Method to Calculate the Retention Index in Hydrophilic Interaction

Subscriber access provided by UNIV OF LOUISIANA

Article

A Method to Calculate the Retention Index in Hydrophilic Interaction Liquid Chromatography Using Normal Fatty Acid Derivatives as Calibrants and Its Application in Metabolomics Analysis Quan-Fei Zhu, Tian-Yi Zhang, Lin-Lin Qin, Xin-Ming Li, Shu-Jian Zheng, and Yu-Qi Feng Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.9b00598 • Publication Date (Web): 03 Apr 2019 Downloaded from http://pubs.acs.org on April 4, 2019

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

1

A Method to Calculate the Retention Index in Hydrophilic

2

Interaction Liquid Chromatography Using Normal Fatty Acid

3

Derivatives as Calibrants

4 5

Quan-Fei Zhu,† Tian-Yi Zhang,† Lin-Lin Qin, Xin-Ming Li, Shu-Jian Zheng, Yu-Qi

6

Feng *

7 8

Key Laboratory of Analytical Chemistry for Biology and Medicine (Ministry of

9

Education), Department of Chemistry, Wuhan University, Wuhan 430072, P.R. China

10

†These authors contributed equally to this work.

11

*To whom correspondence should be addressed. Tel.: +86-27-68755595; fax: +86-27-

12

68755595. E-mail address: [email protected].

13 14

1

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

1

Page 2 of 28

Abstract Hydrophilic interaction liquid chromatography-mass spectrometry (HILIC-MS) is

2 3

a

complementary

technique

to

reversed-phase

liquid

4

spectrometry (RPLC-MS) and has been widely used to expand the coverage of the

5

metabolome in MS-based metabolomics. However, the use of HILIC retention time

6

(HILIC RT) in metabolites annotation is quite limited because of its poor

7

reproducibility. Here, we developed a method to calculate the retention index in HILIC

8

(HILIC RI) for calibration of HILIC RT. In this method, a mixture of 2-

9

dimethylaminoethylamine (DMED)-labeled fatty acid standards with carbon chain

10

length from C2 to C22 were selected as calibrants to establish a linear calibration

11

equation between HILIC RT and carbon number for the calculation of HILIC RI. The

12

calculated HILIC RIs based on a regression equation could efficiently calibrate the

13

retention time shifts for 28 DMED-labeled carboxyl standards and DMED-labeled

14

carboxyl metabolites in rat urine, serum and feces on a HILIC column with different

15

gradient elution conditions. Furthermore, the developed HILIC RI strategy was applied

16

to RT calibration of screened metabolites, the annotation of isomers in HILIC-MS-

17

based metabolomics analysis for real samples, and the correction of isotope effects in

18

chemical isotope labeling-HILIC-MS analysis. Taken together, the resulting HILIC RI

19

strategy is a promising analytical technique to improve the accuracy of metabolite

20

annotation; it would be widely used in HILIC-MS-based metabolome analysis.

21

Keywords: Hydrophilic interaction liquid chromatography, Retention index, Mass

22

spectrometry, Metabolite annotation, Isotope effect 2


chromatography-mass

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


1

INTRODUCTION

2

Liquid chromatography-mass spectrometry (LC-MS) is a powerful tool for

3

metabolomics. It can screen metabolites with high coverage and offer quantitative

4

information on metabolites with high sensitivity1-5. However, in LC-MS-based

5

metabolomics analysis, accurate annotation of metabolites remains difficult especially

6

for the identification of low-abundance metabolites6. In the current LC-MS-based

7

metabolomics analysis, conventional approaches to metabolite annotation are usually

8

based on accurate mass (AM) and/or MS/MS spectra matching with metabolite-specific

9

databases such as Metlin, HMDB, and KEGG; MS/MS spectra and retention times (if

10

available); comparisons via authentic compounds; or postulation-based on MS/MS

11

fragment pathways7-9. However, it is difficult to obtain high-quality MS/MS spectra for

12

low abundance metabolite in complex samples6. In addition, the chromatographic

13

retention time (RT) of metabolites is strongly dependent on the chromatographic

14

conditions; it usually has a large deviation, which limits its utility in measuring retention

15

time10,11.

