Dansylhydrazine Isotope Labeling LC-MS for Comprehensive

Subscriber access provided by University of Sunderland

Article

Dansylhydrazine Isotope Labeling LC-MS for Comprehensive Carboxylic Acid Submetabolome Profiling Shuang Zhao, and Liang Li Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b03435 • Publication Date (Web): 19 Oct 2018 Downloaded from http://pubs.acs.org on October 20, 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 28 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

(Manuscript submitted to Anal Chem on July 31, 2018; Revised on September 30, 2018)

Dansylhydrazine Isotope Labeling LC-MS for Comprehensive Carboxylic Acid Submetabolome Profiling

Shuang Zhao and Liang Li* Department of Chemistry, University of Alberta, Edmonton, Alberta T6G 2G2, Canada

*Corresponding authors. E-mail: [email protected]

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

Page 2 of 28

2

Abstract High-performance chemical isotope labeling (CIL) LC-MS is an important tool for profiling chemical-group-based submetabolomes using different labeling chemistries for quantitative metabolomics. Metabolites containing carboxylic acid groups are a class of molecules playing diverse and significant roles in many metabolic pathways. We report a relatively simple and convenient method for analyzing the carboxyl submetabolome with high coverage. Dansylhydrazine (DnsHz) labeling of the carboxylic acid group in metabolites is used to improve both separation and ionization in LC-MS. Using differential

13

C- and

12

C-DnsHz reagents for

labeling individual samples and a reference sample (e.g., a pooled sample), accurate relative quantification of individual metabolites among comparative samples can be achieved. DnsHz labeling of carboxylic acids could be carried out at room temperature in water-containing solution in two hours. A labeled-carboxyl-standard library consisting of 193 endogenous human metabolites was constructed for metabolite identification, which covers more than 55 pathways. To demonstrate the performance of this method, analysis of triplicate human urine samples with duplicate injections (n=6) was carried out. A total of 2266 peak pairs or metabolites were commonly detected in all six runs and, among them, 81 peak pairs were positively identified and 517 and 1445 peak pairs were matched in the zero- and one-reaction MyCompoundID metabolome libraries, respectively, based on accurate mass search. The possibility of using this method for comprehensive profiling of the carboxyl submetabolome in real world samples was demonstrated in metabolome comparison of 30 urine samples of female and male individuals and absolute quantification of two urinary acids.


Page 3 of 28 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


3

Introduction Liquid chromatography-mass spectrometry (LC-MS) is one of the most commonly used analytical platforms for metabolome analysis. In a conventional LC-MS approach, a combination of two chromatographic conditions [i.e., reversed phase liquid chromatography (RPLC) and hydrophilic interaction liquid chromatography (HILIC)] and two ionization modes (i.e., positive and negative ion detection) is often used to increase the number of detected metabolites, thereby expanding the overall metabolome coverage. To increase quantification accuracy, choosing appropriate internal standards for analytes can correct for sample loss, instrument drift and matrix effect in LC-MS. However, a great variety of metabolites with wide concentration ranges can present a major challenge for detecting and quantifying many analytes simultaneously in complex biological samples. A different approach to address the challenge is to perform multichannel chemical isotope labeling LC-MS (mCIL LC-MS) analysis. In this approach, metabolites containing a certain chemical functional group (i.e., submetabolome) are derivatized using a pair of chemical isotope labeling reagents and then analyzed by LC-MS. During the derivatization, isotopic moiety can be introduced into the labeled metabolites, thereby creating stable isotopic internal standards for all the labeled metabolites for accurate quantification. In addition, the chemical and physical properties of metabolites are altered to improve their separation and detection. If multiple CIL LCMS methods are used for analyzing different chemical-group-based submetabolomes, the combined results would produce a near-complete metabolome profile. There are a number of labeling reagents reported for targeted or untargeted metabolite analysis with varying degrees of performance.1-12 In our previous studies, several highperformance labeling methods have been developed, including 12C-/13C-dansyl chloride (DnsCl)



Page 4 of 28

4 labeling for amine/phenol submetabolome analysis,13 base-activated 12C-/13C-DnsCl labeling for hydroxyl

submetabolome

analysis,14

submetabolome analysis15 and

12

C-/13C-dansylhydrazine

(DnsHz)

for

carbonyl

12

C-/13C-dimethylaminophenacyl (DmPA) bromide labeling for

carboxyl submetabolome analysis.16 Carboxyl-containing metabolites have many important functions in biological systems. Several chemical derivatization methods have been reported for analyzing carboxylic acid using different MS platforms.17-23 Most of the reactions for carboxylic acid derivatization are based on condensation reaction with amines24-31 or esterification reaction.16, 32, 33 Our previously reported DmPA method has been used for profiling carboxyl submetabolome in metabolomics applications. However, based on our experience of using DmPA for metabolome profiling of various types of sample, we found that this method has two drawbacks. DmPA labeling was found to be optimally done in non-aqueous solution, requiring an extraction step from an aqueous biological sample. DmPA labeling also produces a relatively higher background, compared to dansyl-based labeling methods for profiling amines, hydroxyls and carbonyls, potentially causing ion suppression of labeled analytes. We have recently reported a differential

