S-1

Supporting Information

A Lipidomic Workflow Capable of Resolving sn- and C=C Location

Isomers of Phosphatidylcholines

Xue Zhao1, Wenpeng Zhang2, Doinghui Zhang3, Xinwei Liu3, Wenbo Cao3, Qinhua Chen4,

Zheng Ouyang3, Yu Xia1*

1MOE Key Laboratory of Bioorganic Phosphorus Chemistry & Chemical Biological, Department of Chemistry, Tsinghua University, Beijing 100084, China 2Departement of Chemistry, Purdue University, West Lafayette, IN 47907, USA 3State key laboratory of Precision Measurement Technology and Instruments, Department of Precision Instrument, Tsinghua University, Beijing, 100084, China 4Affiliated Dongfeng Hospital, Hubei University of Medicine, Shiyan, Hubei Province 442000, China

To whom correspondence should be addressed:

Prof. Yu Xia, xiayu@mail.tsinghua.edu.cn

Electronic Supplementary Material (ESI) for Chemical Science.This journal is © The Royal Society of Chemistry 2019

S-2

Experimental Section

Lipid nomenclature

Shorthand notation suggested by LIPID MAPS was used for structural annotation of

identified lipids.1 For example, PC 16:0/18:1(9Z) signifies the fatty acyl chain on sn-1

position contains 16 carbons and the fatty acyl chain on sn-2 position contains 18 carbons.

The “0” and “1” after the carbon number refers to the degree of unsaturation of the fatty acyl

chain. The location of C=C is defined by counting from the alpha carbon of the fatty acyl

ester, Z and E is used for annotating C=C bond geometry. “/” means the sn-1/sn-2 position is

determined, while “_” means the position is unknown.

Materials

Lipid standards and the bovine liver polar extract were purchased from Avanti Polar Lipids,

Inc. (Alabaster, AL, USA). Ammonium salts of bicarbonate and formate were purchased from

Aladdin (Shanghai, China). HPLC grade acetone and water were purchased from Fisher

Scientific Company (Ottawa, ON, Canada). Phospholipase A2 (PLA2) from porcine pancreas

was purchased from Sigma-Aldrich (St. Louis, MO, USA).

Human breast cancer tissue samples were supplied by the specimen bank of Dongfeng

Hospital of Hubei University of Medicine. The procedures were complied with ethical

regulations of the Ethical Review Board of Tsinghua University (IRB No. 2017007). Informed

consent was obtained from all participants. The lipid extraction procedure from tissue samples

was followed Folch method.2 Tissue sample (70 mg) was added with 1 mL of deionized water

in a 10 mL-centrifuge tube. It was homogenized by a handheld homogenizer (Jingxin

S-3

Technology, Beijing, China) at 40,000 Hz for 5 min. After homogenization, the lipid was

extracted with 1-mL methanol and 2 mL chloroform. After 5 min vortex, the mixture was

centrifuged at 11, 269 × g for 8 min (Eppendorf, Shanghai, China). The top layer was used

for repeated extraction and the bottom layer was collected and transferred to a 10 mL-glass

test tube. The chloroform layers from the two extractions were combined and dried under

nitrogen flow. The extract was reconstituted into 1-mL methanol and stored at -20 ℃ before

analysis.

PLA2 Digestion

We followed the PLA2 digestion protocol from Sigma-Aldrich3 with some modification.

PLA2 powder was dissolved in 1:1 water-glycerol (v/v) containing 75 mM NaCl and 10 mM

Tris-HCl buffer (pH = 8.0) to produce a PLA2 concentration of 0.89 U μL-1. Five theoretical

mole percentages mixtures of sn- isomers of PC were mixed in methanol while keeping the

total concentration at 10 μM. These mixtures were prepared from the same stock solutions as

those used for direct MS analyses. Each sample was reconstituted with 150 μL of the PLA2

solution and 10 μL of aqueous 100 mM CaCl2 solution. Resulting samples were incubated at

37 ℃ for 7 min, followed by dilution of 50 μL of each sample with 5 mM ammonium acetate

in methanol to a final volume of 1 mL.

Mass spectrometry

For method development, experiments were performed on a triple quadrupole/linear ion trap

(LIT) hybrid mass spectrometer (4500 QTRAP, Sciex, Toronto, Canada), employing

nanoelectrospray ionization (nanoESI). Neutral loss and precursor ion scans were performed

S-4

in triple quadrupole mode. Accurate mass measurement data were collected on a quadrupole

time-of flight (Q-TOF) mass spectrometer (X500R, Sciex, Toronto, Canada), which was

equipped with an ESI source.