16

Hydrophilic interaction liquid chromatography coupled to mass spectrometry

17

(HILIC-MS) is a complementary approach to reversed-phase liquid chromatography-

18

mass spectrometry (RPLC-MS). HILIC-MS has been widely used for the untargeted

19

profiling of polar metabolites12. However, slight changes in chromatographic

20

conditions (such as pH, gradients, and matrices) can cause RT shifts in HILIC and even

21

affect the elution order of the compounds6. In addition, the RTs of many metabolites in

22

HILIC are often very close due to insufficient resolution13,14. Therefore, HILIC RT shift 3



1

between the detected metabolite and the authentic standard was set to less than 0.3 min

2

for peaks matching the metabolite annotation6. RT deviation criteria would increase the

3

error rate of peak matching and reduce the accuracy of metabolite annotation. Hence, a

4

stable and reliable HILIC RT correction method could be helpful to improve the

5

accuracy of metabolite annotation in HILIC-MS-based metabolomics analysis.

6

Several approaches to the alignment of RTs for LC-MS-based metabolomics have

7

been developed. Li’s group employed 22 dansyl-labeled standards as the calibrants, and

8

a local linear regression calibration method was utilized to calibrate the shift in RT15.

9

Similarly, Xu and co-workers adopted a calibration method to reduce the RT difference

10

between experimental data and in-house database by choosing 11 internal standards for

11

positive ion mode and 9 internal standards for negative ion mode during the LC-MS

12

analysis11,16. In addition, a standard quality control (QC) strategy was used to assess

13

reproducibility and instrument performance during HILIC-MS analysis. This method

14

requires several consecutive injections of QC standard samples (usually 510 times)

15

before and during the LC-MS run to determine the stability of the analysis state, which

16

will increase the analysis time and sample consumption17,18.

17

More recently, the retention index (RI) that was already widely used in GC-MS

18

has been introduced into RPLC-MS-based metabolomics analysis to convert a variable

19

retention time to a stable retention value. Quilliam et al. developed a new RI system

20

based on a homologous series of substances, i.e., N-alkyl-3-pyridine-sulfonate (NAPS)

21

standards (n=019) for RPLC-MS19,20. Good accuracies were achieved among five

22

laboratories using different RPLC-MS systems. Recently, our group established the 4


Page 4 of 28

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


1

chemical labeling-based RPLC retention index (CL-RPLC RI) of carboxyl and amino

2

metabolites in CL-RPLC-MS analysis21. This work used a mixture of chemically

3

labeled saturated fatty acids and fatty amines as calibrants. The obtained RIs exhibited

4

much better reproducibility than RTs under different conditions including gradients,

5

columns, and instrument systems for untargeted metabolic profiling. However, only a

6

few approaches have been reported to calibrate the RTs of metabolites in the HILIC-

7

MS-based metabolomics analysis.

8

Here, we proposed a method to calculate retention index in HILIC (HILIC RI) for

9

the improvement of the accuracy of peak matching in metabolomics analysis. Based on

10

the carbon number rule, a mixture of 2-dimethylaminoethylamine (DMED)-labeled

11

fatty acid standards with carbon chain length from C2 to C22 was used as a calibrant

12

for constructing the HILIC RI linear calibration equation. Here, 28 DMED-labeled

13

carboxyls in standard solution and DMED labeled carboxyl metabolites in rat urine,

14

serum and feces were used to evaluate the reproducibility of HILIC RIs and RTs under

15

different chromatographic gradients. The HILIC RI strategy was then applied to RT

16

calibration of screened metabolites and isomers identification in HILIC-MS-based

17

metabolomic analysis, and correction of isotopic effect in chemical isotope labeling

18

assisted HILIC-MS analysis. The results demonstrated that the HILIC RI strategy

19

proposed here is a promising analytical tool to improve the accuracy of metabolites

20

annotation in HILIC-MS-based metabolomics.

21 22

EXPERIMENTAL SECTION 5



1

Page 6 of 28

Chemicals and Reagents

2

Standards of fatty acids were purchased from Sigma-Aldrich (St. Louis, MO,

3

USA), J&K Chemical (Beijing, China), Cayman Chemical (Arbor, MI, USA), and

4

Aladdin (Shanghai, China). Detailed information on the fatty acids and 28 carboxylic

5

acids is listed in Tables S1 and S2. Analytical grade ammonium formate, 2-chloro-1-

6

methylpyridinium

7

dimethylaminoethylamine (DMED) were supplied by Sinopharm Chemical Reagent Co.

