Establishing Quality Control Metrics for Immunodepletion of High-
Abundance Plasma Proteins and Applying to a Large Clinical CohortMeredith Turner; Erik J. Soderblom; J. Will Thompson; Laura G. Dubois; M. Arthur Moseley
Duke Proteomics Core Facility, Institute for Genome Sciences & Policy, Duke University School of Medicine, Durham, NC
INTRODUCTION
Human plasma is a commonly interrogated biological matrix to identify candidate
protein biomarkers because of its wide range of expressed protein functionalities.
These samples are also relatively easy to obtain from large patient cohorts with
routine sample collection procedures. However, measuring changes in biologically
relevant low-abundance proteins by LC-MS strategies is hindered by the large
dynamic range of these biological matrices. In human plasma, the concentrations of
proteins span over ten orders of magnitude with the fourteen most abundant proteins
making up about 94% of the total protein mass. This large dynamic range of protein
concentrations represents a challenge to the identification and characterization of
biologically significant low-abundance proteins due to a “masking” effect. To address
these challenges, immunodepletion of these high-abundance proteins is often
employed and results in a subsequent increase in protein identification depth of
coverage.
As variations in immunodepletion efficiencies can have consequences on
downstream LC-MS based protein quantitation, we have developed a series of quality
control metrics which allow us to assess the efficiency of immunodepletions using
Agilent’s Multi-Affinity Removal System (MARS-14) LC column. The metrics focus on
bound versus unbound peak area ratios from UV chromatograms, pre- and post-
depletion protein concentration ratios, and image analysis of 1D SDS-PAGE
gels. These metrics provide unique insights into basic sample preparation variables
and importantly allow the identification of potential outliers in the cohort prior to LC-
MS/MS analysis. Potential outliers can then be flagged for statistical consideration in
the downstream analysis or excluded from further analysis all together.
METHODS
Fifty uL aliquots of EDTA human plasma from a 243 patient chronic hepatitis C clinical
cohort were subjected to immunodepletion using a 4.6x100 mm Agilent MARS14
HPLC column (part number 5188-6558) on an Agilent 1100 series LC system. Before
immunodepletion, the aliquots were diluted with 200 uL of Agilent Buffer A , and then
100 uL of the 5x dilution samples was injected on the column. The manufacturer’s
recommended LC gradient protocol was followed. Areas under the curve for bound
and unbound fractions (automated peak integration) were recorded from each UV
chromatogram (280 nm) within Agilent ChemStation (v10.02). and their ratios were
utilized for one depletion metric. Bradford assays (Bio-Rad) were performed on
plasma samples pre- and post-depletion (post-depletion done after a buffer exchange
and ~20x concentration into 50 mM ammonium bicarbonate) and the values were
plotted as the total plasma protein concentration versus the percentage of total
protein in the unbound fraction. Finally, 5 ug of each depleted plasma sample was run
on an Invitrogen NuPAGE 4-12% Bis-Tris gel, stained with Colloidal Blue (Invitrogen)
and subjected to semi-automated densiotometery measurements within TotalLab
Quant (Non-Linear Dynamics).
20000
40000
60000
80000
100000
120000
140000
160000
0 20000 40000 60000 80000AU
C U
nb
ou
nd
Fra
cti
on
AUC Bound Fraction
Flow Cell 2
Unbound AUC vs. Bound AUC
– 243 sample HCV study
MARS14 Depletion
min10 15 20 25 30
mAU
-100
0
100
200
300
400
500
600
700
DAD1 A, Sig=280,16 Ref =360,100 (042009\004-0201.D)
11.018
Area: 2387.19
22.150
Area: 647.967
23.455
DAD1 A, Sig=280,16 Ref =360,100 (042009\003-0101.D)
QC Metric 1 – Characterizing
Depletion Chromatogram
QC Metric 2 – Bradford AssayQC Metric 3 – SDS-PAGE
Gel Image Analysis
QC METRIC 1 - Characterizing Depletion Chromatogram
100
80
60
40
20
0
[To
tal S
eru
m P
rote
in]
(ug
/uL
)
3530252015105
Percentage of Total in Unbound Fraction (%)
55 HCV Samples (1797)
34 DARPA Samples (1871)
MARS14 "Typical" Depletion "Average" Serum Protein Conc (55ug/ul)
Unbound (non-depleted)
proteinsBound (depleted) proteins
**The samples
showing the extra
peak were
previously noted to
be cloudy.
Early eluting peak
possible lipid
(micelle)
contamination.
QC METRIC 2 – Bradford Assay
0
10
20
30
40
50
60
70
80
0 5 10
[Pre
-Dep
leti
on
To
tal S
eru
m
Pro
tein
] (u
g/u
L)
Percentage of Total Protein in Unbound Fraction (%)
Depletion QC Based on Bradford Assays
*Samples highlighted are more than two standard deviations away from the average.
QC METRIC 3 - SDS-PAGE Gel Image Analysis
The three sets of depletion QC metrics described are effective at identifying potential outliers to
assess the efficiency of the depletion protocol. The flagged samples can then be set aside in
statistical analyses so as not to affect the biological variability calculations. The QC metrics were
applied to a large clinical cohort and successfully found samples that were potential outliers.
Features of UV (A280) Chromatogram of
Depleted Human Plasma
Integration of
Unbound Peak
20000
40000
60000
80000
100000
120000
140000
160000
0 20000 40000 60000 80000
AU
C U
nb
ou
nd
Fra
cti
on
AUC Bound Fraction
Unbound AUC vs. Bound AUC
–179 sample HCV study
Raw AUC Intensities
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
50.0 55.0 60.0 65.0 70.0
Un
bo
un
d/B
ou
nd
AU
C R
ati
o
% of Total AUC in the Unbound AUC
Depletion QC Based on AUC Ratios
from 243 sample HCV Study
The area under both curves
is included in the AUC
determination.
www.genome.duke.edu/proteomics/
450 Research Drive, Durham, NC 27710
Flow Cell 1
Unbound/Bound AUC Ratio
Unbound AUC / Bound AUC
% of Total AUC in Unbound AUC
[Unbound AUC / (Unbound AUC +
Bound AUC)] * 100
Red Boxes
Samples highlighted in red are not
within two standard deviations of
the average point, so are flagged
for further consideration.
Pre-Depletion Total Serum
Bradford
Post-Depletion Bradford
Percentage of Total Protein in
Unbound Fraction (%)
Seven bands monitored
across all samples
Representative gel from
243 patient HCV cohort
min5 10 15 20 25
mAU
-100
0
100
200
300
400
500
600
DAD1 A, Sig=280,16 Ref =360,100 (040909\019-1501.D)
1-P2-
E-07
1-P2-
E-08
1-P2-
E-09
1-P2-
F-01
1-P2-
F-02
1-P2-
F-03
1-P2-
F-04
0.00
0.10
0.20
0.30
0.40
1 2 3 4 5
Pro
tein
Con
ce
ntr
ation
Bra
dfo
rd A
ssa
y (
A595)
[ug/u
l]
QC 1
• Characterizing Depletion Chromatogram• Samples not within two standard deviations of the average point are flagged
for further examination
QC 2
• Bradford Assay• Samples with an extremely low post-depletion concentration are not
normalized with the other samples
QC 3• SDS-PAGE Gel Image Analysis• Samples showing a hemoglobin band are run at the end of the queue
[(Volume of plasma depleted* pre-depletion
concentration) / (Volume of depleted plasma
obtained * post-depletion concentration)] * 100
OVERVIEW
Analysis of Atypical Chromatogram Features
CONCLUSIONS