Supplementary Information
Multiplexed Separations for Biomarker Discovery in
Metabolomics: Elucidating Adaptive Responses to
Exercise Training
Naomi L. Kuehnbaum,1 Jenna B. Gillen,2 Aleshia Kormendi,1 Karen P. Lam, 1 Alicia
DiBattista, 1 Martin J. Gibala2 and Philip Britz-McKibbin1*
1
Department of Chemistry and Chemical Biology, McMaster University, Hamilton, L8S
4M1, Canada
2
Department of Kinesiology, McMaster University, Hamilton, L8S 4K1, Canada
*
Author correspondence: e-mail: britz@mcmaster.ca; FAX: 1-905-520-9140
Supplemental Information: Supplementary Methods and Materials;
Table S1; Table S2; Fig. S1; Fig. S2; Fig. S3
SUPPLEMENTAL METHODS AND MATERIALS
S1. Chemicals and Reagents.
Ammonium acetate, acetic acid, formic acid, Cl-Tyr, L-carnitine (C0), O-acetyl-Lcarnitine (C2) and hypoxanthine (HyX) were purchased from Sigma-Aldrich (St. Louis,
MO, USA), whereas glutathionyl-L-cysteine mixed disulfide (GSH-Cys-SS) was
obtained from Toronto Research Chemicals (Toronto, ON, Canada).
10 mM stock
solutions of metabolite standards were prepared in water and stored at 4°C. HPLC-grade
acetonitrile (Honeywell, Muskegon, MI, USA) and methanol (Caledon, Georgetown, ON,
Canada) were used for preparation of background electrolyte (BGE) and sheath liquid
respectively. Ammonium acetate was prepared as a 400 mM stock in water and was
adjusted to pH 5.0 using acetic acid. All aqueous buffers and stock solutions were
prepared with deionized water purified using a Thermo Scientific Barnstead EasyPure II
LF ultrapure water system (Cole Parmer, Vernon Hills, IL, USA).
S2. Study Participants.
Nine overweight/obese but otherwise healthy women (age: 21-45 yr, BMI: 25-36 kg m-2)
participated in a 6-week supervised exercise training intervention at McMaster
University. The subjects represent a subset of a group of 16 individuals who took part in
a study that evaluated the impact of pre-exercise nutritional state on adaptations to HIIT
[Gillen et al. Obesity 2013, 21: 2249–2255]. Blood samples at all time intervals were
acquired from only 9 of the 16 subjects from the original trial with exercise
responsiveness being the primary focus of the present investigation. All subjects were
compared against themselves in a repeated measures/cross-over study design in which
pre-exercise nutrition was standardized for a given subject when performing ergometer
cycling trials at the same absolute workload. Subjects were classified as “sedentary”
based on self-reporting of physical activity ≤ 2 sessions/week of exercise with a duration
≤ 30 min. A placebo-controlled arm was not performed in this work due to a large body
of evidence demonstrating the efficacy of HIIT to generate moderate improvements in
baseline cardiorespiratory fitness for sedentary/non-athletic subjects [Weston et al. Sports
Med. 2014, 44: 1005-1017]. Subjects were instructed to maintain a regular diet and
lifestyle under “free-living” conditions throughout the intervention period.
S3. Plasma Collection and Sample Workup.
Venous blood samples were collected via an indwelling catheter inserted in a forearm
vein. Vacutainers (6 mL) with EDTA (10.8 mg) as anticoagulant were used for
collection. Samples were collected prior to onset of exercise (i.e., baseline, 0 min)
immediately after HIIT (i.e., post-exercise, 20 min), and 20 min after HIIT while at rest
(i.e., recovery, 40 min). Blood samples were placed on ice and subsequently centrifuged
at 2,500 g at +4°C for 5 min to fractionate plasma from erythrocytes. Fractionated plasma
was aliquoted and frozen at -80°C until later processing and analysis. Frozen plasma
aliquots were thawed on ice, then vortexed for 30 s to mix. Plasma was diluted 4-fold to a
final concentration of 200 mM NH4Ac (adjusted to pH 5 with HAc) with 25 μM 3chloro-L-tyrosine (Cl-Tyr) as internal standard and vortexed for 30 s. Plasma proteins
were removed using a 3 kDa MWCO Nanosep centrifugal device (Pall Life Sciences,
Washington, NY, USA) at 13,000 g for 15 min. A 20 µL aliquot of the plasma filtrate
sample was subsequently used for analysis by MSI-CE-MS under positive ion mode
conditions.
