Date post: | 21-Jun-2018 |
Category: |
Documents |
Upload: | trinhtuyen |
View: | 226 times |
Download: | 1 times |
1
Comparison of commercial nano LC columns for fast, targeted
proteomics of cancer cells
Tore Vehus1*, Kristina Erikstad Sæterdal1, Stefan Krauss2, Elsa Lundanes1, Steven
Ray Wilson1
1Department of Chemistry, University of Oslo, Post Box 1033 Blindern, NO-0315 Oslo, Norway, 2Unit for
Cell Signaling, SFI-CAST Biomedical Innovation Center, Oslo University Hospital, Rikshospitalet, NO-0027
Oslo.
*Corresponding author: Tore Vehus [email protected] +47 97 51 81 13
Abstract
Targeted proteomics through nanoLC-MS/MS has the potential to measure several proteins
in a complex sample in reasonable analysis time. However, to be a contender to existing
methods; WB and ELISA, sample preparation has to be minimized, sensitivity high and
analysis time short.
We compared four commercially available nanoLC columns with various morphologies
(Chromolith® (silica monolith), PepMap™ (porous particles), Accucore™ (solid core particles),
and PepSwift™ (organic monolith)) for determination of Wnt/beta-catenin related pathway
proteins in unfractionated/undepleted cell samples with gradients of 30 minutes.
We found that the solid core particle packed column outperformed the other columns in
terms of peak capacity (~190 at 50% peak height in 30 minutes). The average LC retention
time shift was also least for the two particle based columns (<1 minute), whereas the
monoliths were more prone to retention time instability when sample complexity increased.
This will affect selectivity of identifications in a complex background, and effectively reduce
assay robustness. Of the particle-packed columns, the solid core particle column had the
best retention time repeatability in complex samples (1 % RSD). Only small difference in
sensitivity was observed, even if the flow-rate was altered in the range of 200 – 600 nL/min.
With the solid core particle packed column, Wnt/beta-catenin signal pathway members
beta-catenin and GSK3beta were easily identified in non-fractionated samples and could be
measured precisely (below 10 % variation) in complex backgrounds using SILAC labeled cell
lines as internal standard which corrects for each sample preparation step and LC-MS
variation.
However, less abundant pathway proteins such as AXIN2 or TNKS2 could not be detected,
perhaps calling for more narrow LC columns for fast trace analysis.
2
Introduction
Nano LC is widely used for comprehensive proteomics to obtain qualitative and quantitative
information of as many proteins as possible. Even though, it is known that in common data-
dependent acquisition, many peptides are not detected due to the vast complexity, dynamic
range and diversity in biological samples (1). Using LC-ESI-MS/MS, typically with
selected/parallel reaction monitoring (SRM or PRM, depending on the instrument), targeted
proteomics aims to selectively identify/quantify specific proteins, preferably within short
analysis times compared to more comprehensive approaches (e.g. 30 minutes instead of 2
hours) (2-4). Therefore, the speed and increased reliability can make targeted approaches
applicable for e.g. clinical settings.
However, the shorter the time used for chromatographing compounds, the greater chance
the target molecules will fail to enter the MS for detection, due to ion suppression by other
(co-eluting) peptides during the electrospray process. This is especially the case if
chromatographed samples are complex (e.g. not pretreated with immunoaffinity-based
isolation). Co-elution can be reduced using high performance LC columns, which produce
narrow peaks and hence increased resolution. A key descriptor of an LC system´s resolution
is its peak capacity (i.e. the number of compounds that can be chromatographically
separated) (5, 6). This number can depend on e.g. the column material/stationary phase,
column length, mobile phase conditions (flow-rate, solvent strength, gradient time etc.),
column packing/polymerization procedure and analytes. In comprehensive proteomics,
correlations between number of identified peptides and peak capacity have been shown (7,
8). Another desired trait of an LC column is a robust/repeatable retention time, which can
support compound identification and allow for narrow SRM/PRM scheduling. Compared to
comprehensive approaches, retention time robustness is arguably more focused upon in
targeted proteomics.