LC-PB-MS/MS

The LC-PB-MS/MS system consisted of an ExionLC AC system (Sciex, Toronto, Canada), a

QTRAP 4500 mass spectrometer (Sciex, Toronto, Canada) and a home-built flow

microreactor. A flow microreactor made from FEP tubing (0.03-in. i.d., 1/16 o.d.) and a low-

pressure mercury lamp (emission at 254 nm wavelength) was used for the post-column PB

derivation. The LC was equipped with a degasser, two pumps, an automatic sampler, and a

column oven. Separation was performed on a HILIC column (150 mm × 2.1mm, silica spheres,

2.7 μm) from Sigma-Aldrich (St. Louis, MO, USA). The column temperature was set at 30 ℃.

Mobile phase consisted of A: acetonitrile (ACN) /Acetone (50:50, v/v) and B: NH4HCO3

aqueous solution (20 mM). A linear gradient was used: 0-8 min: 90% to 85% B; 8-15 min:

85% to 80% B; 15-17 min: 80% to 70% B; 17-23 min: 70% to 70% B; 23-24 min: 70% to

90% B, followed by washing with 90% B at a flow rate of 0.2 mL/min. The injection volume

was 2 μL.

Offline PB reaction

For determining C=C in sn-1 fatty acyls of PCs, PB-MS4 CID was applied to the bicarbonate

adduct of PC. The PB reaction solution was collected from an offline flow microreactor.4 The

flow path was made from fused silica capillaries (363 μm o.d. 100 μm i.d.; Polymicro

Technologies/Molex; Phoenix, AZ, USA) with UV-transparent fluoropolymer coating. A

S-5

low-pressure mercury (LP Hg) UV lamp (254 nm, BHK, Inc.; Ontario, CA, USA) was utilized

to initiate the PB reaction.

OzID on a home-built linear ion trap mass spectrometer

A home-built dual-trap mass spectrometer5 was employed for performing OzID experiments.

First the precursor ion (m/z 820) was trapped in linear ion trap 1 (LIT1) and then mass-

selectively transferred to LIT2. This fragment ion (m/z 419) was formed through beam-type

CID by increasing the potential differences between LIT DC float voltages, which was

followed by isolation in LIT2. Ozone was produced from dielectric barrier discharge of O2

using a home-built ozone generator and injected into LIT2 via a pinch valve for 500 ms.

Two fragment ions at m/z 153 and 169 were observed, proving the existence of C2-C3 double

bond on glycerol backbone.

Hypothesis testing for calibration curves generated from NLS 121 Da of [M+HCO3]-

Hypothesis testing was used to determine whether there were significant differences between

the two calibration curves (Fig. 2e) of the sn-isomers (PC 16:0/18:1 vs. PC 18:1/16:0). F-test

(one-tailed) was first employed to compare the covariances of the two regression lines, 𝑆"#.

T-test (two-tailed) was then employed to compare the slopes (m) of two calibration curves.

F-test for covariance of regression

Covariance of regression, 𝑆"#= ∑ [&'() *'+(-./0')]

3

4+#

For calibration curve of PC 16:0/18:1, 𝑆"5# =6.98e-3

For calibration curve of PC 18:1/16:0, 𝑆"## =1.36e-3

Then, F = 78)3

7833= 5.12

S-6

Fcrit at 95% confidence level: Fcrit (0.05, 4, 4) = 6.39

F < Fcrit

Therefore, the two calibration curves did not show significant difference in covariance.

T-test for slope

Variance of the slope, 𝑆/# = 783

799 (𝑆00=∑𝑥;#-(∑ 0')

3

4)

For calibration curve of PC 16:0/18:1, 𝑆/5# =8.21e-4

For calibration curve of PC 18:1/16:0, 𝑆/## =1.60 e-4

𝑆<==>?@# =𝑆/5# +𝑆/## 𝑆<==>?@=0.031

t = ∆/7BCCDEF

= 0.40

t crit at 95% confidence level (two-tailed): t crit (0.05, 3) = 3.18

t < tcrit

Therefore, there was no significant difference between the two slopes.

The two test results suggest that the two calibration curves of sn-isomers did not show

significant difference at 95% confidence level.

