Integrating MS1 and MS2 Scans in High-Resolution Parallel Reaction

Dec 2, 2016 - Quantification of targeted metabolites, especially trace metabolites and structural isomers, in complex biological materials is an ongoi...
0 downloads 8 Views 2MB Size
Subscriber access provided by UNIV OF GEORGIA

Article

Integrating MS1 and MS2 scans in high-resolution parallel reaction monitoring assays for targeted metabolite quantification and dynamic 13C-labeling metabolism analysis Zhucui Li, Yujing Li, Wujiu Chen, Qichen Cao, Yufeng Guo, Ni Wan, Xiaolong Jiang, Yinjie J. Tang, Qinhong Wang, and Wenqing Shui Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b03947 • Publication Date (Web): 02 Dec 2016 Downloaded from http://pubs.acs.org on December 5, 2016

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 free 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 accessible to all readers and 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.

Analytical Chemistry 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 30

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

Integrating MS1 and MS2 scans in high-resolution parallel reaction monitoring

2

assays for targeted metabolite quantification and dynamic

3

metabolism analysis

4

Zhucui Li1,2#, Yujin Li3#, Wujiu Chen1, Qichen Cao1, Yufeng Guo1, Ni Wan4, Xiaolong

5

Jiang1, Yinjie J. Tang4, Qinhong Wang1, Wenqing Shui2*

6

1

7

300308

8

2

iHuman Institute, ShanghaiTech University, Shanghai 201210, China

9

3

College of Life Sciences, Nankai University, Tianjin 300071, China

10

4

Department of Energy, Environmental and Chemical Engineering, Washington

11

University in St. Louis, St. Louis, MO 63130, USA

13

C-labeling

Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin

12 13

#These authors contribute equally to this work

14 15

*To whom correspondence should be addressed to:

16

Wenqing Shui, iHuman Institute, ShanghaiTech University, Shanghai 201210, China;

17

Tel: 86-21-20685595; email: [email protected]

18

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

Abstract

2

Quantification of targeted metabolites, especially trace metabolites and structural

3

isomers, in complex biological materials is an ongoing challenge for metabolomics.

4

Initially developed for proteomic analysis, the parallel reaction monitoring (PRM)

5

technique exploiting high-resolution MS2 fragment ion data has shown high promise

6

for targeted metabolite quantification. Notably, MS1 ion intensity data acquired

7

independently as part of each PRM scan cycle are often underutilized in the PRM

8

assay. In this study, we developed an MS1/MS2-combined PRM workflow for

9

quantification of central carbon metabolism intermediates, amino acids and shikimate

10

pathway-related metabolites on an orthogonal QqTOF system. Concentration curve

11

assessment revealed that exploiting both MS1 and MS2 scans in PRM analysis

12

afforded higher sensitivity, wider dynamic range and better reproducibility than

13

relying on either scan mode for quantification. Furthermore, Skyline was incorporated

14

into our workflow to process the MS1/MS2 ion intensity data, and eliminate noisy

15

signals and transitions with interferences. This integrated MS1/MS2 PRM approach

16

was applied to targeted metabolite quantification in engineered E. coli strains for

17

understanding of metabolic pathway modulation. In addition, this new approach,

18

when first implemented in a dynamic

19

advantage in capturing and correcting isotopomer labeling curves to facilitate

20

nonstationary 13C-labeling metabolism analysis.

13

C-labeling experiment, showed its unique

21 22 2

ACS Paragon Plus Environment

Page 2 of 30

Page 3 of 30

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

Introduction

2

Mass spectrometry (MS) coupled to different separation techniques such as liquid

3

chromatography (LC), gas chromatography (GC) and capillary electrophoresis (CE)

4

has become the leading platform for large-scale identification and quantitation of

5

metabolites in various biological systems.1-3 Non-targeted MS-based metabolomic

6

surveys mainly employ the high-resolution full scanning instruments to profile

7

hundreds of endogenous and drug metabolites.4-6 In these studies, metabolite

8

identification is typically ensured by MS/MS analysis of isolated precursors using a