12

C-/13C-DnsHz labeling method for analyzing

carbonyl-containing metabolites and demonstrated very high performance of DnsHz labeling for profiling the carbonyl submetabolome.15 Encouraged by this development, we examined the possibility of using DnsHz for labeling carboxylic acids with a hope that the benefits of DnsHz labeling could be similarly expanded to carboxylic acid analysis. In principle, the hydrazide moiety in DnsHz can be linked with carboxylic acids using a coupling reagent. In this report, we present our study of developing DnsHz labeling for LC-MS profiling of the carboxyl submetabolome. The reaction conditions were optimized with labeling of several carboxylic acid standards covering


Page 5 of 28 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


5 different acid classes and structures. A robust workflow including a library of 193 DnsHz-labeled acid standards for positive metabolite identification was developed. The performance of this method for real biological samples was validated using human urine as an example of complex metabolomic systems.

Experimental Section Principle of DnsHz Labeling LC-MS. A general workflow of applying this method for metabolomics study is shown in Figure 1. Supplemental Note S1 provides detailed description of the workflow. Briefly, accurate and precise relative quantification is realized by differential chemical isotope labeling. Individual samples are labeled by 12C-DnsHz (light labeling), while a control sample or a pooled sample is labeled by 13C-DnsHz (heavy labeling). After mixing the 12Clabeled sample and 13C-labeled control, the mixtures are analyzed using LC-MS. IsoMS software is used to extract peak pairs and calculate the peak area ratio between the

12

C-labeled and

13

C-

labeled metabolites. Metabolite identification is performed using a DnsHz-labeled carboxyl standard library and MyCompoundID database (see below). Chemicals and Urine Sample. The 193 metabolite standards (Supplemental Table S1) were obtained from the Human Metabolome Database (HMDB). 12C-DnsHz and 13C-DnsHz are available from mcid.chem.ualberta.ca. Other chemicals were purchased from Sigma-Aldrich (Edmonton, AB, Canada), except those specifically stated. Human urine sample collection and processing is described in Supplemental Note S1. Preparation of Solution. The 193 standards were individually dissolved in water to a concentration of 10 mM. If solubility was too low, the supernatant of a saturated solution was used. The standard solutions were stored in -20 ºC. Labeling reagents, 12C- or 13C-DnsHz, were dissolved



Page 6 of 28

6 in acetonitrile to 10 mg/mL. 2-(N-morpholino)ethanesulfonic acid (MES) and Na2HPO3/ NaH2PO3 were dissolved in water to generate an MES buffer or phosphate buffer solutions (PBS), respectively. The buffer solutions were used to prepare fresh coupling reagents solutions, including 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and other additives (1-hydroxy-7azabenzotriazole, 4-dimethylaminopyridine, or N-hydroxysuccinimide; see Results and Discussion). Copper chloride (CuCl2) was dissolved in water as a quenching reagent. Chemical Isotope Labeling. For metabolite standard labeling, 20 µL of standard solution in water was placed in a centrifuge vial, followed by adding 20 µL of EDC and 20 µL of additives in buffer solution. After mixing, 20 µL of 10 mg/mL 12C-DnsHz was added. After vortexing and spinning down, the mixture was shaken at 20 ºC for 90 min. Then 20 µL of 100 mM CuCl2 solution was added to quench the labeling reagents and incubated at 40 ºC for 30 min. The solution was centrifuged at 20817 g for 10 min and diluted before injecting into LC-MS for analysis. To prepare the 12C-/13C-DnsHz labeled urine sample for method validation, 40 µL of urine was divided to two aliquots: 20 µL of urine was labeled by other 20 µL was labeled by

13

12

C-DnsHz (light labeling) and the

C-DnsHz (heavy labeling) using the same reaction conditions

following procedures described above. After labeling, the

12

C-DnsHz labeled urine was mixed

with the same volume of the 13C-DnsHz labeled urine. The mixture was injected into LC-MS for analysis. The

12

C-/13C- DnsHz labeled blank samples were prepared by labeling LC-MS grade

water with the same protocol. LC-MS and Data Analysis. LC-MS and MS/MS experiments were done using Ultimate 3000 UHPLC (Thermo Scientific, USA) linked to Impact HD Quadrupole Time-of-flight (Q-TOF) MS (Bruker, Billerica, MA). CIL LC-MS data were processed and analyzed using a set of programs (freely available from www.mycompoundid.org) including IsoMS,34 Zerofill,35 and


Page 7 of 28 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


7 IsoMS-Quant36. Metabolite identification was done using MyCompoundID (MCID) at www.mycompoundid.org with zero-reaction and one-reaction MCID libraries.37 Supplemental Note S1 provides more details on LC-MS conditions and data analysis.

