1. The document describes a study analyzing serum protein profiles from 15 individual samples using online two-dimensional low-flow liquid chromatography-tandem mass spectrometry (LC-MS/MS).
2. Over 400 proteins were quantified across the samples, with 237 proteins showing at least a two-fold change in abundance between samples.
3. Principal component analysis separated the samples obtained from one source, indicating the impact sample preparation can have on protein profiles. Deeper profiling of individual samples is needed for precision medicine applications like biomarker discovery.
1. 1 The world leader in serving science
Alexander Boychenko1, Natalia Govorukhina2, Runsheng Zheng1
1Thermo Fisher Scientific, Germering, Germany
2University of Groningen, Groningen, The Netherlands
Precise Quantitative Serum LC-MS/MS Profiling: The Impact of Sample
Preparation and Sample Source on Biomarker Discovery Studies
For Research Use Only. Not for diagnostic purposes.
3. 3
Blood Complexity
https://www.proteinatlas.org/hu
manproteome/blood
94%
6%
> 3000 proteins identified with MS
> 94 % of total protein amount in
serum is occupied by 14 proteins
Protein concentrations in blood
Protein abundances in blood
78 107
228
498
670
27
838
776
10-100 ug/mL
1-10 ug/mL
0.10-1 ug/mL
0.01-0.1 ug/mL
1e-2 - 1e-3 ug/mL
100-1000 ug/mL
1e-3 - 1e-4 ug/mL
< 1e-4 ug/mL
4. 4
Protein Biomarkers in Blood
S.S. Hori, Sci. Transl. Med. 3 109-116 (2011)
Tumor cells
Normal cells
Specific biomarkersNatural biological
variation of protein
concentration
FDA Approved Biomarkers
(mostly nonspecific)
Aberrant secretions, 6% Receptor
ligands,
18%
Tissue
leakage,
25%
Immunogobulins, 6%
Function
on plasma,
45%
N. Leigh Anderson, Clinical Chemistry 56:2 177–185 (2010)
> 80% proteins have less than 2-
fold abundance variation
51% - normal functions of blood
25% - tissue leakage
18% - receptor ligands
6% - tumor markers
Log2 (Protein Abundance)
Frequency
6. 6
Experimental Design: “Triangular” Biomarker Discovery and Validation Pipeline
Analytical
Variability1 Biological
Variability2
Protein
regulation:
cases vs.
controls
3
Verification
100’s samples
100’s analytes
Validation
1000’s samples
10’s analytes
• Lower analytical variability gives more confidence in discovered
protein abundance changes in cases and controls
• Reliably defined biological variability is the key to unbiased
selection of biomarker candidates
• Deeper proteome profiling increases chance of discovering
specific biomarkers
7. 7
“The Devil is in the Detail”: Multicentric Study and Sample Preparation
SeraLab (UK) – 1 group (Healthy)
Bleed bag unit (not specified)
O/N clotting; T +4 ºC
2,800g; 20min; 5 ºC
UMCG (NL) – 2 groups (CIN0, Healthy)
Serum tube: BD 367953
2 - 8 hours clotting; +20 ºC
3,000g, 10 min; RT
Lund (SE) – 1 group (Healthy)
Serum tube: BD 367615
30 min - 2 hours clotting; +20 ºC
2,000g, 10 min; RT
How reliable is your analytical and data processing pipeline?
Can you identify biomarkers in samples from healthy subjects?
For Research Use Only. Not for diagnostic purposes.
8. 8
Fast, Deep, Quantitative Proteomics Platform
Flow,
µL/min
Samples
per 24
hours
MS
utilization,
%
Average
PWHM, sec
Average
PW base,
sec
Asymmetry
60 min 0.300 24 95 10 19 1.23
48 min 0.600 30 90 9 18 1.21
24 min 0.800 60 87 7 13 1.17
14.4 min 1.000 100 85 4 7 1.13
8 min 1.500 180 75 3 6 1.16
75 µm x 150
mm, 3 µm
PN ES800
Thermo Scientific™
UltiMate™ 3000
RSLCnano system
Thermo Scientific™ Q
Exactive™ HF-X Hybrid
Quadrupole-Orbitrap™
mass-spectrometer
5049
10468
14107
18261
24932
1069
1708
2258
2720
3432
0
500
1000
1500
2000
2500
3000
3500
4000
0
5000
10000
15000
20000
25000
30000
35000
8 min 14.4 min 24 min 48 min 60 min
Proteingroups
Peptidegroups
HeLa digest, 200 ng
1071
1703
2063
2474
2767
138
178
217
230
265
0
50
100
150
200
250
300
0
500
1000
1500
2000
2500
3000
3500
4000
8 min 14.4 min 24 min 48 min 60 min
Proteingroups
Peptidegroups
Serum digest, 0.05 uL
Standardized low-flow LCMS methods
60 min low-flow LCMS
method was used for
analysis of 15 crude
serum samples
For Research Use Only. Not for diagnostic purposes.