9

data-dependent acquisition (DDA) strategy.1 However, metabolite quantification

10

relying on MS1 scan data could suffer from low reproducibility due to inadequate

11

sensitivity of MS1 scans for metabolites that are less abundant, prone to matrix

12

interference or have isomeric compounds.3 Very recently, a data-independent

13

acquisition (DIA) workflow using SWATH-MS technique initially developed for

14

proteomic quantification has been adapted to global metabolomics to allow for

15

sensitive, reproducible and high-throughput metabolite quantification based on

16

fragment ion intensities in MS2 scans7.

17

MS2-based quantification of small molecules has been primarily implemented in

18

pharmaceutical labs for specified drug metabolite analysis.8-10 Multiple reaction

19

monitoring (MRM) performed on triple quadrupole instruments is recognized as the

20

“gold standard” MS technique for targeted bioanalyte quantification due to its high

21

sensitivity and robustness.8-10 The MRM technique has been recently adopted in

22

metabolomic studies focusing on a defined subset of metabolites representative of key 3

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

pathways.11-14 This type of targeted metabolomics is a valuable tool for assessing

2

changes of metabolic pathways or networks resulting from genetic mutation, altered

3

gene expression, protein dysfunction or environmental perturbation. For example,

4

Wei and colleagues developed an LC/MS/MRM platform to monitor 205 endogenous

5

metabolites and applied it to targeted metabolite analysis in plasma samples.11

6

Nevertheless, MRM assays performed on low-resolution mass spectrometers

7

sometimes are limited by ambiguous metabolite identifications as a result of

8

interfering peaks from complex matrix that resemble the compound of interest under a

9

specific Q1/Q3 transition.15

10

Technological improvements in high-resolution MS have led to new acquisition

11

methods which have challenged the dominance of MRM for targeted MS.16 One

12

approach, parallel reaction monitoring (PRM), has emerged as a capable alternative to

13

MRM. In a PRM assay, the precursor ion is isolated in Q1 and fragmented in Q2, and

14

all generated fragment ions are monitored in MS2 scans on a high resolution, accurate

15

mass spectrometer.17 Then fragment ion intensities from MS2 spectra are used for

16

specific precursor quantification. The PRM approach has been successfully

17

implemented in a panel of targeted proteomic studies and demonstrated superior

18

specificity, accuracy and multiplexing capability compared to the MRM assay.18-21 In

19

contrast, there has been rare report of PRM application to targeted metabolomic

20

analysis except the very latest research from Zhou et al22 and Qiu et al23. Both groups

21

developed PRM assays on a hydrid Q-Exactive instrument (Thermo) for measurement

22

of targeted metabolites. Zhou et al reported monitoring 237 endogenous metabolites 4

ACS Paragon Plus Environment

Page 4 of 30

Page 5 of 30

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

for which structural identification solely relied on MS/MS spectra available in public

2

databases.22 Qiu et al constructed PRM assays using chemical standards for

3

quantification of 25 intracellular primary metabolites.23 It is worth noting that

4

PRM-based quantification is carried out using ion intensities of MS2 fragments, yet

5

the role of MS1 scans have largely been ignored, despite the fact that the respective

6

ion signals of certain precursors could be stronger than those of fragments. Systematic

7

evaluation of MS1 vs MS2 quantification in PRM assays has not been reported for

8

either metabolite or peptide analysis.