S-7

Scheme S1. The possible mechanism for neutral loss of acraldehyde (NL_56 Da) from MS3

CID of the sn-1 fragment.

Scheme S2. The “zwitterion intermediate” pathway for the formation of ethyl radical ion

MS3

NL_56Da

-H2CO3

S-8

Figure S1. Negative ion mode MS2 CID of formate adduct of 10 μM PC PC 16:0/18:1 (m/z

804.2, CE=30 eV).

Figure S2. Accurate mass measurement of the sn-1 fragment (m/z 419). The MS2 CID

spectrum of bicarbonate adduct of PC 16:0/18:1 (m/z 820.2, CE=30 eV) was collected on a

Q-TOF instrument.

Rel

. Int

. (%

)

-CH3COOH

m/z

0

50

100744.1

281.3

255.2 480.1 804.2

804

[M+HCOO]-NLS_ketene 18:1

100 500 900300 700

506.1

255.2322

281.2477419.2554

673.4792

699.4945

758.5680

100 300 500 700 9000

50

100C21H40O6P

Rel

. Int

. (%

)

m/z

820

S-9

Figure S3. Negative ion mode MS3 CID of (a) m/z 758.4, (b) m/z 673.3, (c) m/z 699.5, (d)

m/z 700.6 from the MS2 CID of the bicarbonate adduct of 10 μM PC 16:0/18:1.

Figure S4. (a) Ion/molecule reactions between the ethyl radical ion and molecular oxygen.

Ethyl radical ion at m/z 728.5 was generated from MS2 CID of PC 18:0/18:1 ([M +HCO3]-,

m/z 848.5), while oxygen was trace residual in the vacuum system due to atmospheric

sampling. Zoomed-in mass spectra after the ethyl radical ion (m/z 728.5) being trapped in Q3

LIT for (b) 2 ms and (c) 200 ms. The peak at m/z 760.4 suggests O2 addition to the ethyl

radical ion.

100 500 900m/z

281.3

758.4255.2

494.30

50

100R

el. I

nt. (

%)

820758

a

281.4

700.6255.4

435.2417.4 443.3

100 500 9000

50

100

Rel

. Int

. (%

)

m/z

820699

c100 500 900

m/z

255.2673.3

409.1391.2

281.3417.1435.1153.0

0

50

100

Rel

. Int

. (%

)

820673

b

281.4

255.4

419.3437.3363.4

100 500 9000

50

100

Rel

. Int

. (%

)

820

700699.5700.6

699.6AF

m/z

716 732 748 764m/z

x 6.0

728.5760.4

0.0

50

100

Rel

. Int

. (%

)

716 732 748 764m/z

x 6.0

728.5

0.0

50

100

Rel

. Int

. (%

)

Trapped time: 2 ms848728

Trapped time: 200 ms848728

O2a

b

c

S-10

Figure S5. Negative ion mode MS2 CID of bicarbonate adduct of 10 μM D9-labeled PC

16:0/16:0 (m/z 803.7, CE=30 eV), where all nine hydrogen atoms in trimethylamine group

were labeled with deuterium atoms.

Figure S6. (a) The break down curve derived from MS2 beam-type CID of bicarbonate adduct

of PC 16:0/18:1 (m/z 820). (b) MS2 ion-trap CID of bicarbonate adduct of PC 16:0/18:1 (m/z

820, AF2 = 0.1, 25 ms).

200 400 600 800

673.5552

419.3055

739.6784674.5619

740.6850647.5368255.2653 722.6405

721.6341420.3114 803.69690

50

100

419.3055

420.3114

673.5552

674.5619647.5368

675.5681

739.6784

740.6850722.6405

721.6341

-AH+CD2H(CD3)2N

-D2CO3

-HDCO3

-AH+(CD3)3N

-AH+CDH2(CD3)2N-A+(CD3)3NCHCH2

Rel

. Int

. (%

)

803

m/z

300 500 900

699.2281.4673.4 758.4

255.4

820.30

50

100

100 700

Rel

. Int

. (%

)

m/z10 20 30 40 50 60

m/z 820 m/z 699 m/z 419

CE energy /eV

Inte

nsity

/%

0

50

100a b 820

S-11

Figure S7. Negative ion mode MS2 CID of bicarbonate adduct of (a) 10 μM PC 16:0/18:0

(m/z 822.2, CE = 30 eV), (b) 10 μM PC 16:0/18:1 (m/z 820.2, CE = 30 eV), and (c) 10 μM

PC 18:0/20:4 (m/z 870.2, CE = 30 eV).