9. 9
Analytical Variability: Precision of Low-Flow LCMS Analysis
Analytical
Variability1 Biological
Variability2
Proteins
regulation:
cases vs.
controls
3
RSD, %
log2 Protein abundance ratios
• 3 LCMS replicates per sample
• Label-free quantification
Proteins identified (1% FDR): 265
Proteins quantified: 248
• RSD < 15% 218 (88%)
• 95% of proteins have abundance
variation < 50%
For Research Use Only. Not for diagnostic purposes.
10. 10
Experimental Design: Biological Variability
Analytical
Variability1 Biological
Variability2
Proteins
regulation:
cases vs.
controls
3
Analytical variability
Biological variability Sweden
UMCGCIN0
Seralab
CIN0
Seralab
UMCG
Sweden
• Biological variability is
significantly larger in comparison
to analytical variability
• Still 80% of proteins have less
than a 2-fold change
For Research Use Only. Not for diagnostic purposes.
11. 11
Experimental Design: Regulated Proteins
Analytical
Variability1 Biological
Variability2
Proteins
regulation:
cases vs.
controls
3
Seralab
CIN
UMCG
CIN
Sweden
CIN
UMCG
Seralab
Sweden
Seralab
Sweden
UMCG
UP and DOWN regulated proteins
Protein ratio distribution, selection criteria
80%
90%
1-1
Cauchy σ=0.310
Reliability of quantification, Biological functions
Data Analysis pipeline
2-2
Boichenko et al. J Proteome Res. 2014 Nov 7;13(11):4995-5007
Selection criteria
• 94 proteins had at
least one abundance
ratio > 2
• 22 proteins had
abundance ratio > 4
• 16 of them can be
mapped to blood atlas
For Research Use Only. Not for diagnostic purposes.
12. 12
Why Did We Find Proteins Regulated in Healthy Groups?
KRT20
FGG
LPA
PPIA
APOD
SERPINA5
KRT2PF4V1
PPBP
HBD
CNDP1
HBB
HBA1
ORM1
THBS1
CAT
-6
-5
-4
-3
-2
-1
0
1
2
3
4
0 500 1000 1500 2000 2500 3000 3500
LOG2(PROTEINCONCENTRATION)
PROTEIN NO.(SORTEDBY ABUNDANCE)
Protein Concentration (Human Blood Atlas)
Regulated (p >0.8), fold change > 2
Regulated (p > 0.9), fold change > 4
Significant regulation of classical blood proteins in healthy groups is very unlikely and can lead to
• Artificial protein abundance alteration in one or multiple individual samples (storage, analysis error, etc.)
• Artefacts arising from sample preparation
KRT20
FGG
LPA
PPIA
APOD
SERPINA5
KRT2
PF4V1
PPBP
HBD
CNDP1
HBB
HBA1
ORM1
THBS1
CAT
-3
-2
-1
0
1
2
3
4
0 200 400 600 800 1000
LOG2(PROTEINCONCENTRATION)
PROTEIN NO. (SORTED BY ABUNDANCE)
Protein Concentration (Human Blood Atlas)
Regulated (p >0.8), fold change > 2
Regulated (p > 0.9), fold change > 4
For Research Use Only. Not for diagnostic purposes.
13. 13
Protein Regulation in Individual Samples
HBD
IGKV1D-13
APOD
IGKV2D-30
KRT2
CAT
HBB
HBA1
PF4V1
PPBP
P0DOX6
CNDP1
THBS1
ORM1
IGHG3
IGLV2-23
PPIA
KRT20
SERPINA5
LPA
IGHD–noref.