Nano LC columns employed in proteomics are typically defined as having inner diameters
(IDs) ranging from 10 to 100 µm, with mobile phase flow-rates operating between 10 and
500 nL/min. The narrow IDs and low flow-rates can significantly enhance sensitivity/reduce
ion suppression by decreasing radial dilution during chromatography and allowing
production of smaller electrospray droplets, respectively (9, 10).
As the case is with larger bore-LC, nano-columns for reversed phase (RP) LC are by far most
common. Several types of nano RPLC columns are commercially available, e.g. packed with
solid core particles or totally porous particles, silica-based monoliths and organic monoliths
(11-13). Today, totally porous particles for LC are often 2 μm or smaller in diameter, as these
sizes are able to provide higher efficiencies (also at higher speeds) compared to the more
traditional 3.5-5 μm particles (14). However, smaller particles can generate higher
backpressures. Columns packed with solid core particles have comparable speed and
resolution to small porous particles, with lower backpressures, but possibly with lower
loading capacity (15). Monoliths are single piece skeletal structures that are synthesized
3
within the column, and are mostly organic- or silica-based (16). These structures can also
provide excellent efficiency and increased speeds due to their high permeability (17).
However, these columns are more challenging to prepare reproducibly (18, 19).
All of these column variants can provide high-resolution separations for comprehensive
proteomics (13, 14, 20, 21). But to the authors´ knowledge, the performance of the various
types of nano RPLC columns has not been compared regarding targeted proteomics
experiments (e.g. relatively fast gradients, complex samples and increased focus on LC
robustness/repeatability). We wished to investigate which of four commercially available
columns with the above-mentioned morphologies provided the best chromatographic
performance (peak capacity, peak shape, carry-over, retention time repeatability etc.)
utilizing 30-minute gradients of very complex samples (as opposed to e.g. only assessing
protein standards). Also, we wanted to investigate whether these traits were associated with
success in identifying central proteins of the cancer associated Wnt/beta-catenin-pathway
(22) in complex samples, that also included a universal internal standard solution.
Materials and methods
Sample preparation
Recombinant APC (H00000324-Q01) and AXIN2 (H00008313-Q01) were purchased from
Abnova (Tapei City, Taiwan). Glycogen synthase 3β (GSK3β) were from Life Technologies
(Carlsbad, CA, USA) and beta-catenin (12-537) from Millipore (Billerica, MA, USA). The poly-
ADP-ribosylation polymerization (PARP)-domain of human tankyrase2 (TNKS2) was produced
as described in (23). All amino acid sequences can be found in Supplementary Table 2.
Each standard protein was digested with trypsin (Promega, Madison, WI, USA). Briefly, 10 μg
of each protein was dissolved in 1 mL 8 M urea (Sigma-Aldrich) dissolved in 50 mM Tris-HCl
pH 8.0 (Sigma Aldrich). The samples were reduced in 5 mM dithiothretiol (DTT, Sigma Aldrich)
at 37°C for 30 minutes and alkylated with 15 mM iodoacetamide (IAM) for 15 minutes in the
dark. Trypsin was added to a protein:enzyme ratio of 1:20, and incubated over night at 37°C.
The digested standards were desalted using solid phase extraction (SPE) on RP C18 cartridges
(Bond Elut C18, 100 mg, Agilent, Santa Clara, CA, USA) with water (Millipore) and eluted in 1
mL 80 % acetonitrile (ACN, HiPerSolv CHROMANORM, VWR, Radnor, PA, USA) with 0.1 % FA
(Sigma Aldrich) (v/v) and dried with a SpeedVac (Thermo Fischer Scientific; former Savant,
Waltham, MA, USA). Each standard were reconstituted in 0.1 % (v/v) TFA (Sigma Aldrich) to a
final concentration of 10 μg/mL.
A set of external standard mixtures (ExSMix) containing 1, 0.5, 0.1, 0.05, 0.01, 0.005, 0.001,
0.0005 and 0.0001 μg/mL of each protein standard were prepared by appropriate dilution
with 0.1 % (v/v) TFA.
