SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary...

8
SUPPORTING INFORMATION Metabolomics for Improved Treatment Monitoring of Phenylketonuria: Urinary Biomarkers for Non-invasive Assessment of Dietary Adherence and Nutritional Deficiencies Jennifer Wild 1 , Meera Shanmuganathan, 1 Mika Hayashi 1 , Murray Potter 2 and Philip Britz- McKibbin 1,2 1 Department of Chemistry and Chemical Biology, McMaster University, Hamilton, Ontario, Canada 2 Department of Pathology and Molecular Medicine, McMaster University, Hamilton, Ontario, Canada * Corresponding author: Philip Britz-McKibbin, [email protected] Table S1; Figure S1-Figure S6 Electronic Supplementary Material (ESI) for Analyst. This journal is © The Royal Society of Chemistry 2019

Transcript of SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary...

Page 1: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

SUPPORTING INFORMATION

Metabolomics for Improved Treatment Monitoring of

Phenylketonuria: Urinary Biomarkers for Non-invasive Assessment

of Dietary Adherence and Nutritional Deficiencies

Jennifer Wild1, Meera Shanmuganathan,1 Mika Hayashi1, Murray Potter2 and Philip Britz-McKibbin1,2 1 Department of Chemistry and Chemical Biology, McMaster University, Hamilton, Ontario, Canada

2 Department of Pathology and Molecular Medicine, McMaster University, Hamilton, Ontario, Canada

* Corresponding author: Philip Britz-McKibbin, [email protected]

Table S1; Figure S1-Figure S6

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

Page 2: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

2 of 8

Table S1. Median concentration and their ranges for urinary Phe and three other catabolites from classic PKU patients (n=16) and their correlation with urinary Phe excretion.

Phe Catabolite m/z:RMT:mode Median (Range) Concentrationa Correlation (R2)

o-Hydroxyphenylacetic acid 151.040:0.969(-) 324 (41-2835) µM 0.469, n=16

Phenylpyruvic acid 163.040:0.981(-) 650 (52-5818) µM 0.808, n=12

Phenylalanine 166.086:0.913(+) 300 (28-1280) µM 0.833, n=13*

N-Phenylacetylglutamine 263.104:0.801(-) 1264 (167-5194) µM 0.889, n=16

a Median concentrations and their ranges for urinary Phe and related catabolites from single-spot urine samples of a sex-balanced cohort of PKU patients (median age of 14 y; 8 M, 8F), where Pearson correlation of urinary Phe concentrations with other urinary metabolites was performed with the exception of urinary Phe, which was correlated with plasma Phe concentrations from matching PKU patients (median = 410 µM; range of 38-1548 µM).

Page 3: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

3 of 8

Figure S1. Representative UPLC-UV run that is used for amino acid analysis following pre-column chemical derivatization for plasma Phe (and Tyr) determination as required for confirmatory testing and treatment monitoring of PKU patients. All measurements were performed independently at McMaster Children’s Hospital (Hamilton, ON, Canada) using standardized protocols/reagents supplied from Waters Inc. (Mass Trak Amino Acid Analysis).

AMQ

Phe

+ R

R R

UPLC-UV (260 nm) Amino Acid Analysis Solution using Waters AccQ-Tag Derivatization Kit:

Phe-AMQ

Plasma Amino Acid Profile from PKU Patient

* 45 min run time/sample with

column reconditioning

445 µM Phe

Page 4: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

4 of 8

Figure S2. Overview of the long-term technical precision of MSI-CE-MS for 40 runs performed over 8 days when using MSI-CE-MS (n=136), which included plasma (n=83), urine (n=28) and repeated analysis of pooled samples as QC (n=25). Overlay of CE current traces for 40 runs over 8 days by for metabolomic analyses of plasma and urine samples with full-scan data acquisition in (A) positive and (B) negative ion mode detection. There was excellent inter-day precision in CE current traces in positive and negative ion mode with average CV of 1.8% and 2.6%, where upper and lower action limits are indicated by dashed line (± 3s). Also, excellent robustness and long-term precision is also demonstrated by control charts for the quantification of the recovery standard, F-Phe that was added to all samples analyzed (n=136), where the solid line represents the average relative peak area, and the dotted lines represent represent the upper and lower action limits (± 3s).