9

Our study, for the first time, has developed an MS1/MS2-combined PRM 13

10

workflow for targeted metabolite quantification as well as

11

analysis on an orthogonal QqTOF instrument (TripleTOF 6600). Sixty-five targeted

12

metabolites include central carbon metabolism (CCM) and shikimate pathway

13

intermediates, and free amino acids. After comparing the sensitivity, reproducibility

14

and linearity of metabolite quantification based on MS1 vs MS2 ion chromatograms

15

derived from PRM cycles, we concluded that exploiting both MS1 and MS2 scans in

16

PRM assays achieves the best quantitative performance. Furthermore, Skyline can be

17

incorporated into our workflow to process the MS1/MS2 ion intensity data, and

18

eliminate noisy signals and transitions with interferences.24 The integrated MS1/MS2

19

PRM approach was applied to targeted metabolite quantification in engineered E. coli

20

strains for understanding of metabolic pathway modulation. In addition, this approach

21

was first implemented in a dynamic 13C-labeling experiment to reveal its capability of

22

accurate measurement of authentic isotopomers for tracing cell metabolism.

23

5

ACS Paragon Plus Environment

C-labeling metabolism

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

Experimental Section

2

Chemicals, details in sample preparation, and data processing can be found in

3

Supporting Information.

4

Metabolite Extraction from E. coli Cells. A wild-type E. coli strain AB2834 (WT)

5

and two engineered strains (TIB01 and TIB02) with improved production of

6

3-dehydroshikimic acid were grown in LB media aerobically at 30 °C, 250 rpm to

7

produce the seed culture. The seed cultures were then transferred into a 250 ml flask

8

containing the NBS minimal complete media and grew under 37 °C, 250 rpm till

9

OD600 reached 2.0. Bacteria cultures of 5~6 ml were quenched by mixing with the

10

same volume of frozen saline buffer and harvested through centrifugation at 4 °C for

11

3 min at 3500 g. The cell pellet was washed twice with PBS and extracted using a

12

pre-chilled solvent mixture of methanol: acetonitrile: water (4:4:2, v/v) with 0.1% FA

13

which contains d4-succinic acid as the internal standard. The metabolite extracts were

14

snap frozen in liquid nitrogen, thawed, and then sonicated in water bath. The

15

freeze-thaw-sonication cycle was repeated twice. Then the samples were incubated at

16

-20 °C for 2 h. After centrifugation at 16000 g for 20 min at 4 °C, the supernatant was

17

harvested and dried out by speed vacuum and stored in -80 °C. Before HILIC-MS

18

analysis, the metabolite extracts were reconstituted into the acetonitrile/water (1:1; v/v)

19

solvent. Each strain was prepared in biological duplicate and metabolite extraction

20

was processed in duplicate for each biological replicate.

21

HILIC-QqTOF Analysis. Metabolite standards or total cell extracts were

22

analyzed on a Nexera HPLC system (SHIMADZU, Japan) connected to TripleTOFTM 6

ACS Paragon Plus Environment

Page 6 of 30

Page 7 of 30

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

6600 mass spectrometer (AB SCIEX, USA). LC separation was carried out on a

2

ZIC-HILIC column (100 mm × 2.1 mm, 3.5 µm, Merck, Germany). The mobile phase

3

A was 10 mM ammonium acetate in water, and mobile phase B was acetonitrile. The

4

linear gradient used was as follows: 0-3 min, 90% B; 3-18 min, 90%-60% B; 18-25

5

min, 60%-55% B; 25-27 min, 55%-50% B; 27-30 min, 50% B; and the column was

6

re-equilibrated for 10 min. The flow rate was 0.2 ml/min and the sample injection

7

volume was 5 µl. Each metabolite standard was also injected into the mass

8

spectrometer through direct infusion to acquire reference MS/MS spectra. Mass

9

spectra were acquired in negative ion ESI mode within a mass range from 100 to 1000

10

m/z. The ion source parameters are ion spray voltage, -4500 V; curtain gas, 35 psi;

11

nebulizer gas, 55 psi; heater gas, 55 psi; source temperature, 550 °C. Rolling CE and

12

fixed CE were assessed using standards to acquire optimal CEs for individual

13

metabolite targets.