4
HCT15 cells (American Type Culture Collection, ATCC, VA, USA) were cultured in RPMI 1640
medium (Life Technologies) supplemented with 10 % fetal bovine serum (FBS, Life
Technologies) and penicillin streptomycin (PS, Life Technologies) and harvested with trypsin
EDTA (Life Technologies) at 80 % confluence. The cells were counted and washed in
phosphate-buffered-saline (PBS, Oslo University Hospital, Oslo, Norway). The proteins were
extracted and digested using the filter-aided sample preparation (FASP) protocol (24). Briefly,
1 million cells were re-suspended in 200 μL lysis buffer, heated for 15 minutes at 70°C and
sonicated for 5 minutes. Debris was removed with centrifugation at 13000 rpm for 10
minutes in a thermostated centrifuge at 20°C (Eppendorf, Hamburg, Germany). The protein
concentration was determined using Bradford Assay (Bradford Quick Start Assay, Bio-Rad,
CA, USA) at absorbance of 595 nm with bovine serum albumin (BSA, Sigma Aldrich) used as
calibration standards. About 100 μg protein was added to 10 kDa 0.5 mL filter devices
(Millipore). The FASP two-step digestion protocol with Trypsin-LysC mix (Promega) (1:20
protein:enzyme ratio) was followed (24). The peptides were desalted using the above-
described SPE-procedure. Peptide concentrations were determined using a NanoDrop2000
instrument (Thermo Fisher Scientific) with absorbance at 205 nm with 31 mg/mL absorption
coefficient. The samples were diluted to a final concentration of 1 mg/mL with 0.1 % (v/v)
TFA.
An internal standard protein solution (ISprot) was prepared by stable isotope labeling of
amino acids in cell culture (SILAC) labeling of HEK293 (ATCC) and HCT15 cells according to
the procedure described by Ong & Mann (25) with 13C615N4-arginine and 13C6-lysine (+10.008
and +6.020 Da, respectively) (Thermo Fisher Scientific) supplemented to RPMI1640 Media
for SILAC acquired from Thermo Fisher Scientific. The labeled cell lines were subsequently
lysed as described above and added to samples prior to protein digestion. Heavy amino acid
incorporation was verified with data-dependent LC-MS/MS analysis of non-labeled cell lines
and labeled cell lines (data not shown).
Treatment of cell lines with G007-LK
The colon carcinoma cells were seeded (100,000 cells/well) in 6-well plates (Nunc™ Cell-
Culture Treated Multidishes, Thermo Fisher Scientific). RPMI1640 were used for HCT15 and
COLO320DM (ATCC) cells with incubation in 5 % CO2, and Leibowitz L-15 medium (Thermo
Fisher Scientific) for SW480 (ATCC) cells with incubation in 0 % CO2. After 24 hours, the
medium was removed and the tankyrase inhibitor G007-LK (23) (dissolved in
dimethylsulfoxide (DMSO, Sigma-Aldrich)) was added to a final concentration of 1 μM in the
cells’ respective medium. An equal volume of DMSO was added as negative control. After 24
hours of incubation the cells were harvested and processed as described above.
Three biological replicates were made and analyzed, with exception of COLO320DM and
SW480 cells, were treated cells were analyzed in duplicates.
LC instrumentation
5
A NanoLC1000 pump from Thermo Fisher Scientific was used in this study. The mobile phase
A (MP A) contained 0.1 % (v/v) FA in H2O (Optima® LC/MS, Fisher Scientific, part of Thermo
Fisher Scientific) and mobile phase B (MP B) contained 0.1 % (v/v) formic acid (FA) in ACN.
The pre- and analytical columns were coupled through a stainless steel T-piece (Valco, VICI
AG International, Schenkon, Switzerland). Each gradient of 30 minutes was followed with a
linear increase to 95 % MP B for 10 minutes and a 10 to 15 minute hold at 95 % MP B. Each
pre- and analytical column was equilibrated with at least 6 column volumes.