Results and Discussion Optimization of Labeling Reaction. Carbodiimide-mediated coupling reaction is one of the most popular chemistries for crosslinking carboxyl group to amine group. In this reaction, the carboxyl group present in a compound is activated by a condensation reagent (i.e., carbodiimide), followed by reacting with an amine group in a labeling reagent, resulting in a stable amide bond. 1-Ethyl-3-(-3-dimethylaminopropyl)carbodiimide (EDC) is chosen in our work as carbodiimide, because it is water-soluble and allows labeling reaction to be carried out in an aqueous solution, which is good for metabolomics since water is often present in biological samples. Direct labeling of a metabolome sample without removing water is more convenient. For the reaction mechanism,38 the carboxyl group is first activated by EDC to form an O-acylisourea intermediate that contains a better leaving group and can be displaced by nucleophilic attack from an amine group. Although the reaction can be carried out rapidly in mild condition, the hydrolysis and rearrangement of the intermediate limit the reaction efficiency and introduce side reactions. Thus, other additives such as N-hydroxysuccinimide (NHS) are employed to assist the reaction and reduce side reactions.39 To optimize the labeling reaction for various carboxyl-containing metabolites in a metabolome sample, seven carboxylic acid standards from different categories (Supplemental Figure S1) were mixed to generate a standard mixture, including propionic acid representing shortchain fatty acids, pyruvic acid representing keto-acids, benzoic acid representing aromatic acids,



Page 8 of 28

8 succinic acid representing polycarboxylic acids, cholic acid representing bile acids, glucuronic acid representing sugar acids and hydroxyoctanoic acid representing hydroxyl acids. The standard mixture was labeled with 12C-DnsHz under different reaction conditions. The same mixture was labeled with 13C-DnsHz under one condition, as internal standards, and then spiked to 12C-labeled samples with the same amount. The labeled samples were analyzed by LC-MS and the peak area ratio of each 12C-/13C-labeled metabolite was used to evaluate the relative efficiencies of different labeling conditions. Using this approach, we optimized the reaction buffer, reagents, time and temperature (Figure 2). Slightly acidic pH is required for the formation of O-acylisourea intermediate.38 Thus it is important to control the pH for carbodiimide-mediated coupling reaction. In this study, two types of buffer solution were tested with different pH values, including PBS with pH 3.5, 5.5, 7 and 8 and MES with pH 3.5 and 5.5. Figure 2A shows the effect of buffer solutions and pH on labeling efficiency. For all seven metabolites, reaction in MES solution behaved better than that in PBS. In the pH values tested, the results of using PBS as buffer solution indicated that pH 3.5 and 5.5 provided higher labeling efficiency than pH 7 and 8, confirming that the reaction needs relatively acidic condition. The results of using MES buffer indicated that reactions of hydrophilic metabolites performed better at pH 3.5 (i.e., labeling glucuronic acid, succinic acid and pyruvic acid), while more hydrophobic ones were better at 5.5 (i.e., labeling propionic acid and cholic acid). Although there was no consistent trend for all seven metabolites, the labeling results of solution at pH 3.5 were found to be still acceptable even for those hydrophobic acids. Therefore, MES solution at pH 3.5 was considered to be an optimal buffer solution in our method. We examined the effect of EDC concentration on the labeling reaction (Figure 2B). As expected, without EDC (0 mM), little if any labeled metabolites could be detected (except for


Page 9 of 28 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


9 glucuronic acid). Within the concentration range tested, 100 mM of EDC solution provided good labeling results for all seven metabolites and therefore was chosen as the optimal condition. In addition to carbodiimide, other additives can be used in coupling reaction to enhance labeling efficiency and reduce side reactions. In this work, we tested three additives: 1-hydroxy-7azabenzotriazole (HOAT), 4-dimethylaminopyridine (DMAP) and N-hydroxysuccinimide (NHS). We found that, for all metabolites, the use of HOAT generated consistently the highest labeling efficiency (Figure 2C). We then studied the effect of HOAT concentration on the labeling reaction (Figure 2D). The results suggested that, for a majority of the tested standards, HOAT was necessary for the reaction. However, the concentration of HOAT did not affect the labeling efficiency dramatically, except for pyruvic acid. Taken all together, we chose 10 mM HOAT solution for the labeling reaction. However, it should be noted that, for analyzing a special type of sample (e.g., cell extracts of small numbers of cells, sweat, etc.), it is best to fine-tune the reagent concentrations of both EDC and HOAT to produce an optimal workflow. The other two parameters affecting the reaction process were reaction temperature and time (Figures 2E and 2F). For temperature screening, we found that reaction at 20 ºC was best for all the tested metabolites. Using this mild condition is beneficiary to protect thermally unstable metabolites and prevent loss of volatile acids. At 20 ºC, reaction for 90 min was suitable for most of the standards. Although the amounts of labeled metabolites slightly increased after 90 min for some metabolites, considering both labeling efficiency and sample throughput, we chose 20 ºC for 90 min as the optimal labeling conditions. At last, a quench method was developed for DnsHz labeling of carboxylic acids in order to remove the remaining reagents that may interfere with the detection of labeled metabolites. DnsHz is a stable compound and can produce an intense signal in MS detection in positive ion mode.