14

MS data acquisition for PRM analysis was programmed in the product ion mode,

15

and consisted of one MS1 scan (150 ms) with the mass range from 50 to 900 m/z,

16

followed by targeted MS2 scans (50 ms) under the optimal CE condition across the

17

mass range from 30 to X (X = precursor mass + 10 Da) m/z. The maximal cycle time

18

was 3.25 sec for a total of 65 targets. In the dynamic

19

cell extracts sampled at different time points were also analyzed by HILIC-Q/TOF in

20

the PRM mode. For certain metabolites having suspected interferences, targeted MS2

21

scans were performed on specific isotopomers of interest.

13

C-labeling experiment, total

22 23

7

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

Results and Discussion

2

Establishing PRM assays of 61 metabolites on TripleTOF 6600 mass

3

spectrometer

4

To develop PRM assays for quantification of 61 intracellular metabolites including 44

5

central carbon metabolism (CCM) intermediates and 17 amino acids (full name and

6

abbreviation in Table S1), we first analyzed a mixture of the synthetic metabolite

7

standards on the TripleTOF 6600 mass spectrometer in PRM mode. Each metabolite

8

exceeding the MS1 intensity threshold was subjected to MS/MS fragmentation under

9

default sweeping collision energy (-35±15 eV). However, we noticed that the

10

MS/MS spectra generated under default CE on certain metabolites were of low quality

11

as a result of either insufficient or excessive fragmentation of the precursor ions

12

(Figure S1A). This is different from peptide PRM assays in which the majority of

13

peptides of diverse sequences are fragmented effectively under theoretically predicted

14

CE based on the precursor m/z and charge state.17, 21 To improve MS/MS spectral

15

quality, we manually optimized CE for each metabolite standard in a way similar to

16

MRM assays on small molecules. For both NADPH and acetyl CoA, an increased CE

17

relative to the default setting gave rise to a greater number of fragment ions as well as

18

diminished precursor peaks (Figure S1B). For another metabolite TPP, two distinct

19

fragment ions at m/z 176.938 and 301.970 of 10-fold higher intensity than the default

20

condition were generated from the precursor subject to a lower CE (Figure S1B). The

21

optimal CE values ranged from -15 to -60 eV for specific metabolites (Table S2).

22

We then examined the optimized MS/MS spectra to select one to six fragments 8

ACS Paragon Plus Environment

Page 8 of 30

Page 9 of 30

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

of relatively high abundance and large m/z for target quantification. Skyline software

2

was used here to automatically extract MS2 ion chromatograms after we defined

3

specific transitions for each metabolite precursor. As MS1 full scan was also

4

performed in each cycle of the PRM assay, quantitative data were extracted for a

5

given precursor ion using the same “Transition List” module in Skyline. Therefore we

6

were able to extract both MS1 and MS2 transitions in a simultaneous data-processing

7

workflow with Skyline.

8

The reproducibility of MS2-based quantification of 45 tested metabolites was

9

significantly impacted by CE conditions. With optimized CE, fragment peak areas for

10

38 metabolites showed CVs < 20% in triplicate measurements whereas the number of

11

metabolites with fragment peak area CVs < 20% dropped to 22 under the original CE

12

condition (Figure S1C). It is noteworthy that the variation of CVs has little to do with

13

the data sampling points over extracted ion chromatograms. In fact, a minimum of 18

14

data points were acquired for all targeted metabolites considering the wide HILIC

15

peak width (1-2 min) and the maximal cycling time including all MS2 acquisitions

16

(~3.25 sec). Therefore, we would recommend optimization of CE settings for

17

individual metabolites to ensure sensitivity and reproducibility of targeted PRM

18

analysis, at least on a QqTOF instrument, even though rolling CE or constant CE is

19

typically adopted in data-dependent acquisition for non-targeted metabolomics

20

studies.25, 26

21 22

Comparison of MS1 and MS2 quantification in PRM using concentration curves 9

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

To compare the quantitative performance of MS1 and MS2 chromatogram extraction

2

from PRM acquisitions, we carried out a series of dilution experiments in both simple

3

and complex matrix. A mixture of 15 metabolite standards were diluted in simplex

4

matrix (50% MeOH) spanning a concentration range from 5 pg to 100 ng on column.