The Acclaim® PepMap RSLC (PepMap™) 75 μm x 20 mm and 50 μm x 150 mm particle
packed pre- and analytical columns, the PepSwift® 200 μm x 5 mm and 100 μm x 250 mm
monolithic poly-styrene divinylbenzene (PS-DVB) pre- and analytical columns, the Accucore®
75 μm x 150 mm solid core particle packed column were from Thermo Fisher Scientific and
the 100 μm x 50 mm and 50 μm x 150 mm Chromolith® Caprod® silica monolithic pre- and
analytical columns were from Merck-Millipore.
MS and MS/MS-parameters
The electrospray voltage was set to 1.8 kV for the 2 μm ID stainless steel emitter (Thermo
Fisher Scientific), the 5 and 8 μm ID New Objective Emitters (New Objective, Woburn, MO,
USA). The Accucore® and PepSwift® were connected to the MS through PicoTip ESI emitters
fitted for the column flow-rate used. The PepMap™ and Chromolith® columns were
connected to stainless steel emitters.
A Q Exactive™ Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific)
was used for the entire study. Two main methods were used, full-MS with subsequent data-
dependent MS/MS (ddMSMS) and targeted-MS/MS with selected ions from the proteotypic
peptides chosen. In full-MS, the resolution was set to 140,000 @ m/z 200, automatic gain
control (AGC) to 1,000,000, maximum inject time to 100 ms and scan range m/z 350-1850.
The 10 most intense ions were selected and fragmented using normalized collision energy
(NCE) of 25 % and the MS/MS scans were acquired with; resolution of 70,000, AGC target
value of 100,000, and maximum inject time of 500 ms. Dynamic exclusion was enabled for
20 seconds, and charge states of 1 and >7 were rejected for fragmentation.
For targeted MS/MS, each target was monitored in a retention time window of +/- 4 minutes
relative to the retention time determined by ddMSMS, and either operated in single or
duplexing mode. For single ion fragmentation, the maximum injection time was set to 500
ms, with an AGC target value of 100,000, and a resolution of 140,000. The isolation width
was set to m/z 4.0 and the NCE to 25 eV.
In duplex mode, the resolution was decreased to 35,000, the maximum inject time lowered
to 100 ms and the AGC target value kept unaltered. Each target was fragmented with an
isolation of m/z 2.0 and NCE of 25 eV.
Peptide identification and proteotypic peptide selection
6
Using the Skyline software (v2.5) (26), a theoretical tryptic digest of each protein standard
was made, and together with data-dependent LC-MS/MS runs for each column, a set of
proteotypic peptides was selected for each protein (Table 1). The peptides selected were
checked for uniqueness against the Uniprot database (27) and only peptides selectively
representing the proteins of interest were chosen as proteotypic peptides. Peptides
containing cysteine, methionine, serine, tyrosine and threonine were avoided if other
possibilities were available.
Data processing
All chromatograms were analyzed with Xcalibur software (v2.1, Thermo Fisher Scientific) and
Proteome Discoverer (v1.4, Thermo Fisher Scientific) with the sequest algorithm were used
for searching against the protein sequences found in Supplementary Table 1. For
identification in ddMSMS, maximum two missed cleavages, maximum 10 ppm precursor and
0.6 Da mass tolerance, respectively, were allowed, combined with a false discovery rate of
0.01. Carbaiodomethylation was set as a static modification and oxidation of methionine as
dynamic. Targeted MS/MS data acquired at resolutions of 140,000 and 35,000 were
extracted using 7 and 25 ppm, respectively. Peak widths, heights and areas were measured
manually, and peak capacity (nc) at half peak height was calculated according to Equation 1,
𝒏𝒄 =𝒕𝑹,𝒍𝒂𝒔𝒕−𝒕𝑹,𝒇𝒊𝒓𝒔𝒕
𝒘𝟎.𝟓̅̅ ̅̅ ̅̅ Equation 1
where 𝒘𝟎.𝟓̅̅ ̅̅ ̅̅ is average peak width at half peak height, tR,first and tR,last are the retention times
for the first and last eluting peak, respectively.