Table S1. A 6-week HIIT intervention for improving cardiorespiratory fitness and body
composition of a cohort of overweight/obese women. Despite between-subject differences in
exercise responsiveness, an overall improvement in aerobic endurance capacity outcomes was
measured relative to baseline (in brackets) that also contributed to a modest decrease in body fat.
Age
BMI
Fat
VO2 max
Workload
Av. Heart Rate
Subject
2
(yrs)
(kg/m )
(%)
(mL/kg/min)
max (W)
(bpm)
S1
24
25.5
33.8 (+0.2)
39.3 (-0.90)
248 (+18)
164 (-8.8)
S2
23
27.8
32.0 (-0.6)
37.5 (0.0)
257 (+32)
158 (-4.1)
S3
23
25.7
31.7 (-0.6)
26.8 (+9.2)
266 (+20)
169 (-8.1)
S4
41
27.8
40.1 (-0.3)
29.3 (+3.1)
205 (+41)
150 (-9.0)
S5
23
27.4
47.1 (-1.8)
29.8 (+11)
173 (+54)
182 (-20)
S6
33
25.6
35.1 (-1.8)
33.9 (+4.8)
228 (+17)
174 (-19)
S7
41
36.1
50.1 (-1.2)
19.4 (+3.0)
190 (+40)
156 (-12)
S8
21
25.1
37.1 (-0.4)
30.4 (+7.6)
194 (+12)
174 (-6.5)
S9
23
30.0
50.5 (-2.4)
24.1 (+5.3)
170 (+30)
170 (-10)
Median
23
27.4
37.1 (-0.6)
29.8 (+4.8)
205 (+30)
169 (-9)
Table S2. Summary of 41 unique plasma metabolites after dilution trend filtering of a pooled
plasma sample when using MSI-CE-MS, which were consistently detected (average CV is 11%
for 9 QCs) in all individual subjects across all time intervals irrespective of training/feeding
status. Each metabolite is annotated by its characteristic m/z:RMT as a paired variable, whereas
identification was confirmed by spiking plasma samples with authentic chemical standards.
Annotated Ion
Precision (QCs)
Molecular Formula
Plasma Metabolite
m/z:RMT
CV% (n=9)
76.0393:0.691
C2 H5 N O2
Glycine (Gly)
6.6
90.0550:0.743
C3 H7 N O2
Alanine (Ala)
7.6
104.0706:0.789
C4 H9 N O2
-7.6
106.0499:0.836
C3 H7 N O3
Serine (Ser)
4.2
114.0662:0.585
C4 H7 N3 O
Creatinine (Cretn)
8.4
116.0706:0.864
C5 H9 N O2
Proline (Pro)
8.9
118.0863:0.818
C5 H11 N O2
Valine (Val)
20
118.0863:0.954
C5 H11 N O2
Betaine (Bet)
13
119.0161:0.970
C4 H6 O2 S
-9.6
120.0655:0.868
C4 H9 N O3
Threonine (Thr)
20
132.0768:0.723
C4 H9 N3 O2
Creatine (Cret)
8.9
132.1019:0.832
C6 H13 N O2
Isoleucine (Ile)
7.2
132.1019:0.843
C6 H13 N O2
Leucine (Leu)
3.7
133.0608:0.882
C4 H8 N2 O3
Asparagine (Asn)
4.9
133.0972:0.567
C5 H12 N2 O2
Ornithine (Orn)
5.9
134.0448:0.980
C4 H7 N O4
Aspartic acid (Asp)
7.5
134.0448:1.343
C4 H7 N O4
-33
137.0457:1.133
C5 H4 N4 O
Hypoxanthine (HyX)
8.0
144.1019:0.970
C7 H13 N O2
-19
147.0764:0.905
C5 H10 N2 O3
Glutamine (Gln)
8.4
147.1128:0.562
C6 H14 N2 O2
Lysine (Lys)
8.5
148.0604:0.920
C5 H9 N O4
Glutamic acid (Glu)
6.6
150.0583:0.889
C5 H11 N O2 S
Methionine (Met)
4.7
156.0768:0.606
C6 H9 N3 O2
Histidine (His)
6.7
162.1125:0.691
C7 H15 N O3
Carnitine (C0)
18
166.0863:0.918
C9 H11 N O2
Phenylalanine (Phe)
7.3
170.0924:0.618
C7 H11 N3 O2
Methylhistidine (MeHis)
7.4
173.0921:0.745
C7 H12 N2 O3
-13
175.1190:0.579
C6 H14 N4 O2
Arginine (Arg)
4.9
176.1030:0.930
C6 H13 N3 O3
Citrulline (Cit)
6.0
182.0812:0.951
C9 H11 N O3
Tyrosine (Tyr)
2.7
192.1594:0.764
C9 H21 N O3
-9.7
204.1230:0.735
C9 H17 N O4
Acetylcarnitine (C2)
15
205.0972:0.917
C11 H12 N2 O2
Tryptophan (Trp)
5.9
209.0921:0.872
C10 H12 N2 O3
Kynurenine (Kyn)
6.9
218.1387:0.756
C10 H19 N O4
Propionylcarnitine (C3)
31
241.0311:0.935
C6 H12 N2 O4 S2
Cystine (Cys-SS)
4.9
276.1199:1.122
C10 H17 N3 O6
-10
286.2013:0.840
C15 H27 N O4
-30
298.0526:0.788
C18 H15 N3 O5 S2 CysGly-Cys-SS (mixed disulfide)
9.5
427.0952:0.997
C13 H22 N4 O8 S2
GSH-Cys-SS (mixed disulfide)
41
(b) HCA/Heat Map: Data Overview
(a) PCA: 2D Scores Plot
(c)!PCA:!2D!Scores!Plot!