Figure S8. Correlations of sn-isomer quantitation by PLA2 hydrolysis and sn-1 fragment

method: (a) PC 16:0_18:1, (b) PC 16:0_18:0, and (c) PC 18:0_18:1.

100 500 900m/z

749.6808.7

723.6793.6

303.4283.4 447.4317.4 670.6508.3

0

50

100

Rel

. Int

. (%

)

PC 18:0/20:4 749.6750.6

100 500 900m/z

701.3

419.0 760.3

675.2283.2255.1 447.1 822.2

0

50

100

Rel

. Int

. (%

)

701.3PC 16:0/18:0a

b

100 500 900m/z

699.2

758.3419.2

281.3673.2255.2

363.1 445.0 820.20

50

100

Rel

. Int

. (%

)

PC 16:0/18:1699.2

700.3

c

822

820

870

y = 1.0118x - 0.0033R² = 0.9999

0 0.5 1

% P

C 1

8:1/

18:0

from

sn

-1 fr

agm

ent

% P

C 1

8:1/

16:0

from

sn

-1 fr

agm

ent

y = 1.0323x - 1.0618R² = 0.9997

0 50 100% PC 18:1/16:0 from PLA2 digest

0

50

100 y = 0.9835x + 0.0422R² = 0.9999

0 0.2 0.4 0.6 0.8 1

% P

C 1

8:0/

16:0

from

sn

-1 fr

agm

ent

0

50

100

0

50

100

% PC 18:0/16:0 from PLA2 digest

ba c

% PC 18:1/18:0 from PLA2 digest

S-12

Figure S9. Limit of detection (LOD) for PC 16:0/18:1 based on signal to noise ratio higher

than 3 using different MS/MS methods. (a) LOD =10 pM from NLS 121 Da of [M +HCO3]-,

(b) LOD =1.0 pM from PIS m/z 184 of [M +H]+, and (c) LOD = 5 nM from NLS 60 Da of

[M +HCO2]-.

Figure S10. (a) NLS 121 Da of [M +HCO3]- derived from an equal molar mixture of PC

15:0/15:0, PC 16:0/18:0, and PC 18:0/20:4 (5 μM each, CE=30 eV). (b) Calibration curve for

PC 18:0/20:4 using NLS 121 Da with IS (PC 15:0/15:0) kept at 0.5 μM.

CIDNLS 121 Daa

818.0 820.0 822.0m/z

0

10.0

19.0

Inte

nsity

, cps

820.610 pM

NLS 121 Da[M+HCO3]-

802.0 804.0 806.0m/z

0

2.0

4.0

5.8804.6

805.4

NLS 60 Da5 nM

[M+HCOO]-

Inte

nsity

, cps

c CIDNLS 60 Da

755.5 758.0 761.0 764.0m/z

0100200300400

760.9

761.4762.6

Inte

nsity

, cps

[M+H]+ PIS m/z 1841.0 pM

bm/z 184

700 750 800 850 900m/z

0.0

4.0e4

8.0e4

Inte

nsity

, cp

s

766.8

822.8 870.8

PC 15:0/15:0

PC 16:0/18:0 PC 18:0/20:4

y = 0.6503x - 0.0972R² = 0.9799

0

1

2

0 1 2 3Concentration/ (μM)

Ratio

(I/I IS

)

a b

S-13

Figure S11. Comparisons of the PB reaction kinetic curves of PC 18:0/18:1 (5 μM) prepared

in 1:1 acetone: H2O with (a) 5 mM NH4HCO3 added, and (b) without NH4HCO3 added. (c)

The PB reaction kinetic curve of PC 18:1/18:0 (5 μM) prepared in 1:1 acetone: H2O.

Figure S12. Comparisons of negative ion mode LC-MS from polar lipid extract of bovine

liver with (a) 20 mM NH4HCO3 and (b) 20 mM NH4HCO2 added in the mobile phase as

buffer, while all other parameters were kept the same. Retention of different classes of lipids

was comparable between the two buffer conditions. (c) MS spectrum of a low abundance PC

species, PC 31:1, from condition (a) and (d) corresponding extracted ion chromatogram (XIC).