For statistical analysis, F-test and two-sided student t-test were used. LC-MS/MS raw data is
available on the following link http://www.mn.uio.no/kjemi/supplementary-
data/bach/comparison-of-commercial-nano-lc-columns-for-fast-/
Results
Framework and plan of study
The RP columns investigated were; Chromolith® CapRod (C18 silica monolith), PepMap™ (C18
porous particles), Accucore™ (C18 solid core particles) and PepSwift™ (PS-DVB organic
monolith). The columns had to be compatible with a (commonly used) nano EASY-nLC 1000
pump (Thermo Fisher Scientific), the pump available to us at the time of study (and not
equipped with a column oven); this excluded the testing of instrument specific units as the
Agilent CHiP systems and Waters nanoAcquity column systems. The columns had some
variations in dimensions (50-100 µm IDs and 15-25 cm length); identical dimensions were
not commercially available for all columns. The linear separation gradient was set to 30
minutes (=tG), as this was considered to be an acceptable compromise between speed and
risk of ion suppression in tryptic digest of un-fractioned human cell lysate (containing
between 10,000 – 30,000 protein species). For each column, the gradient was adjusted so
7
that the last eluting peptide of interest eluted at tG, allowing for a fair comparison between
the chromatographic properties. A standard mobile phase consisting of water, 0.1 % FA, and
ACN were used for all columns. Proteins studied were AXIN2, beta-catenin, GSK3beta and
TNKS2. Additionally, the C-terminal of APC was included, and served as a negative control for
protein identification in the APC-mutated colon carcinoma cell lines. Although this is a
limited number of analytes, these proteins represent a central selection of Wnt-related
proteins and relevant range (for signaling pathway analysis) of endogenous concentrations,
where the low-abundant concentrations may not be detected in ddMSMS) (28). Also, these
analytes are also available for use as external standards (vital for several of the
investigations performed here, but also for “tier 1” level targeted proteomics in general (29)).
The proteotypic peptides generated from these proteins had molar mass of 0.8-1.9 kDa, and
GRAVY index values of -2.32 - 1.24, which are arguably representative for tryptic peptides in
general. The amounts of complex samples loaded onto each system were between 0.5-1 μg,
amounts which are common in proteomics experiments (30, 31) and well below the pre-
column capacities reported by the manufacturers. The LC-MS/MS system set-up was
considered “healthy” as 40 proteins/separation minute could be identified in comprehensive
mode (data not shown).
The main steps of the study were: Optimization of peak capacity for each column and
evaluation of other basic LC traits; Evaluation of repeatability/robustness by comparing
system behavior/performance between that of an external standard mix of the targeted
proteins (“ExSMix”) and lysates spiked with the standards; Investigation of which target
peptides (endogenously present at trace –to moderate levels) were possible to detect in an
unspiked sample; Evaluation of relative quantification by LC-MS/MS with isotopically labeled
internal standards and comparison with established Western blotting (WB) methods.
Columns/hardware were installed by personnel with considerable experience in low nano
flow-rate instrumentation. Just one column per column type were tested; however, columns
used for experiments performed satisfactorily with respect to the manufacturers´
specifications upon installment (or were replaced if they did not; this was the case for one
column), and batch-to-batch issues were not explored.
Chromatographic performance
To compare the nano LC columns´ performance for 30 minute gradients, the flow-rate and
gradient composition were optimized for each column with regards to peak capacity
according to the procedure described by Wang et al. (6). The definition used for peak
capacity is found in Eq.1 (Materials and Methods) A key point is to fully exploit the
separation window, ensuring that the most hydrophobic analytes elute at the end of the
gradient (tfinal peak = tG). The sample was the ExSMix (see Materials and Methods).
Table 1 shows the peak capacity and peak asymmetry for each column set-up, as well as the
enabling solvent conditions. (Table 1 here) The highest peak capacity was obtained with the
solid core particle packed column set-up, with an average peak capacity close to 190 (Figure
1); approximately 1.5 times larger compared to the second-best performing column (the
8
silica monolith, peak capacity = 130) with our conditions. The solid core particle packed
based column also provided the least peak tailing and the narrowest peaks of the columns
included in this study, with an average asymmetry of 1.1 and a base peak width of 10
seconds. (Figure 1 here) The two monolithic columns tested displayed more peak tailing
compared to the particle-based columns (1.4-1.5 vs 1.1-1.3).