FGG
CIN0_01
CIN0_02
CIN0_03
Seralab_01
Seralab_02
Seralab_03
Seralab_04
Seralab_05
Sweden_01
Sweden_02
Sweden_03
UMCG_01
UMCG_02
UMCG_03
-6.0
-3.0
0.0
3.0
6.0
Log2(ProteinAbundance)
Proteins downregulated in Seralab samples represent classical blood proteins expressed in specific blood cells
released during the serum preparation process
A deeper dive into the proteome is required to find specific biomarkers
PCA of 15 individual samples
For Research Use Only. Not for diagnostic purposes.
15. 15
Deeper Proteome Profiling with Comprehensive Online-2D Low-Flow LC-MS/MS
75 µm x 150 mm,
3 µm
PN ES800
UltiMate 3000 RSLCnano
Online 2D configuration
Thermo Scientific™
Orbitrap Exploris™ 480
2nd Dimension
1st Dimension
PepSwift, Monolithic
Column 100 µm x 250 mm
2 fractions
(135 min)
4 fractions
(225 min)
8 fractions
(405 min)
2292
407
1600 • Automated sample analysis (no manual
sample handling)
• Orthogonal selectivity in 1st and 2nd
dimension
• Low overlap of peptides elution between
fractions
• Deeper complex proteome profiling
810
1403
1519
782810
1221 1266
595
810
2031
3297
3892
0
500
1000
1500
2000
2500
3000
3500
4000
4500
0
200
400
600
800
1000
1200
1400
1600
1 2 3 4
Fraction #
UniPep New.unipep Acc.unipep
For Research Use Only. Not for diagnostic purposes.
16. 16
Increased Dynamic Range with Online 2D Low-Flow LC-MS/MS
2840 2884
3764
4825309
265
329
377
0
50
100
150
200
250
300
350
400
0
1000
2000
3000
4000
5000
6000
Merged 15
samples x 60
min, 15
hours
0.25uL
serum, 2
frac, 2.25
hours
0.25uL
serum, 4
frac, 3.75
hours
0.50uL
serum, 8
frac, 6.75
hours
Peptide Groups Protein Groups
-6
-4
-2
0
2
4
0 500 1000 1500 2000 2500 3000 3500
LOG10(PROTEINCONCENTRATION)
PROTEIN NO. (SORTED BY ABUNDANCE)
Human Blood Atlas
Merged 15 samples x 60 min, 24 hours
2 fractions, Online 2D, 2.25 hours
4 fractions, Online 2D, 3.75 hours
8 fractions, Online 2D, 6.75 hours
• Higher loading and peak capacity of online 2D low-flow LC-MS/MS analysis results in deeper serum
proteome coverage
• Protein and peptide identifications increase linearly with number of fractions
• Automated on-line 2D LC-MS/MS analysis can be easily adjusted to any number of fractions and combined
with DDA and DIA MS acquisition techniques
For Research Use Only. Not for diagnostic purposes.
17. 17
From Groups to Individual Samples: Focus on Precision Medicine
• 15 individual samples were analysed
with low-flow online 2D LC-MS/MS
• 442 proteins were quantified based
on 3536 unique peptides
• 110 protein ratio combinations were
computed for 15 samples
• 237 proteins had at least one
abundance ratio > 2
• 151 of these proteins were mapped
to the Human Blood Atlas
• PCA shows clear separation of samples obtained from Seralab
• There is no clear differentiation between other sample groups
For Research Use Only. Not for diagnostic purposes.
18. 18
Pathway Mapping for Serum Regulated Proteins https://reactome.org/
Most UP and DOWN regulated proteins in individual samples were related to clotting or generic immune system
response
For Research Use Only. Not for diagnostic purposes.
19. 19
Protein Regulation in Individual Samples
CIN0_01
CIN0_02
CIN0_03
Seralab_01 (ref.)