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1

2.2

2.3

0 20 40 60 80 100

Rel

ativ

e Pe

ak A

rea

(RPA

) of F

-Phe

Sample Number

PLASMA PLASMA, QC

0.9

1.1

1.3

1.5

1.7

1.9

2.1

0 10 20 30 40

Rel

ativ

e Pe

ak A

rea

(RPA

) of F

-Phe

Sample Number

URINE URINE, QC

CV = 4.7%CV = 6.5%

A B(C) (D)

23

22

21

20

19

18403020100

30

28

26

24

22

20

403020100

Day 5 Day 6 Day 7 Day 8

BA

n = 2020.5 ± 0.4 µA

n = 2025.4 ± 0.7 µA

Time (min) Time (min)

Cur

rent

(µA)

Cur

rent

(µA)

Day 1 Day 2 Day 3 Day 4

(A) (B)

Page 5: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

5 of 8

PCA: 2D Scores Plot of Plasma from PKU Patients

QC, n=15 Median CV = 15%

PKU, n=16 Median CV = 37%

56 plasma metabolites, autoscaled/glog transformed Median plasma Phe = 254 µM; M=9; F=7; 3 with no diet control

Phe

FC > 1.5; p < 0.05

C3

Tyr ABA Arg

Volcano Plot: PKU Dietary Adherence

FC = 5.1 p = 1.11 E-4

q < 0.05

FC = 0.44 p = 5.45 E-4

q = 0.05

Poor Optimal

Poor dietary adherence/suboptimal à Plasma [Phe] > 360 µM

Poor Optimal

FC = 0.47 p = 1.87 E-3

q < 0.05

Poor Optimal

FC = 0.45 p =4.10 E-3

q < 0.05

Poor Optimal Poor Optimal

FC = 0.59 p = 2.28 E-3

q < 0.05

PC1 (26.9%)

PC2

(15.

5%)

Phenylalanine (Phe) Propionylcarnitine (C3) Tyrosine (Tyr) Aminobutyric acid (ABA) Arginine (Arg)

Poor diet control (n=9): Mean Plasma [Phe] = 757 µM

Mean Age = 9.7 y

Optimal diet control (n=7): Mean Plasma [Phe]= 153 µM

Mean Age = 23 y

AAA Crt

(A)

Figure S3. Data overview when using principal component analysis (PCA) and top-ranked metabolites in (A) plasma, as well as (B) urine from PKU patients reflecting poor adherence to dietary Phe-restriction that exceed recommended therapeutic range (> 360 µM plasma Phe) based on volcano plots (FC > 1.5; p < 0.05). Overall, elevated plasma Phe in PKU patients with poor dietary maintenance was inversely correlated to circulating levels of Tyr, propionylcarnitine (C3), arginine (Arg) and α-aminobutyric acid (ABA), whereas urine reflected increased excretion of Phe, and several Phe catabolites, including phenylpyruvic acid (PPA), phenylacetylglutamine (PAG), as well as hydroxyphenylpyruvic acid (OH-PPA) and phenyllactic acid (PLA), together with lower levels of aminohippuric acid (AHA), as well as a histidine catabolite, uraconic acid (UCA). Overall, poor dietary adherence with corresponding elevated plasma Phe concentrations were associated with adult PKU patients not following specialized diets for Phe restriction (e.g., amino acid supplemental formula).

QC, n=5 Median CV = 12%

PKU, n=16 Median CV = 74%

84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254 µM; M=8; F=8; 5 with no diet control

PPA

FC > 1.5; p < 0.05

UCA

Volcano Plot: PKU Dietary Adherence

FC = 47 p = 3.30 E-5

q < 0.05

FC = 4.3 p = 1.95 E-4

q < 0.05

Poor Optimal

Poor dietary adherence/suboptimal à Plasma [Phe] > 360 µM

Poor Optimal

FC = 2.3 p = 6.16 E-4

q < 0.05

Poor Optimal Poor Optimal

FC = 0.55 p = 5.61 E-3

PC1 (18.4%)

PC2

(15.

2%)

Phenylpyruvic acid (PPA) Phenylacetylglutamine (PAG) Phenylalanine (Phe) Uraconic acid (UCA)

Poor diet control (n=9): Mean plasma [Phe] = 863 µM Mean urine [Phe] = 557 µM

Mean age = 25 y

q < 0.05 q < 0.05

FC = 0.40 p = 4.51 E-3

Poor Optimal

Aminohippuric acid (AHA)

PAG

Phe

Trigonelline AHA ILA

OH-PPA PLA

C0 C4 CS

TMAO

q < 0.05

PCA: 2D Scores Plot of Urine from PKU Patients (B)

Optimal diet control (n=7):

Mean plasma [Phe]= 252 µM Mean urine [Phe]= 112 µM

Mean age = 13 y

Page 6: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

6 of 8

Figure S4. Box-whisker plots for the 7 top-ranked plasma metabolites associated with PKU phenotype that was associated with poor diet control (plasma Phe > 360 µM, n=9) as compared to optimal diet control (plasma Phe < 360 µM, n=7) when using nontargeted metabolite profiling of plasma by MSI-CE-MS. As expected, plasma Phe was independently confirmed to be significantly elevated among poor diet control PKU sub-group with the largest fold-change (FC) increase in response, and strong effect size. Additionally, 6 other plasma metabolites were inversely correlated to Phe and found to be significantly lower/deficient in poor diet control PKU sub-group as compared to treated PKU patients following a Phe-restricted diet and amino acid supplementation as summarized in Table 2.