Table 1 - Optimal flow-rate, gradient composition, peak capacity, peak width and asymmetry based on data-
dependent MS/MS scans of separated 1 ng ExSMix from recombinant APC, AXIN2, beta-catenin, GSK3beta and
TNKS2 for the four column set-ups. Retention time RSD values are based on 3 technical replicates.
Column
Precolumn Analytical column Precolumn/analytical
column material
Optimal flow-rate
(nL/min/ linear
velocity in
cm/second
Gradient end
(% (v/v) ACN)
Retention
time
variation
(RSD, %)
Peak
capacity
at half peak
height
Peak width
at half peak
height
(sec)
Asymmetry
ID
(μm)
Length
(mm)
ID
(μm)
Length
(mm)
Chromolith®C
apRod® 100 50 50 150
C18 silica monolithic/
C18 silica monolithic 200/ 0.8 36 0.3 130 19.8 1.4
PepMap™ 75 20 50 150 C18, 3µm, 100Å particles/
C18, 2µm, 100Å particles 200/ 0.9 36 0.4 106 16.2 1.1
Accucore™ 75 20 75 150
C18, 3µm, 100Å particles/
C18, 2.6µm, 80 Å solid core
particles
300/ 0.7 36 0.2 189 10.2 1.1
PepSwift™ 200 50 100 250 PS-DVB monolithic/
PS-DVB monolithic 500/ 0.5 20 1.0 100 16.2 1.5
9
Figure 1 – A : EIC of peptides from 1 ng standard tryptic mixture chromatographed using the 75 µm x 15 cm
Accucore™ solid core particle packed column set-up. B : EIC of HSSPGTVAAR with peak widths at 10 % and 50 %
peak height and overall peak capacity nC, average peak width ( 𝑤0.5̅̅ ̅̅ ̅ ) and average asymmetry factor (𝐴𝑆̅̅ ̅).
Supplementary Figures 1 – 4 demonstrate each columns’ response in peak capacity and
asymmetry to flow-rate, showing that the silica monolith could be operated in a larger span
of flow-rates without loss in chromatographic performance (i.e. peak capacity of 130 +/- 5
from 200 – 500 nL/min), in contrast to the other columns which had more a more narrow
range of optimal flow-rates. The optimal linear gradient composition was the same (36 %
solvent B at tG ≈ 30 minutes) for the columns with C18 stationary phases. For the less
hydrophobic PS-DVB column, the end gradient composition needed to be significantly lower
10
(20 % solvent B at tG) to ensure tfinal ≈ tG. With the flow-rates investigated, no major effects
on signal intensity were observed (2-fold changes or less, Supplementary Figures 5 and 6).
The correlation between concentration (1 – 0.001 ng/µL) and peak area (not corrected with
an internal standard (IS), and measured with proteotypic peptides with a range of
hydrophobicity, see below) was R2>0.97, implying that the trapping efficiency of each pre-
column was satisfactory (Supplementary Figures 7-11 and Supplementary Table 3). The
average RSDs for the tR in the ExSMix were 0.3, 0.2, 0.4, and 1.0 % for the Chromolith®,
Accucore™, PepMap™, and PepSwift™, respectively. Carry-over was negligible (< limit of
detection, LOD) for all columns regarding the analytes and concentrations investigated in
this study.