Page 10 of 28

10 According to a recent report of using DnsHz labeling for developing a Cu2+-selective fluorescence signaling probe, copper (II) was found to speed up the hydrolysis of DnsHz in water without affecting labeled compounds.40 In our study, we found that adding copper (II) chloride could effectively remove the extra DnsHz in the labeling reaction solution. Supplemental Figure S2A shows the LC-UV chromatograms of DnsHz and CuCl2 mixtures prepared at different incubation times at 40 ºC. It is clear that CuCl2 could significantly reduce the DnsHz signal after incubating for 30 min. This result was confirmed using LC-MS. Supplemental Figure S2B shows the LC-MS chromatograms of 20 mM DnsHz incubated with or without CuCl2. Without adding CuCl2, a broad, saturated and tailed peak from DnsHz could be observed around 9 min, which may cause ion suppression for the detection of co-eluting labeled metabolites. However, this problem could be reduced by adding CuCl2 and incubating at 40 ºC for 30 min. Note that there are several additional peaks detected in LC-MS chromatograms. They were from background reactions which were largely suppressed in a real sample containing analytes (see, for example, labeled urine samples as shown below). These blank peaks could be filtered out in data processing. Validation of Labeling Conditions with Urine Sample. Urine was chosen as a representative complex metabolomic system to validate whether the optimized method is applicable for real biological samples. LC-MS analysis of 1:1 mixture of

12

C-/13C-labeled urine

was used to determine and compare the number of peak pairs detected to evaluate the performance of the labeling conditions. For each condition, data from six replicates were collected, including experimental triplicate of sample labeling and injection duplicate for each labeled sample. Figure 3A shows the validation of the buffer solution selection. We examined the performance of using MES solution, PBS solution and water (no buffering) for labeling. For the MES buffer, combinations of two pH values (pH 3.5 and pH 5.5) and two concentrations (0.1 M


Page 11 of 28 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


11 and 0.4 M) were tested. For the PBS buffer, three pH values of 0.1 M solution were tested (pH 3.5, 5.5 and 7). The results confirmed the finding from the standard mixture experiments that the MES buffer solution performed better than the PBS buffer. However, we also found that the concentration and pH of the MES buffer solution had no significant effects on urine labeling, in terms of the number of peak pairs detected. Even without adding any buffer (the H2O group), the peak pair number was similar to that of the MES buffer group. This can be attributed to the internal urinary buffering system, which keeps normal urine pH values ranging from 4.5 to 8. We tested the urine samples in this study and found that the pH value was about 5.0, which is in the optimal pH range for DnsHz carboxyl labeling. Thus, even without adding external buffer solutions, comparable results of labeling reaction were generated. However, for practical purpose of analyzing different urine samples (i.e., in case of encountering urine with higher pH in a metabolomics study), the use of 0.1 M MES buffer solution is recommended. We examined the effects of reagent type and concentration for urine sample labeling. The results were in agreement with those of standard mixture experiments, i.e., 100 mM of EDC solution produced the highest peak pair numbers (Figure 3B). Among all the additives tested, adding HOAT into the reaction mixture increased the peak pair numbers (Figure 3C). And within the concentrations we tested, the concentration of HOAT had little effect on the peak pair numbers (Figure 3D). Figures 3E and 3F show the results of reaction time and temperature validation, respectively. They were also in agreement with those of standard mixture labeling. Incubating at 20 ºC for 1.5 h was therefore chosen as an optimized condition. Overall, based on the results of standard mixture and urine labeling experiments, it can be concluded that the optimal labeling condition is to use 100 mM of EDC as a coupling reagent and 10 mM of HOAT as the additive. Both reagents are dissolved in 0.1 M MES buffer solution at pH



Page 12 of 28

12 3.5. Reaction mixtures are shaken at 20 ºC for 90 min, followed by adding 100 mM of CuCl2 as a quenching reagent and incubating at 40 ºC for another 30 min. The entire labeling process can be completed in about 2 hr. Metabolite Detectability. To demonstrate the detectability of the developed method for profiling complex samples, the urinary carboxyl submetabolomes from experimental triplicates of labeled urine samples were analyzed. Figure 4A shows a typical base-peak ion chromatogram of 12

C-/13C-DanHz labeled human urine. Many peaks were detected within the whole gradient. We

also optimized the sample injection amount in order to detect as many labeled metabolites as possible. To do this, different volumes of 12C-/13C-DanHz labeled urine were injected into LC-MS for analysis and the peak pair numbers were then compared. Figure 4B shows the number of peak pairs as a function of injection volume. A plateau was observed at 8 µL injection. Venn diagram was used to evaluate the labeling reproducibility among experimental triplicates of sample processing (Figure 4C). An average of 2597±28 pairs per run were detected with 2266 pairs in common, indicating high reproducibility of peak pair detection. The distribution of absolute intensities of these common peak pairs is presented in Figure 4D (in black). As the peak intensity decreased, more peak pairs could be detected, which suggests that the developed method could effectively detect metabolites with very low abundance or low signal intensity, leading to high metabolome coverage. It should be noted that the number of peak pairs detected was similar to that from DmPA labeling. In addition, in this example, we did not observe any coincident overlapping of a peak from a

13

C-labeled metabolite with a peak from a

12

C-labeled

different metabolite at a given retention time. It suggests that although differential isotope labeling increases the number of peaks detected, it does not cause additional peak overlapping interference when high resolution LC separation and m/z detection are used. The dansylhydrazine labeling