Trained#
QC#
Naïve##
!Each!run!has!QC!to!assess!overall!system!dri5 !
CV!=!11%!(n=9,!41!metabolites)!
Figure S1. (a) PCA 2D scores plot highlighting the good reproducibility of MSI-CE-MS based on tight
clustering of 9 QCs relative to large biological variability in plasma metabolome derived from female
subjects in their naïve and trained states. (b) 2D heat map with hierarchical cluster analysis (HCA)
providing a data overview of the dynamic plasma metabotype for each subject based on their training status
and exercise time course during standardized HIIT trials.
(b) PLS-DA: 2D Scores Plot
(a) Volcano Plot: Paired t-Test
* FC > 1.5; p < 0.01
C0
* 95% CI ellipses, R2 = 0.9968; Q2 = 0.2272
GSH-Cys-SS
Trained
C8:1
Normalized Responses
Naive
C0, p = 8.25 E-5
GSH-Cys-SS, p = 5.50 E-4
C8:1, p = 3.42 E-3
* log10 transformed+ auto-scaled data
N
T
VIP Ranking
N
N
T
T
(c) ROC Curves
C0, AUC = 0.951; p = 2.92 E-4
His/C0, AUC = 0.981; p = 2.81 E-5
C0
NT
C8:1
C2
Bet
Cret
Cretn
GSH-Cys-SS
Thr
CysGly-Cys-SS
Cys-SS
Figure S2. Complementary statistical methods for classification of plasma metabolites associated with
adaptive responses to HIIT for subjects while at rest (0 min/baseline), including (a) paired student’s t-test
that takes advantage of the cross-over design, (b) PLS-DA with variable importance in projection (VIP)
ranking of metabolites and (c) receiver operating characteristic (ROC) curves derived from single and
ratiometric plasma metabolites associated with training status. All methods confirm that plasma L-carnitine
(C0) levels are upregulated in trained subjects after a 6 week HIIT intervention.
(b) PLS-DA: 2D Scores Plot
(a) Volcano Plot: Paired t-Test
C2
* FC > 1.5; p < 0.01
* 95% CI ellipses, R2 = 0.9999; , Q2 = 0.2797
Trained
HyX
Normalized Responses
Naive
C2, p = 6.00 E-4
HyX, p = 5.94 E-3
* log10 transformed+ auto-scaled data
N
N
T
VIP Ranking
T
C2
(c) ROC Curves
C2, AUC = 0.901; p = 3.25 E-3
HyX/C2, AUC = 0.975; p = 4.50 E-5
NT
HyX
Bet
Cretn
C0
Cys-SS
C8:1
Cret
His
Figure S3. Complementary statistical methods for classification of plasma metabolites associated adaptive
training responses to HIIT for subjects after strenuous exercise (20 min/post-exercise), including (a) paired
student’s t-test that takes advantage of the cross-over design, (b) PLS-DA with variable importance in
projection (VIP) ranking of metabolites and (c) receiver operating characteristic (ROC) curves derived
from single and ratiometric plasma metabolites associated with training status. All methods confirm that
plasma hypoxanthine (HyX) and O-acetyl-L-carnitine (C2) levels are significantly attenuated and
upregulated after HIIT, respectively.