Chromatography of unfractionated cell lysates in targeted MS/MS mode
Using the optimized LC conditions and data-dependent MS/MS acquisition, 17 proteotypic
peptides (Table 2) of proteins associated with the beta-catenin degradasome were selected
for further targeted MS/MS studies. Table 2 – Peptide identification for each chromatographic column investigated. 1) is 0.5 ng ExSMix, 2) is 0.5 ng
ExSMix spiked into 1 µg HCT15 tryptic cell lysate and 3) is 1 μg HCT15 tryptic cell lysate (all in triplicates). Green
equals identified, red not identified and grey not observed with ddMSMS. MS/MS extraction was done with 7
ppm mass accuracy. Minimum 3 transitions were required for positive identification and not more than 0.5
minutes shift in tR between 2 and 3 were allowed for positive identification. The effect of sample
complexity-related retention time robustness (important for e.g. MS/MS scheduling/aiding
selectivity) was examined by comparing retention times of the simple ExSMix proteotypic
peptides with that of the ExSMix spiked to tryptic HCT15 cell lysate (serving as a complex
matrix). The median change in retention time of the peptides in the ExSMix and the ExSMix
spiked cell sample was significantly larger for the monoliths (-2.2 and 1.7 min, for the organic
and silica monolith, respectively) compared to the particle packed columns (-0.4 and -1.0
min, for the totally porous and solid core particle packed columns, respectively). The average
retention time RSDs of complex samples were 0.8 (1.9), 0.6 (1.0), 2.6 (2.8), and 1.3 (2.3) %
for the Chromolith®, Accucore™, PepMap™, and PepSwift™, respectively (largest variation in
parenthesis). (Table 2 here) Figure 2 shows extracted ion chromatograms (EIC) of the
representative beta-catenin peptide LLNDEDQVVVNK in each sample chromatographed on
the four column set-ups. The peak widths were not changed as the sample complexity
increased (between ExSMix and ExSMix added to the cell lysate), indicating no sign of
column overload.
11
Table 2 – Peptide identification for each chromatographic column investigated. 1) is 0.5 ng ExSMix, 2) is 0.5 ng
ExSMix spiked into 1 µg HCT15 tryptic cell lysate and 3) is 1 μg HCT15 tryptic cell lysate (all in triplicates).
Green equals identified, red not identified and grey not observed with ddMSMS. MS/MS extraction was done
with 7 ppm mass accuracy. Minimum 3 transitions were required for positive identification and not more than
0.5 minutes shift in tR between 2 and 3 were allowed for positive identification.
Column Chromolith® PepMap™ Accucore™ PepSwift™
Protein Peptide 1 2 3 1 2 3 1 2 3 1 2 3
APC
HSGSYLVTSV
HSSPSGTVAAR
VTPFNYNPSPR
AXIN2
AQSLTLGHFK
ILGKVER
ILGKVERID
Beta-
catenin
LLNDEDQVVVNK
HAVVNLINYQDDAELATR
NEGVATYAAAVLFR
ATVGLIR
GSK3beta
DIKPQNLLLDPDTAVLK
DSSGTGHFTSGVR
LLEYTPTAR
VIGNGSFGVVYQAK
TNKS2
DGGHAGGIFNR
EVSEENHNHANER
SFLQFSAMK
ID count
11 5 6 3 6 5 9 5
Median shift in retention time (min)
between ExSMix and ExSMix in 1 μg
HCT15 cell lysate
-1.7 -0.5 -1.2 -2.2
Average retention time variation (RSD
(%)) in lysates (2 and 3) 0.8 2.6 1.0 2.5
12
Figure 2 – EIC of LLNDEDQVVVNK corresponding to beta-catenin in 0.5 ng ExSMix, 0.5 ng ExSMix spiked in 1 µg
tryptic peptide extract from HCT15, and 1 µg tryptic peptide extract from HCT15 for each of the
chromatographic systems investigated. (Chromatographic conditions and MS-settings as described in Materials
and Methods).
The proteotypic peptides were subsequently searched for in an unspiked cell sample. For
identification minimum three intense/descriptive MS/MS transitions were required. An
additional criterion was that the retention time variation was maximum ±0.5 minutes
relative to that of the ExSMix in spiked sample (matrix matching, see Figure 3) to ensure very
confident identification.
Figure 3 – EIC of the high-abundance peptide (HAVVNLINYQDDAELATR – beta-catenin), with transitions from
m/z = 1021.51898 to 1) m/z = 775.3958, 2) m/z = 1181.545, 3) m/z = 1295.588 in spiked and unspiked tryptic
digested HCT15 extract. Grey background equals positive identification. Samples were chromatographed on the
Accucore™ column set-up with parameters as described in Materials and Methods.