Page 13 of 28 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


13 reaction is not reversible. After labeling, the labeled samples were very stable and could meet the needs of long term studies (e.g., running samples over one month) (see Supplemental Note S3). Quantification. In CIL LC-MS method, relative quantification of metabolites is realized by measuring peak ratio between the 12C-labeled metabolite from individual samples and its 13Clabeled counterpart in the pooled sample. We note that the overall labeling yield of a given metabolite may be dependent on the metabolite type and the concentration. However, in our work, the reagent amount is always kept very high to drive the reaction to the maximal extent. More importantly, for relative quantification, the pooled sample has an average concentration of a given metabolite, which is not very different from that of an individual sample (the fold-change is usually 1.5). We plotted the peak ratios of all pairs against their absolute signal intensities (Figure 5C). To make the plot clearer, logarithm transformation was used to convert the peak ratios and intensity values. Most of the pairs with large ratio errors were the low intensity peak pairs [i.e., absolute intensity of less than 2×105 or log (peak intensity) < 5.3]. By plotting the distributions of the absolute intensities of peak pairs that have ratios of larger than 1.25 or smaller than 0.8 (Figure 4D, in red), we can observe more clearly that low intensity pairs had higher percentage of values outside the ±20% range. This might be caused by poor peak shape of low intensity peak, which made peak area calculation more difficult, resulting in larger errors in ratio calculation. This observation suggests that, for low intensity peaks, great care should be exercised in relative quantification. This is actually true even for targeted analysis of metabolites; quantification limit is often defined as the lowest signal at S/N=10, while detection limit is defined at S/N=3. Future work of investigating the relation between signal intensity and quantification accuracy for different ratios of 12C-labeled sample vs. 13C-labeled pool is warranted. Nevertheless, the above results already indicate that good accuracy and precision could be obtained for most of the peak pairs. We carried out a proof-of-concept project involving metabolome comparison of 30 urine samples collected from 10 individuals (three daily collections per individual) and absolute quantification of butyric acid and adipic acid in urine. Supplemental Note S2 describes the results showing the capability of DanHz labeling LC-MS for quantitative metabolomics. Metabolite Identification. Metabolite identification is another important aspect in metabolome analysis. To perform high confidence metabolite identification, usually at least two independent properties of unknown metabolites from samples should be matched to those of authentic standards. We constructed a DnsHz-labeled carboxyl standard library which currently


Page 15 of 28 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


15 contains 193 human endogenous carboxyl-containing metabolites. These metabolites cover 55 metabolic pathways (Supplemental Table S2). Each standard was labeled with DnsHz individually and then analyzed by LC-MS and LC-MS/MS. Triple-parameters information for each standard was collected: accurate mass, retention time (RT) and MS/MS fragmentation spectrum (Supplemental Table S1). To perform identification, the measured mass, RT and/or MS/MS of an unknown labeled metabolite from a sample could be used to search against the library using a website-based program DnsHz-ID.15, 41 Both of the library and search program are freely available from a publicly accessible website at www.mycompoundid.org. Accurate mass and MS/MS matches can be performed directly after uploading the corresponding information of unknown samples to the website. However, RT is easily affected by many possible variations in LC-MS conditions, such as instrument or column brand, connection tubing length, etc. To use RT as an identification parameter, a RT normalization method reported previously41 was used to correct the RT shift caused by these variations. A set of DnsHz labeled compounds, which was initially developed for identification of DnsHz-labeled carbonyl-containing metabolites15, was used as RT calibration mixture since the same LC gradient was used for the analysis of the two submetabolomes. The RTs of 193 standards collected for the library were all normalized to this RT calibration mixture. When performing metabolome analysis using the presented method in a different instrument, the same RT calibration mixture needs to be run on the new instrumental setup. Then the RTs of unknown metabolites will be corrected automatically using the DnsHz-ID program based on the local RT calibration mixture, followed by identification using RT. Putative identification based on the use of accurate mass match alone to the existing metabolite database is less confident. However, it can still provide valuable identity information of unknown metabolites and hints for further confirmation, such as purchasing or synthesizing



Page 16 of 28

16 standards. Therefore, in this study, we used a two-tier identification approach with positive identification in tier 1 and putative identification in tier 2. In tire 1, among the common 2266 peak pairs detected in six replicates of labeled urine samples, we positively identified 81 peak pairs based on accurate mass and retention time matches using the constructed labeled metabolite library (Supplemental Table S3). Note that the current library of labeled carboxyl standards does not include any metabolites that are already in the existing libraries of labeled amines/phenols, carbonyls and hydroxyls. Thus, some of metabolites with two or more labeliable groups such as glucuronic acid were not shown in Table S3, as they could already be positively identified in other labeling methods (e.g., the labeled carbonyl standard library contains glucuronic acid). In tier 2, the remaining peak pairs were searched, based on accurate mass match, against the MyCompoundID (MCID) library composed of 8,021 known human endogenous metabolites (zero-reaction library) and their predicted metabolic products from one metabolic reaction (375,809 compounds) (one-reaction library). 517 and 1445 peak pairs were matched in the zero- and one-reaction libraries, respectively (Supplemental Tables S4 and S5). Thus, out of 2266 unique peak pairs detected, 2043 pairs (90.2%) could be positively identified or putatively matched. It is not surprising that we observed some peak pairs were matched to carbonyl-containing metabolites in putative identification. As reported before, 12C-/13C-DnsHz can be used for carbonyl submetabolome profiling using different reaction conditions (40 ºC for 60 minutes) and reagents (hydrogen chloride).15 Although the reaction buffers for two submetabolomes are different, both of the reactions are carried out in acidic condition. Therefore, it is possible to label some carbonylcontaining metabolites when profiling carboxyl submetabolome using this method. When we constructed the labeled carboxyl standard library, we actually found some metabolites containing