(e) MS spectrum of PC 31:1 from condition (b) and (f) corresponding XIC. Detection of PC

31:1 in the form of [M+HCO3]- (m/z 778) is about 8 times more abundant than in the form of

[M+HCO2]- (m/z 762).

0

10

20

30

40

0 4 8 120

20

40

60

0 4 8 12UV Exposure Time, s

0

10

20

30

40

0 4 8 12UV Exposure Time, s

[PBPC 18:0/18:1+H]+ [PBPC 18:1/18:0+H]+b c[PBPC 18:0/18:1+HCO3]-

UV Exposure Time, s

Yiel

d/%

a

S-14

Figure S13. PIS m/z 184 of [M +H]+ (CE=35 eV) of PCs in bovine liver extract (200 ppm.)

Figure S14. (a) Relative quanitation of PCs at subclass level from normal (N=3) and

cancerous (N=3) human breast tissue. (b) Changes of sn-isomeric ratio of PC 16:0_16:1 from

normal (N=3) and cancerous (N=3) human breast tissue. (c) Relative changes of ∆9/∆11

ratios of C18:1 in PCs from normal (N=3) and cancerous (N=3) human breast tissue.

Differences between normal and cancerous samples were evaluated for statistical significance

700 750 800 850 900

786.8

812.8760.8

838.7

PC 34:1

PC 36:2

PC 38:3

PC 40:4

0

50

100

Rel

. Int

. (%

)

m/z

732.7PC 32:1

PIS m/z 184 Da

0

4

8

12normal cancer

***

***

** *

a

Rat

io(I/

I IS)

PC species

0

3

6

9

PC 16:0_18:1

***

Rat

io (I

∆9/I ∆

11)

02468

10

PC 17:0_18:1

**

0

3

6

9

PC 18:1_18:2

**

0

3

6

9

PC 18:1_18:1

***

0

5

10

PC 18:0_18:1

***

0

2

4

6

8

PC 19:0_18:1

**

normal cancerc

PC species

b

0

2

4

6

PC 16:0_16:1Rat

io (I

16:0

/16:

1/I16

:1/1

6:0)

PC species

***

S-15

using the two-tailed student’s t-test (* P<0.05, ** P<0.01, *** P<0.001). Error bar represents

± s.d. (N=3).

S-16

Table S1. Comparison of sn- purities for commercial PC standards by PLA2 digestion and

sn-1 fragment derived from MS2 CID of [M +HCO3]-

Lipids

PLA2 (mol%) sn-1 fragment (mol%)

fragment

%Dev

PC 16:0/18:1 92.5±0.4 93.2±1.4 +1%

PC 18:1/16:0 85.3±0.4 87.2±1.4 +2%

PC 16:0/18:0 87±3 83±5 -5%

PC 18:0/16:0 86±4 89±4 +3%

PC 18:0/18:1 93±4 93±3 0%

PC 18:1/18:0 83±4 84±3 +1%

Table S2. Comparison of tsn-purities for commercial PC standards by PLA2 digestion and

MS3 CID of [M +HCOO]- (all data were taken from publication by Ekroos et al.) 6

Lipids

PLA2 (mol%) Ion trap MS3 CID

(mol%) fragment

%Dev

PC 16:0/18:1 88 83 -5%

PC 18:1/16:0 83 79 -5%

PC 16:0/18:0 88 83 -6%

PC 18:0/16:0 94 93 -1%

PC 18:0/18:1 96 89 -7%

PC 18:1/18:0 81 75 -7%

Reference

1. G. Liebisch, J. A. Vizcaino, H. Kofeler, M. Trotzmuller, W. J. Griffiths, G. Schmitz, F. Spener and M. J. O. Wakelam, J. Lipid Res., 2013, 54, 1523-1530.

2. J. Folch, M. Lees and G. H. S. Stanley, J. Biol. Chem., 1957, 226, 497-509. 3. Sigma-Aldrich, 1993, Enzymatic Assays of Phospholipase A2.

4. E. T. Franklin, S. K. Betancourt, C. E. Randolph, S. A. Mcluckey and Y. Xia, Anal. Bioanal. Chem., 2019, 411, 4739-4749.

5. X. Liu, X. Wang, J. Bu, X. Zhou and Z. Ouyang, Anal. Chem., 2019, 91, 1391-1398.

6. K. Ekroos, C. S. Ejsing, U. Bahr, M. Karas, K. Simon and A. Shevchenko, J. Lipid Res., 2003, 44, 2181-2192.