8

Ltd. (Shanghai, China). HPLC-grade acetonitrile (ACN) and methanol (MeOH) were

9

obtained from Merck (Darmstadt, Germany). Water was purified by a Milli-Q water

10

purification apparatus (Bedford, MA, USA). The isotope-labeling reagents for d4-

11

DMED were synthesized in our group according to our previous work22. The stock

12

solutions of DMED (20 μmol/mL), d4-DMED (20 μmol/mL), CMPI (20 μmol/mL), and

13

TEA (20 μmol/mL) were prepared in HPLC-grade ACN. The stock solutions of all

14

standards were prepared in HPLC-grade ACN at 1 mg/mL for each. All stock solutions

15

were stored at -20°C.

16

Sample Collection and Preparation

iodide

(CMPI),

triethylamine

(TEA),

and

2-

17

Rat serum, feces, and urine were collected from one Sprague Dawley (SD) male

18

rat supplied by Wuhan Institute of Physics and Mathematics, Chinese Academy of

19

Sciences (CAS, Wuhan, Hubei, China). The detailed sample collection procedures were

20

described in Supporting Information.

21

Different extraction procedures were employed to achieve efficient extraction of

22

metabolites from serum, feces, and urine of rat with reference to reported methods23,24. 6


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


1

Detailed procedures for extraction can be found in Supporting Information.

2

The DMED and d4-DMED labeling reactions were performed under optimized

3

reaction conditions referencing our previously reported method25. The detail reaction

4

conditions can be found in Supporting Information.

5

LC-MS Analysis

6

The LC-MS analysis was performed on two instrument systems: a Shimadzu MS-

7

8045 triple quadrupole mass spectrometer (Tokyo, Japan) coupled with a Shimadzu

8

LC-30AD UPLC system (Tokyo, Japan) as well as an Orbitrap Fusion Tribrid mass

9

spectrometer (Thermo Fisher Scientific, Rockford, IL USA) coupled with an UltiMate

10

3000 UHPLC System (Thermo Fisher Scientific, USA). The detailed MS experimental

11

parameters were described in Supporting Information. The LC separation was

12

performed at 35°C on four columns: (1) a Waters Acquity UPLC® BEH HILIC (2.1 ×

13

100 mm, 1.7 μm, serial number 03003634818313); (2) a Waters Acquity UPLC® BEH

14

HILIC (2.1 × 100 mm, 1.7 μm, serial number 03493714518642); (3) a Waters Acquity

15

UPLC® BEH Amide (2.1 × 100 mm, 1.7 μm, serial number 01673813418667); and (4)

16

a Merck SeQuant® ZIC®-HILIC (2.1 × 150 mm, 3.5 μm, serial number 843160). The

17

flow rate was set at 0.3 mL/min. The 10 mM ammonium formate in water (solvent A)

18

and ACN (solvent B) was employed as a mobile phase for analysis of metabolites. Two

19

gradients were used: (1) 01 min 95% B, 125 min 9565% B, 2532 min 65% B,

20

3235 min 6595% B, and 3550 min 95% B; (2) 03 min 95% B, 330 min 9585%

21

B, 3035 min 8565% B, 3540 min 65% B, 4041 min 6595% B, and 4160 min

22

95% B. 7



1

Data Processing

2

Untargeted HILIC full-scan MS data were extracted by the software Compound

3

Discover (version 2.0, Thermo Fisher Scientific, Rockford, IL, USA). The peak

4

detection results including RT, m/z, and peak intensity were exported as an Excel table.

5

The raw data were then matched with a defined mass difference (4 Da for DMED/d4-

6

DMED-labeled carboxyl metabolites); similar RTs or RIs and intensities ratios were

7

within 0.661.33 using in-house MATLAB-based software.