Page 17 of 28 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


17 both carbonyl and carboxylic acid groups could be labeled with two DnsHz tags, such as pyruvic acid. In this kind of situation, dansylhydrazine could react with carbonyl groups to generate a sulfonyl hydrazone structure. However, achieving the highest selectivity is not the main goal of the CIL LC-MS approach for high-coverage metabolome profiling, as eventually we will combine data from different submetabolomes to generate the entire metabolome results. On the other hand, it also provides an opportunity to combine the two reaction products for a combined LC-MS analysis of the carbonyl and carboxyl submetabolomes which should increase the throughput of sample analysis. We will report a detailed study of the comparison of carbonyl and carboxyl submetabolomes of various types of samples, as well as the comparison of individual submetabolome analysis vs. combined analysis in the future.

Acknowledgements This work was supported by the Natural Sciences and Engineering Research Council of Canada, the Canada Research Chairs program, Canada Foundation for Innovations, Genome Canada, Genome Alberta and Alberta Innovates.

Disclosure The authors declare no competing financial interest.

Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: (TBA)



Page 18 of 28

18 •

Notes for experimental, proof-of-concept project and stability test; Tables showing the standard list, pathways covered and metabolite identification results; Figures showing the chemical structures of standards, LC-UV and LC-MS results.


Page 19 of 28 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


19

References 1. Arrivault, S.; Guenther, M.; Fry, S. C.; Fuenfgeld, M.; Veyel, D.; Mettler-Altmann, T.; Stitt, M.; Lunn, J. E., Synthesis and Use of Stable-Isotope-Labeled Internal Standards for Quantification of Phosphorylated Metabolites by LC-MS/MS. Anal Chem 2015, 87, 6896-6904. 2. Chu, J. M.; Qi, C. B.; Huang, Y. Q.; Jiang, H. P.; Hao, Y. H.; Yuan, B. F.; Feng, Y. Q., Metal Oxide-Based Selective Enrichment Combined with Stable Isotope Labeling-Mass Spectrometry Analysis for Profiling of Ribose Conjugates. Anal Chem 2015, 87, 7364-7372. 3. Dai, W. D.; Huang, Q.; Yin, P. Y.; Li, J.; Zhou, J.; Kong, H. W.; Zhao, C. X.; Lu, X.; Xu, G. W., Comprehensive and Highly Sensitive Urinary Steroid Hormone Profiling Method Based on Stable Isotope-Labeling Liquid Chromatography Mass Spectrometry. Anal Chem 2012, 84, 10245-10251. 4. Hao, L.; Johnson, J.; Lietz, C. B.; Buchberger, A.; Frost, D.; Kao, W. J.; Li, L. J., Mass Defect-Based N,N-Dimethyl Leucine Labels for Quantitative Proteomics and Amine Metabolomics of Pancreatic Cancer Cells. Anal Chem 2017, 89, 1138-1146. 5. Leng, J. P.; Wang, H. Y.; Zhang, L.; Zhang, J.; Wang, H.; Guo, Y. L., A highly sensitive isotope-coded derivatization method and its application for the mass spectrometric analysis of analytes containing the carboxyl group. Anal. Chim. Acta 2013, 758, 114-121. 6. Mochizuki, T.; Todoroki, K.; Inoue, K.; Min, J. Z.; Toyo'oka, T., Isotopic variants of light and heavy L-pyroglutamic acid succinimidyl esters as the derivatization reagents for DLamino acid chiral metabolomics identification by liquid chromatography and electrospray ionization mass spectrometry. Anal. Chim. Acta 2014, 811, 51-59. 7. Tayyari, F.; Gowda, G. A. N.; Gu, H. W.; Raftery, D., N-15-Cholamine-A Smart Isotope Tag for Combining NMR- and MS-Based Metabolite Profiling. Anal Chem 2013, 85, 8715-8721. 8. Wagner, M.; Ohlund, L. B.; Shiao, T. C.; Vezina, A.; Annabi, B.; Roy, R.; Sleno, L., Isotope-labeled differential profiling of metabolites using N-benzoyloxysuccinimide derivatization coupled to liquid chromatography/high-resolution tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2015, 29, 1632-1640. 9. Wong, J. M. T.; Malec, P. A.; Mabrouk, O. S.; Ro, J.; Dus, M.; Kennedy, R. T., Benzoyl chloride derivatization with liquid chromatography-mass spectrometry for targeted metabolomics of neurochemicals in biological samples. J. Chromatogr. A 2016, 1446, 78-90. 10. Yang, W. C.; Regnier, F. E.; Jiang, Q.; Adamec, J., In vitro stable isotope labeling for discovery of novel metabolites by liquid chromatography-mass spectrometry: Confirmation of gamma-tocopherol metabolism in human A549 cell. J. Chromatogr. A 2010, 1217, 667-675. 11. Yuan, W.; Edwards, J. L.; Li, S. W., Global profiling of carbonyl metabolites with a photo-cleavable isobaric labeling affinity tag. Chem. Commun. 2013, 49, 11080-11082. 12. Yuan, W.; Zhang, J. X.; Li, S. W.; Edwards, J. L., Amine Metabolomics of Hyperglycemic Endothelial Cells using Capillary LC-MS with Isobaric Tagging. Journal of Proteome Research 2011, 10, 5242-5250. 13. Guo, K.; Li, L., Differential C-12/C-13-Isotope Dansylation Labeling and Fast Liquid Chromatography/Mass Spectrometry for Absolute and Relative Quantification of the Metabolome. Anal Chem 2009, 81, 3919-3932. 14. Zhao, S.; Luo, X.; Li, L., Chemical Isotope Labeling LC-MS for High Coverage and Quantitative Profiling of the Hydroxyl Submetabolome in Metabolomics. Anal Chem 2016, 88, 10617-10623.