5

The MS1 peak area of the precursor ion and sum of MS2 peak areas of selected

6

fragment ions provided independent quantitative measure of the targeted metabolite at

7

MS1 and MS2 levels. The sensitivity and linear response range of metabolite

8

quantification in simple matrix was summarized in Table S3, where 9 out of 15

9

metabolites showed a lower LOQ when quantified by MS2 than by MS1. For instance,

10

the LOQ of FAD quantification by extracting six MS2 transitions was 5 pg yet the

11

LOQ was 10-fold higher (50 pg) for MS1 chromatogram extraction (Figure 1A). At

12

low concentrations, the MS1 precursor ion chromatogram of FAD was more prone to

13

ion suppression and interferences, leading to increased variability and reduced

14

sensitivity. In contrast, the MS2 fragment ions of FAD appeared to be more robust and

15

selective, thus offering a lower LOQ and a wider dynamic range (Figure 1A). Among

16

the tested metabolites in simple matrix, five showed similar LOQ levels and linear

17

response range between MS1 and MS2 quantification in PRM assays (Table S3). For

18

example, the precursor and fragment ion chromatograms of L-malic acid (Mal)

19

displayed comparable robustness in the low concentration range (Figure 1B). Notably,

20

for 5-ALA, MS1 precursor quantification enabled much better sensitivity than MS2

21

fragment quantification. The LOQ of 5-ALA quantification with MS2 scans was

22

20-fold higher than that with MS1 scans, which was attributed to very weak signals of 10

ACS Paragon Plus Environment

Page 10 of 30

Page 11 of 30

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

all detected fragment ions of this amino acid (Figure 1C).

2

PRM quantification of all 61 targeted metabolites was also evaluated in complex

3

matrix by preparing up to 210 fold dilution of an E. coli total extract in extraction

4

solvent. Forty-one metabolites of our interest showed better sensitivity, consistency

5

and wider dynamic range in MS2 vs MS1 quantification (full data in Table S4). In a

6

representative case of D-ribose-5-phosphate (R5P), its MS1 precursor ion response

7

plateaued at the low end of the concentration curve whereas the MS2 ion

8

chromatograms encountered little interference and retained strong signals in the high

9

dilution-fold region (Figure 2A). However, for other metabolites such as Val, very

10

poor fragment ion chromatograms were observed with increase of the dilution fold,

11

thus rendering MS2 quantification ineffective (Figure 2B). It is of our notion that

12

metabolites of larger molecular weight including most sugar phosphates, organic acids

13

and energy/redox cofactors that tend to generate a greater number of fragments of

14

strong intensity are more amenable to MS2 quantification. In contrast, smaller

15

metabolites especially amino acids producing fewer and weaker fragments are better

16

to be quantified in MS1 scans. Depending on the quantitative reproducibility,

17

sensitivity and linear response range of the total cell extract sample, we classified 50

18

metabolites to be MS2 quantifiable and 11 to be MS1 quantifiable targets (Table S4).

19

When each metabolite was measured in its optimal mode (MS1 vs MS2), 89% of

20

all targeted metabolites diluted in complex matrix had a linear regression coefficient

21

(R2) of the concentration curves above 0.995, and 95% of metabolites had r2 above

22

0.990 across 2-3 orders of magnitude (Figure 2C, Table S4). Moreover, the median 11

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

CV of metabolite quantification in triplicate measurements was below 20% over the

2

entire concentration range down to 210-fold dilution of the original cell extract (Figure

3

2D). These results indicated that excellent linearity, reproducibility and sensitivity of

4

the PRM assay can be achieved by combining MS1 and MS2 scans.

