 4 °C refrigerator and returned to room temperature before use. Due to the presence of endogenous levels of amines in human plasma, the calibration standard samples and quality control (QC) samples were prepared in a surrogate matrix, 1% BSA in PBS. Sample Preparation. 10 µl IS mixture (100 µg/ml for alanine-d4 and 10 µg/ml for 1,6-Diaminohexane), 10 µl methanol and 20µl plasma were transferred to an Eppendorf tube successively, then 80 µl water was added to dilute the plasma. And 400 µl of methanol (containing 0.2% formic acid (v/v)) was added to precipitate the proteins. After vortex for 30s and centrifugation (4 °C, 15,000 rpm) for 5 min, 10 µl of the supernatant was collected to ACS Paragon Plus Environment


be derivatized with the AccQTag reagents in the light of the manufacturee’s protocol procedure. Finally, 2µl of the derivative solutions was transferred for HPLC-MS/MS analysis. HPLC-MS/MS Analysis. HPLC-MS/MS Analysis was performed on an XR LC-20AD ProminenceTM HPLC system (Shimadzu, Japan) combined with a QTRAPTM 4000 MS/MS system (AB Sciex, USA). The chromatography separation was conducted through a Kinetex XB-C18 column (4.6×100 mm, 2.6 µm; Phenomenex, USA) and gradient elution with the mobile phase (A) 0.1% formic acid in water, (B) 0.1% formic acid in acetonitrile at 35 °C. The gradient eluting procedure was 5% B → 35% B at 0.01–12.00 min; 5% B at 12.01–15.00 min with the flow rate of 0.4 ml/min. An electrospray positive ionization (ESI +) and multiple reaction monitoring mode (MRM) were employed in the MS/MS detection of analytes. Nitrogen was used as gas 1, gas 2, and curtain gas at 50, 50, and 20 psi. The general parameters of ion source were as follows: source temperature, 500 °C; ion spray voltage, 5500V. Compound-specific parameters were listed in Table S1. Method Validation. The method was validated according to US FDA and EMA guidance for bioanalytical method validation33,34 concerning specificity, calibration curve, lower limit of quantification (LLOQ), accuracy and precision, recovery, matrix effect and stability. And considering the particularity for the analysis of endogenous compounds24, 1% BSA in PBS was used as surrogate matrix in the absence of analyte-free authentic matrix. Data Processing. The amines concentrations of study dataset were calculated according to the standard curves and represented as mean ± SD. Multiple statistical data mining approaches were applied to screen metabolites with the best discernment and form amino acid multivariate index namely amino-index. As displayed in Figure 1, the visual data analysis process was summarized. Firstly, the differences in amines profiles between cancer patients and healthy volunteers were assessed by Student’s t-test (normal ACS Paragon Plus Environment

Page 6 of 20

Page 7 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


distribution) and the nonparametric Mann-Whitney U test (abnormal distribution) using SPSS 19.0 software (SPSS Inc., Chicago, IL). Then for further screening of the markers of cancer and generating amino-index, PLS-DA and ROC curve analysis were implemented with SIMCA 13.0 (Umetrics AB, Umea, Sweden) and SPSS 19.0, respectively. Finally, to verify the combined prediction ability of indexes discovered by PLS-DA and ROC analysis, prediction probability P value was calculated by logistic regression analysis and then the ROC curve was plotted using the P value. After the amino-index was produced, we quantified it by multiplying the raised markers and dividing the reduced markers to improve its clinical practicality. In order to assess the clinical utility of the amino-index, amine profiles of 8 lung cancer patients were also compared before and after treatment by PCA analysis using SIMCA 13.0. RESULTS AND DISCUSSION Method Development. The optimization and comparison of two classic methods for amine submetabolome. In this work, two classic methods, HILIC and pre-column derivatization, were optimized and compared. HILIC-MS/MS conditions were listed in the Supporting Information. And in HILIC mode, the separation was poor without buffer in mobile phase (Figure 2). Better separation and narrower peaks for arginine, histidine, lysine, aspartic acid and asparagine were obtained in the presence of buffer. However, all amines showed sharply decreasing peak areas possibly due to the ion suppression effect of the buffer (see Figure 2) 35. While in the pre-column derivatization approach, different chemical reagents were often used, such as OPA (o-phthaldialdehyde), DNS-Cl (diethyl dansyl chloride), DEEMM (ethoxymethylenemalonate), FMOC (9-fluorenylmethyl chloroformate). But these derivatization reagents had more or less disadvantages, including tedious procedure, long reaction time, side reactions and poor stability. An alternative of AccQTag Ultra reagent was chosen due to its ability of labelling primary and secondary amines, mild conditions and good stability of derivatives.7,36,37 The schematic diagram of AQC derivatization of amines and ACS Paragon Plus Environment


Page 8 of 20

fragmentation pathways of derivatives was shown in Figure S17. Owing to better retention and higher sensitivity shown in Figure 2, AQC derivatization was chosen. The AQC-labeled rules of amine submetabolome. Through the optimization of MS conditions, derivatization rules of amine containing metabolites were summarized to make a reference for untargeted and targeted amine submetabolome, shown in Figure 3. The fragment ion of m/z 171 could be produced by every derivative, which may possibly due to a loss of AMQ group (Figure S1). For monoamine amino acid, both primary and secondary amine could react with AQC and produce the parent ion of [M+Aqc+H] +. And diamine containing amino acids were doubly derivatized, demonstrated by lysine and ornithine, which displayed the addition of two AQC units with a double charged ion [M+2Aqc+2H]