Page 20 of 28

20 15. Zhao, S.; Dawe, M.; Guo, K.; Li, L., Development of High-Performance Chemical Isotope Labeling LC-MS for Profiling the Carbonyl Submetabolome. Anal Chem 2017, 89, 6758-6765. 16. Guo, K.; Li, L., High-Performance Isotope Labeling for Profiling Carboxylic AcidContaining Metabolites in Biofluids by Mass Spectrometry. Anal Chem 2010, 82, 8789-8793. 17. Yang, W. C.; Sedlak, M.; Regnier, F. E.; Mosier, N.; Ho, N.; Adamec, J., Simultaneous Quantification of Metabolites Involved in Central Carbon and Energy Metabolism Using Reversed-Phase Liquid Chromatography-Mass Spectrometry and in Vitro C-13 Labeling. Anal Chem 2008, 80, 9508-9516. 18. Bollinger, J. G.; Thompson, W.; Lai, Y.; Oslund, R. C.; Hallstrand, T. S.; Sadilek, M.; Turecek, F.; Gelb, M. H., Improved Sensitivity Mass Spectrometric Detection of Eicosanoids by Charge Reversal Derivatization. Anal Chem 2010, 82, 6790-6796. 19. Yang, W. C.; Regnier, F. E.; Adamec, J., Comparative metabolite profiling of carboxylic acids in rat urine by CE-ESI MS/MS through positively pre-charged and H-2-coded derivatization. Electrophoresis 2008, 29, 4549-4560. 20. Yang, W. C.; Adamec, J.; Regnier, F. E., Enhancement of the LC/MS analysis of fatty acids through derivatization and stable isotope coding. Anal Chem 2007, 79, 5150-5157. 21. Kloos, D.; Derks, R. J. E.; Wijtmans, M.; Lingeman, H.; Mayboroda, O. A.; Deelder, A. M.; Niessen, W. M. A.; Giera, M., Derivatization of the tricarboxylic acid cycle intermediates and analysis by online solid-phase extraction-liquid chromatography-mass spectrometry with positive-ion electrospray ionization. J. Chromatogr. A 2012, 1232, 19-26. 22. Koulman, A.; Petras, D.; Narayana, V. K.; Wang, L.; Volmer, D. A., Comparative HighSpeed Profiling of Carboxylic Acid Metabolite Levels by Differential Isotope-Coded MALDI Mass Spectrometry. Anal Chem 2009, 81, 7544-7551. 23. Williams, J.; Pandarinathan, L.; Wood, J.; Vouros, P.; Makriyannis, A., Endocannabinoid metabolomics: A novel liquid chromatography-mass spectrometry reagent for fatty acid analysis. Aaps Journal 2006, 8, E655-E660. 24. Bartok, T.; Komoroczy, R.; Borcsok, G.; Sagi, F.; Hegyes, P.; Bartok, M., A new derivatization reaction for the identification and determination of organic acids by positive-ion electrospray mass spectrometry. Rapid Commun. Mass Spectrom. 1997, 11, 133-135. 25. Zhu, Q. F.; Zhang, Z.; Liu, P.; Zheng, S. J.; Peng, K.; Deng, Q. Y.; Zheng, F.; Yuan, B. F.; Feng, Y. Q., Analysis of liposoluble carboxylic acids metabolome in human serum by stable isotope labeling coupled with liquid chromatography-mass spectrometry. J. Chromatogr. A 2016, 1460, 100-109. 26. Yuan, B. F.; Zhu, Q. F.; Guo, N.; Zheng, S. J.; Wang, Y. L.; Wang, J.; Xu, J.; Liu, S. J.; He, K.; Hu, T.; Zheng, Y. W.; Xu, F. Q.; Feng, Y. Q., Comprehensive Profiling of Fecal Metabolome of Mice by Integrated Chemical Isotope Labeling-Mass Spectrometry Analysis. Anal Chem 2018, 90, 3512-3520. 27. Marquis, B. J.; Louks, H. P.; Bose, C.; Wolfe, R. R.; Singh, S. P., A New Derivatization Reagent for HPLC-MS Analysis of Biological Organic Acids. Chromatographia 2017, 80, 17231732. 28. Jiang, R. Q.; Jiao, Y.; Zhang, P.; Liu, Y.; Wang, X.; Huang, Y.; Zhang, Z. J.; Xu, F. G., Twin Derivatization Strategy for High-Coverage Quantification of Free Fatty Acids by Liquid Chromatography-Tandem Mass Spectrometry. Anal Chem 2017, 89, 12223-12230.