8 9 10

RESULTS AND DISCUSSION Establishment of the HILIC Retention Index

11

Previous studies have shown that the retention behavior of DMED-labeled fatty

12

acids on RPLC follows the classical carbon number rule: The RTs of DMED-labeled

13

FAs increase with their carbon number21,26. Herein, we investigated the retention

14

behavior of DMED-labeled FAs on a HILIC column (Acquity UPLC® BEH HILIC

15

column, serial number 03003634818313) and found that the retention of DMED-

16

labeled FAs (C2C22) exhibited the opposite trend of that on the reverse-phase LC

17

column. The retention time of DMED-labeled FAs on the HILIC column decreased

18

with increasing carbon number of FAs (C2C22) covering the retention time window

19

from 7.37 to 18.58 min (Figure 1A). In addition, the logarithmic retention time of

20

DMED-labeled FAs (C2C22) linearly decreases with increasing logarithmic carbon

21

number (Figure 1B).

22

A linear regression equation (equation 1) can be obtained by analyzing the 8


Page 8 of 28

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


1

relationship of carbon number of FA and RT: y = 1.41 - 0.40x, where y denotes log10RT,

2

and x denotes log10C. The coefficient of determination R2 of the fitting equation was

3

0.9950, indicating a good correlation between the carbon number (C) and RT. If we

4

define the retention index as RI = 100 × log10C, then the fitting equation can be used to

5

calculate the RI according to equation 2 from the measured RT of the metabolites in the

6

HILIC-MS run:

7

log10𝑅𝑇 = 𝑏 + 𝑎log10𝐶

8

𝑅𝐼 = 100 × log10𝐶 = 100 ×

9

where the values of a and b represent the slope and intercept of equation 1, respectively.

10

The resulting model to calculate HILIC RI has two advantages compared with the

11

local linear regression method: 1) The establishment of HILIC RI can be achieved with

12

a few FAs and does not require a large number of neighboring FA standards for local

13

calculation; 2) RIs of metabolites whose RTs outside the calibrating RT window can

14

also be obtained using the calibration equation.

(1) log10𝑅𝑇 ― 𝑏

(2)

𝑎

15

To verify the applicability of the established HILIC RI strategy on different HILIC

16

separation columns, we investigated the retention behavior of DMED-labeled FAs

17

(C2C22) on four widely used HILIC columns with different batches, manufacturers,

18

and bonded phases (detailed information on the columns are presented in the

19

Experimental Section). The extracted ion chromatograms of DMED-labeled FAs

20

(C2C22) were shown in Figure S1, and their retention behavior also followed the

21

carbon number rule. These results were used to create linear calibration curves (log10RT

22

versus log10C) for the four HILIC columns (Table S3). The coefficients a and b of these 9



Page 10 of 28

1

linear equations obtained on the columns with different ligands are different probably

2

due to different retention mechanisms; however, the coefficients a and b were almost

3

identical on the columns with the same ligand but from different batches such as

4

Acquity

5

03003634818313/03493714518642), suggesting that the HILIC RIs of metabolites can

6

be shared between different batches of columns with same ligands (Figure S2).

7

Reproducibility of RTs and RIs

UPLC®

BEH

HILIC

columns

(serial

number

8

To evaluate the reproducibility of HILIC RI, 28 DMED-labeled carboxyl

9

standards (Table S2) were chosen as analytes, and DMED-labeled FAs were used as

10

calibrants. A rat serum extract spiked with these analytes and calibrants was used as a

11

test sample. The sample was analyzed by HILIC-ESI-orbitrap MS within three-days

12

under the same conditions (n=5, three times on the first day, and once daily for the next

13

two days). The calculation equations of HILIC RI over five runs were obtained

14

according to the above-established strategy, and the RIs of the 28 DMED-labeled

15

carboxyl compounds were obtained by calculating their RTs at each run. Relative

16

standard deviations (RSDs) under these five runs for RTs and RIs were compared

17

(Table S4). The RTs shifts of 28 DMED-labeled carboxyl compounds ranged from

18

±0.13 to ±0.49 min with RSDs ranging from 0.44% to 2.35%; deviations of RIs were

19

less than 0.8 with the RSDs ranging from 0.13% to 0.71%, indicating that the RIs

20

showed better reproducibility. In addition, it was worth of noting that the calibrants

21

were spaced very unequally over the retention time window and C10-C22 were eluted

22

closely as shown in Figure 1; The slighter measurement error in RT will lead to a larger 10


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


1

error in RI (Figure S3). However, the RI error was found to be less than one unit of RI

2

(RSD

Method to Calculate the Retention Index in Hydrophilic Interaction

Recommend Documents