2+

. It was also

observed that there was no reactivity for amide. Furthermore, polyamines (e.g., spermine, spermidine) had multiple reactive sites and displayed a ion of [M + nAqc + nH] n+, where n equals total number of primary and secondary amines. For acetylated polyamines, all primary and secondary amines could be derivatized except for acetylated

amines,

such

as

N-acetylspermine,

N,N-diacetylspermine,

N-acetylspermidine. As shown in Figure 3, N,N-diacetylspermidine presented no derivatives of any form surprisingly, which might be caused by the two adjacent amides. Besides, catecholamine and tryptamine were also derivatized as shown in Figure 3. The optimization of sample pretreatment. Considering the strong polarity of amine submetabolome, protein precipitation was preferred for the pretreatment of plasma samples before derivatization. And then acetonitrile, methanol and isopropanol with various concentrations of formic acids were tested. As a result, methanol showed better sensitivity and reduced matrix effects. In addition, increased recoveries of amino-containing compounds were observed with increasing formic acid (v/v) concentration from 0.05% to 0.2% in precipitation solvent while recovery was decreased with the concentration of formic acid higher than 1%. It was speculated that the AQC-labeled process was affected in low pH condition. Thus, methanol (with 0.2% formic acid) was finally chosen with acceptable ACS Paragon Plus Environment

Page 9 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


recoveries and reproducible for all analytes. Method Validation. The optimal HPLC-MS/MS method was then validated for selected analytes. As displayed in Table S1, the validated analytes were all amino acids, which was due to their 100-1000 times higher concentrations comparing to biogenic amines. A large dilution factor was adopted to reduce matrix effects, which led the concentrations of biogenic amines were below the limits of quantification. Also, our previous study had determined and compared biogenic amine metabolite profiles between cancer patients and healthy controls.38 Therefore, only amino acids were quantified and validated finally. The choice of surrogate matrices. Without true blank samples, three types of surrogate matrices including 1% BSA in PBS, plasma stripped of analytes by active charcoal, and plasma processed through dialysis were compared. All three surrogate matrices had similar matrix effect and recovery for each analyte as original matrix observed possibly due to the large dilution factor in sample preparation. However, a certain number of analytes were still detected from plasma disposed by active charcoal or dialysis, while the concentrations of amino acids in 1% BSA was very low to be considered as insignificant (Figure 4). Therefore, 1% BSA in PBS was finally chosen as the surrogate matrix. The analytical performance of the developed method. Specificity was tested by observing whether the retention time of each analyte in standard solution and blank plasma was identical in the absence of authentic blank matrix. Chromatography indicated there was no interference on retention time of target analytes or IS. As listed in Table S5, the calibration regression coefficients for analytes were all more than 0.9901. QC samples of three levels and LLOQs were used to evaluate intra- and inter-day precision and accuracy. The relative standard deviation (RSD) were all less than 19.8 % and the relative error (RE) were all less than 19.3 % (Table S6). The recoveries of analytes from plasma were all higher than 60.3 % at different levels with RSD less than 14.9 %. The IS-normalized matrix effects were between 80.1–114.0 %, with RSD less than 14.8 %. (Table S7) The plasma samples ACS Paragon Plus Environment


were stable at −80 ℃ for a month and through three freeze – thaw cycles. The prepared samples were stable for 24 h at 4 ℃ in autosampler. (Table S8) And the derivatives of analytes were also stable at room temperature for 1 week. In summary, the developed HPLC-MS/MS method was well validated to quantify amino-containing compounds in plasma. Amine Submetabolome. The differentiation between cancer patients and healthy volunteers. The amine profiles for study dataset were summarized in Table S9. Student’s t-test and Mann-Whitney U test were used to compare the amine metabolome in patients with certain type of cancer vs healthy volunteers. From the results, most of amino acids were decreased in gastric and colorectal cancer but increased in lung and breast cancer compared with healthy controls. More specifically, 9 decreased and 1 increased in gastric cancer; 13 decreased and 2 increased in colorectal cancer; 4 decreased and 11 increased in breast cancer; 2 decreased and 12 increased in lung cancer. It was speculated that digestive system cancer might change the gastrointestinal absorption capacity39 or protein-rich body fluids oozed from the ulcerative cancer masses.40 PLS-DA and ROC curve analysis were used to discover more meaningful cancer biomarkers. Healthy controls and each particular type of cancer were separated clearly based on the plasma amine profiling (R2Y ≥ 0.868, Q2 ≥ 0.832) in the PLS-DA score plots (Figure 5). The discrimination ability of each amino acid were evaluated by PLS-DA coefficient and area under the ROC curve (Taking lung cancer for example in Figure 6). And the multivariate amino-index for cancer prediction was generated by setting the coefficient above 0.12 and AUC value above 0.8 or below 0.2. Sarcosine, glutamic acid, citrulline, tryptophan, glutamine and histidine were for lung cancer; proline, alanine, sarcosine, glutamic acid, γ-aminobutyric acid, and histidine for breast cancer; isoleucine, proline, citrulline, glycine and glutamine for colorectal cancer; asparagine, citrulline, tryptophan, histidine, glutamine and alanine for gastric cancer, respectively. To ensure the utility of the amino-indexes, multivariate ROC curve analyses were ACS Paragon Plus Environment

Page 10 of 20

Page 11 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


carried out using P values as test variables obtained from logistic regression analysis. The AUC of amino-indexes were all 1.000 (P

Highly Sensitive Quantification Method for Amine Submetabolome

Recommend Documents