PhenylalanineFC:5.1

p=1.11E-04Effectsize:2.71

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Poordietcontrol

Op.maldietcontrol

Rel.Pe

akArea

ArginineFC:0.43

p=5.45E-04Effectsize:2.24

TyrosineFC:0.47

p=1.87E-03Effectsize:2.01

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Poordietcontrol

Op.maldietcontrol

Rel.Pe

akArea

2-AminobutryicacidFC:0.59

p=2.28E-03Effectsize:1.86

Propionylcarni+neFC:0.44

p=4.41E-03Effectsize:1.69

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Crea%neFC:0.36

p=8.36E-03Effectsize:1.62

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Poordietcontrol

Op.maldietcontrol

Rel.Pe

akArea

AminoadipicacidFC:0.46

p=8.15E-03Effectsize:1.51

Page 7: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

7 of 8

Figure S5. Extracted ion electropherogram overlay for Phe, N-phenylacetylglutamine, p-cresol sulfate, phenylsulfate, phenyllactate, phenylpyruvate, and o-hydroxyphenylacetate for a representative pooled urine QC sample using MSI-CE-MS with a serial dilution trend filter. Phe was detected in positive ion mode with ion responses scaled down by a factor of 2, whereas all other anionic metabolites were detected in negative ion mode.

3.5 x105

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Ion

Cou

nts

3025201510Time (min)

N-Phenylacetylglutamine263.1037:0.801

Phenylsulfate172.9913:1.158

Phenylpyruvate163.0401:0.981

o-Hydroxyphenylacetate151.0400:0.969

Phenyllactate165.0557:0.923

p-Cresol sulfate187.0070:1.068

Phenylalanine166.0863:0.913

Page 8: SUPPORTING INFORMATION · Median CV = 12% PKU, n=16 Median CV = 74% Mean age = 25 y 84 urinary metabolites, autoscaled/glog transformed, sum normalization Median plasma Phe = 254

8 of 8

Poordietcontrol

Op.maldietcontrol

Rel.Pe

akArea

UrocanicacidFC:0.55

p=5.63E-03Effectsize:1.68

TMAOFC:0.44

p=4.03E-02Effectsize:1.31

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

PhenylpyruvicacidFC:20.54

p=3.68E-06Effectsize:3.77

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

PhenylacetylglutamineFC:2.29

p=6.16E-04Effectsize:2.24

Rel.Pe

akArea

Poordietcontrol

Op5maldietcontrol

PhenylalanineFC:4.30

p=1.10E-03Effectsize:2.42

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Imidazolelac+cacidFC:3.09

p=6.44E-03Effectsize:1.53

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

CresolsulfateFC:0.40

p=3.47E-02Effectsize:1.20

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Hydroxyphenylace.cacidFC:3.79

p=6.16E-03Effectsize:1.46

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Carni&neFC:0.29

p=1.73E-02Effectsize:1.35

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Butyrylcarni+neFC:0.45

p=3.75E-02Effectsize:1.15

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

AminohippuricacidFC:0.44

p=1.72E-03Effectsize:2.04

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

TrigonellineFC:9.74

p=1.35E-02Effectsize:1.45

Poordietcontrol

Op5maldietcontrol

Rel.Pe

akArea

Figure S6. Box-whisker plots for the 12 top-ranked urinary metabolites associated with PKU phenotype that was associated with poor diet control (plasma Phe > 360 µM, n=9) as compared to optimal diet management (plasma Phe < 360 µM, n=7) when using nontargeted metabolite profiling of single-spot urine samples by MSI-CE-MS. As expected, urinary Phe was independently confirmed to be elevated among poor diet control PKU sub-group; however, the major urinary catabolite of Phe excreted among PKU patients was phenylpyruvic acid, which had with the largest fold-change (FC) increase in response with a strong effect size. Additionally, a series of 10 other urinary metabolites were positively (PAG, OH-PAA, ILA, TGL) or inversely (AHA, UCA, C0, CS, TMAO, C4) correlated to PKU phenotype and found to be significantly higher or lower/deficient in poor diet control PKU sub-group, respectively, as compared to treated PKU patients following a Phe-restricted diet as summarized in Table 2.