Seralab_02
Seralab_03
Seralab_04
Seralab_05
Sweden_01
Sweden_02
Sweden_03
UMCG_01
UMCG_02
UMCG_03
UMCG_04
C4A
SAA2
CA1
CD5L
ACTC1
LYZ
KRT14
CFHR5
MBL2
PCOLCE
ICAM2
PFN1
PRDX2
PEPD
CAT
MASP2
LCN2
PTPRJ
LAMP2
NRP1
LTF
TAGLN2
PRSS3
SERPINA11
BLVRB
TPM3
MAN1A1
CACNA2D1
MMRN1
PCSK9
COL6A1
PDLIM1
PRG2
TREML1
PGD
SRGN
SPTA1
ADAMTS13
CD99
APP
INHBC
PRL
HBB
HBD
HBA2
PF4V1
APMAP
CRP
PPBP
SPARC
SAA1
APOD
ALDOB
AZGP1
ORM2
PRTN3
APOC3
PZP
TIMP1
MCAM
PON3
CHL1
RBP4
FLNA
TSKU
APOC2
COLEC11
APCS
TPM4
ADIPOQ
KRT2
NCAM1
AGT
PROZ
SERPINA6
APOC4
F11
EFEMP1
GPX3
ALDOA
TLN1
PPIA
KRT10
H6PD
ORM1
HIST1H1C
NID1
AOC3
IGF1
ENPP2
PRG4
APOM
MMRN2
KNG1
GP1BA
LUM
PI16
VNN1
TFRC
IL1RAP
FCN3
DBH
LCP1
DSG2
PRKCSH
CRTAC1
COLEC10
TUBB3
OAF
FYN
LGALS3BP
CFHR4
FBLN1
CNTN1
CFHR1
KRT9
VCAM1
PLA2G7
HSPD1
MASP1
YWHAZ
TNXB
VASN
SERPINA5
BPIFB1
MASP1
SERPING1
LDHB
LPA
PLXDC2
PRDX1
PIGR
F13A1
SHBG
HIST1H4A
POSTN
GGH
ICAM1
ENG
CPS1
MUC5B
COMP
CA2
ANG
HSP90AB1
PKM
FGB
FGG
THBS4
FCGR3A
KRT17
The analysis of individual samples is complicated by a broad variation in expression across healthy individuals,
sample preparation, analytical variability
A number of proteins in Seralab samples show significant abundance alternations in comparison with other groups
For Research Use Only. Not for diagnostic purposes.
20. 20
Significant Sample Specific Regulations: Potential Biomarkers
Comparison of Targeted Mass Spectrometry Techniques with an Immunoassay: A Case Study for HSP90α
Guzel C., et al Proteomics Clin. Appl. 2018, 12, 1700107
HSP90AB
PKM
Deeper blood proteome profiling is required to retrieve sample protein abundance changes
for disease specific biomarkers
HSP90AB PKM
Low tissue, cell specificity
Low cancer specificity
CIN0 Seralab Sweden UMCG
CIN0 Seralab Sweden UMCG
Low tissue, cell specificity
Low cancer specificity-4
-3
-2
-1
0
1
2
3
4
CIN0_1
CIN_02
CIN_03
Seralab_01
Seralab_02
Seralab_03
Seralab_04
Seralab_05
Sweden_01
Sweden_02
Sweden_03
UMCG_01
UMCG_02
UMCG_03
UMCG_04
PZP
For Research Use Only. Not for diagnostic purposes.
21. 21
• Standardized high-throughput low-flow LC-MS methods can be used for routine profiling of TOP
200 proteins or development targeted assays with high sensitivity
• Standardized sample preparation of biofluids and critical analysis of potential validation targets is
essential for biomarkers discovery studies.
• Profiling of high abundant blood (serum, plasma) proteins has limited benefit for the discovery of
new biomarkers or population related changes due to high stability of the blood proteome and non-
specific changes related to the immune system response
• Comprehensive high pH RP x low pH RP low-flow online 2D LC-MS/MS analysis shows great
potential for deep, automated quantitative blood proteome profiling required to discover cases
specific biomarkers
Conclusions
For Research Use Only. Not for diagnostic purposes.
22. 22
Acknowledgements
RUG, Groningen, NL UMCG, Groningen NL
R. Bischoff A. van der Zee
N. Govorukhina K. ten Hoor
H. Klip
University of Groningen
Erasmus MC, Rotterdam
Dr. T. Luider
C. Guzel
Germering, Germany
Bremen, Germany
T. Arrey
P. Jehle
W. Decrop
C. Pynn
R. Zheng
For Research Use Only. Not for diagnostic purposes.