Page 21 of 28 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


21 29. Han, J.; Gagnon, S.; Eckle, T.; Borchers, C. H., Metabolomic analysis of key central carbon metabolism carboxylic acids as their 3-nitrophenylhydrazones by UPLC/ESI-MS. Electrophoresis 2013, 34, 2891-2900. 30. Han, J.; Lin, K.; Sequeira, C.; Borchers, C. H., An isotope-labeled chemical derivatization method for the quantitation of short-chain fatty acids in human feces by liquid chromatography-tandem mass spectrometry. Anal. Chim. Acta 2015, 854, 86-94. 31. Han, J.; Liu, Y.; Wang, R. X.; Yang, J. C.; Ling, V.; Borchers, C. H., Metabolic Profiling of Bile Acids in Human and Mouse Blood by LC-MS/MS in Combination with PhospholipidDepletion Solid-Phase Extraction. Anal Chem 2015, 87, 1127-1136. 32. Harder, U.; Koletzko, B.; Peissner, W., Quantification of 22 plasma amino acids combining derivatization and ion-pair LC-MS/MS. J. Chromatogr. B 2011, 879, 495-504. 33. O'Maille, G.; Go, E. P.; Hoang, L.; Want, E. J.; Smith, C.; O'Maille, P.; Nordstrom, A.; Morita, H.; Qin, C.; Uritboonthai, W.; Apon, J.; Moore, R.; Garrett, J.; Siuzdak, G., Metabolomics relative quantitation with mass spectrometry using chemical derivatization and isotope labeling. Spectr.-Int. J. 2008, 22, 327-343. 34. Zhou, R.; Tseng, C. L.; Huan, T.; Li, L., IsoMS: Automated Processing of LC-MS Data Generated by a Chemical Isotope Labeling Metabolomics Platform. Anal Chem 2014, 86, 46754679. 35. Huan, T.; Li, L., Counting Missing Values in a Metabolite-Intensity Data Set for Measuring the Analytical Performance of a Metabolomics Platform. Anal Chem 2015, 87, 13061313. 36. Huan, T.; Li, L., Quantitative Metabolome Analysis Based on Chromatographic Peak Reconstruction in Chemical Isotope Labeling Liquid Chromatography Mass Spectrometry. Anal Chem 2015, 87, 7011-7016. 37. Li, L.; Li, R. H.; Zhou, J. J.; Zuniga, A.; Stanislaus, A. E.; Wu, Y. M.; Huan, T.; Zheng, J. M.; Shi, Y.; Wishart, D. S.; Lin, G. H., MyCompoundID: Using an Evidence-Based Metabolome Library for Metabolite Identification. Anal. Chem. 2013, 85, 3401-3408. 38. Nakajima, N.; Ikada, Y., MECHANISM OF AMIDE FORMATION BY CARBODIIMIDE FOR BIOCONJUGATION IN AQUEOUS-MEDIA. Bioconjugate Chem. 1995, 6, 123-130. 39. Staros, J. V.; Wright, R. W.; Swingle, D. M., ENHANCEMENT BY NHYDROXYSULFOSUCCINIMIDE OF WATER-SOLUBLE CARBODIIMIDE-MEDIATED COUPLING REACTIONS. Anal. Biochem. 1986, 156, 220-222. 40. Choi, M. G.; Kim, J.; Hong, J. M.; Chang, I. J.; Ahn, S.; Chang, S. K., Cu2+-selective ratiometric fluorescence signaling probe based on the hydrolysis of dansylhydrazine. Tetrahedron Lett. 2016, 57, 975-978. 41. Huan, T.; Wu, Y. M.; Tang, C. Q.; Lin, G. H.; Li, L., DnsID in MyCompoundID for Rapid Identification of Dansylated Amine- and Phenol-Containing Metabolites in LC-MS-Based Metabolomics. Anal Chem 2015, 87, 9838-9845.



Page 22 of 28

22

Figure Captions Figure 1.

Workflow of differential CIL LC-MS method using

12

C-/13C-DnsHz labeling for

carboxyl submetabolome analysis. Figure 2.

Comparison of relative efficiency for labeling a standard mixture under different reaction conditions: effect of (A) buffer solution and pH, (B) EDC concentration, (C) additive type, (D) HOAT concentration, (E) reaction temperature, and (F) reaction time. Data are presented as the mean ± S.D. of six replicates (n=6).

Figure 3.

Peak pair numbers detected from

12

C-/13C-DnsHz labeled urine samples prepared

under different conditions: effect of (A) buffer solution and pH, (B) EDC concentration, (C) additives type, (D) HOAT concentration, (E) temperature and (F) time. Data are presented as mean ± S.D. from experimental triplicate and injection duplicate (n=6). Figure 4.

(A) LC-MS ion chromatogram of 12C-/13C-DnsHz labeled urine. (B) Peak pair number detected as a function of injection amount of 12C-/13C-DnsHz labeled urine. Data are presented as mean ± S.D. (n=3). (C) Venn diagram of the peak pair numbers detected in triplicate labeled urine. (D) Peak pair number detected as a function of peak intensity (black: all peak pairs; red: peak pairs with peak ratio >1.25 or

Dansylhydrazine Isotope Labeling LC-MS for Comprehensive

Recommend Documents