5 6

Quantification of isomeric metabolites by PRM MS2 scans

7

MS2 scans are particularly useful for differentiation of co-eluting structural isomers

8

of identical or very similar molecular weight.27 Within our target list, two pairs of

9

isomeric metabolites (citrate/isocitrate and leucine/isoleucine) that completely or in

10

most part co-elute in HILIC separation, were indistinguishable based on their MS1

11

precursor chromatograms. However, these isomers could be differentiated with the

12

isomer-specific fragment ions. For instance, two ions at m/z 72.995 and 117.019 are

13

unique fragments derived from isocitrate whereas the abundant ion at m/z 87.009

14

specifically originates from citrate (Figure 3A). In this way the unique transitions

15

allowed for differentiation and quantification of this isomeric pairs in complex matrix

16

even though their MS1 chromatograms were overlapped (Figure 3B). Likewise, Leu

17

and Ile both having their unique fragment ions were quantified with distinct MS

18

chromatograms of specific isomers (Figure 3C, 3D)

19

With PRM MS2 scans, strong linear response of these metabolite isomers were

20

obtained in both simple and complex matrix (R2 > 0.98 for all four metabolites, see

21

detailed data in Table S3 and S4). Notably, it is difficult to differentiate these two

22

pairs of isomeric metabolites in traditional MRM assays11 because MRM typically 12

ACS Paragon Plus Environment

Page 12 of 30

Page 13 of 30

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

selects the most abundant fragment ion which is common to both isomers in this case.

2

Therefore, PRM enables quantification of specific isomers by selecting less abundant

3

and unique transitions that are of adequate sensitivity and free of interferences from

4

the complex background owing to the high-resolution MS2 scans. The previous study

5

by Zhou J et al also demonstrated the advantage of PRM in increased specificity of

6

metabolite identification and flexibility of post-acquisition assay refinement.22

7 8

Integrated MS1 and MS2 for metabolomic study of engineered E. coli strains

9

To further explore the benefits of integrated MS1 and MS2 scans in PRM assays, we

10

performed targeted metabolite analysis in three different E. coli strains of WT, TIB01

11

and TIB02. Strains TIB01 and TIB02 were genetically engineered from the wild-type

12

(WT) in the shikimate pathway for improved production of 3-dehydroshikimic acid

13

(DHS).28 DHS is a key hydroaromatic intermediate in the biocatalytic conversion of

14

glucose into aromatic bio-products and a variety of industrial chemicals.29 In addition

15

to the CCM intermediates and amino acids, we established PRM methods for

16

measurement of DHS and three related intermediates (DAHP, DHQ and SK, full name

17

in Table S1) in its biosynthetic pathway. Then we conducted PRM analysis of all 65

18

targeted metabolites in the cell extracts from strains WT, TIB01 and TIB02 in four

19

replicates. As MS1 and MS2 ion intensity data were acquired simultaneously in each

20

PRM scan cycle, we extracted the quantification data for each metabolite from

21

separate MS1 and MS2 scans in Skyline. Evaluation of peak area CV distribution for

22

all targeted metabolites revealed that combining MS1 and MS2 quantification 13

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

achieved the best precision (Figure 4A). The median CVs of the combined MS1/MS2

2

PRM assays for all targets were 20.3%, 18.1% and 13.9% for WT, TIB01 and TIB02

3

strains respectively. This result reflects remarkable analytical reproducibility of our

4

approach if considering relatively large variation in cell culture and sample

5

preparation. Thus it is advantageous to exploit MS1 scan data in PRM assays which is

6

often underutilized so as to further enhance robustness and sensitivity for targeted

7

metabolome analysis.

8

The pretreated data matrix from targeted metabolites quantified in three strains

9

was used to develop an unsupervised PCA model (R2=0.914, Q2=0.944). The result

10

exhibited a well-segregated, closely clustered pattern among different groups in the

11

PCA score plot, indicating a specific molecular signature of each strain that was

12

distinguishable between each other (Figure 4B). Replicates of the same strain were all

13

clustered together, again suggesting good reproducibility of our targeted metabolite

14

quantification. Significantly modulated metabolites in the CCM and shikimic acid

15

pathways as well as free amino acids in two engineered stains vs WT were then

16

identified according to dual criteria of VIP score in a supervised OPLS-DA model > 1

17

and p-value in the ANOVA test