13
Beta-catenin (downstream target of Wnt/signaling) and GSK3beta (a kinase crucial for N-
terminal phosphorylation of beta-catenin that leads to its degradation) (~60 ng/μg and ~0.8
ng/μg sample), respectively) were clearly identifiable with all the columns with the above-
mentioned criteria. With the criteria employed the two other trace proteins (AXIN2, TNKS2)
were however not identified. Notably, using just two intense/descriptive MS/MS transitions
would cause false positives in our assay; For example, the C-terminal HSSPSGTVAAR peptide
of intact APC (not present in HCT15 colon cancer cells, due to a premature stop codon (32))
was falsely identified with otherwise same criteria (data not shown).
Quantification with isotopically labeled internal standard – impact on LC-MS/MS
performance
For quantification in complex samples, isotopically labeled ISs are often added, either as
peptides or proteins (e.g. AQUA peptides (33) or stable isotope labeling by amino acids in
cell culture (SILAC) (25)). In this study, SILAC was used to label two cell lines which were
pooled and used as an internal standard protein solution (ISprot) in following experiments
(preparation and workflow described in Materials and Methods and Supplementary Figure
12). In contrast to e.g. spiking with single IS peptides, the labeled mix is a “universal” IS, also
provides correction for protein digestion, SPE clean-up and MS-response. Also the approach
is simpler than producing recombinant labeled IS proteins (34). The use of complex ISprot
affects the LC-MS/MS performance in two ways; sample load and MS/MS complexity.
In all experiments, the sample load (in mass) was kept below 1 μg. As the ISprot was added to
an approximate ratio of 1:1 to the sample, this means a lower load of endogenous peptides,
making detection of the low-abundant proteins less achievable.
When comparing the columns, the resolution was kept at 140,000 to have the highest
possible selectivity. This setting combined with an injection time of 500 ms lead to a MS
cycling speed of 14.4 seconds. As the MS/MS complexity doubled, the MS/MS was set to
fragment both labeled and non-labeled peptides simultaneously (duplexing). Therefore, the
resolution and inject time were lowered to reach a higher scan rate (2.4 seconds cycling
speed) (Supplementary Figure 13A). Using a sample containing ExSMix spiked into the ISprot,
no significant difference (p = 0.6) between quantification based on peak area or fragment
intensity at a given time was observed (Supplementary Figure 13B-D). Hence, fragment
intensity was used to measure response in further quantification experiments. Practically,
this allows reliable quantification with fewer points per peak (i.e below 8 points for correct
area measurement, as recommended in (35)).
To see whether the solid core column would allow detecting changes in proteotypic peptide
levels of beta-catenin and GSK3beta, we selected the colon carcinoma cell lines HCT15,
COLO320 and SW480 cells. Each were treated with a selective tankyrase inhibitor (G007-LK
(23)) and subjected to quantification with LC-MS/MS and WB. Beta-catenin (the subject of
the AXIN2/tankyrase2/APC/GSKbeta-containing destruction complex) could be relatively
quantified with excellent precision (RSD 8 %) in these solutions with double complexity
14
(sample +ISprot) well within 20 minutes: Reduction of beta-catenin following treatment with
the selective tankyrase inhibitor G007-LK was observed in SW480 cells and correlated with
results obtained with an established WB protocol (see Figure 4, Supplementary Figure 15 for
raw WB and (36)).
Figure 4 – A : EICs of LLNDEDQVVVNK (blue color, endogenous) and LLNDEDQVVVNk (red color, internal
standard) in DMSO and in 1 µM G007-LK treated SW480 cells, respectively, analyzed by LC-MS/MS on the
Accucore™ column using standard gradient conditions and MS/MS detection in duplexing mode with resolution
at 35,000 and maximum injection time at 100 ms. B – WB of total beta-catenin and actin in DMSO and 1 µM
G007-LK treated SW480 cells (Procedure in Supplementary Materials and Methods). C – Relative quantification
based on LC-MS analysis and WB analysis (Procedure in Materials and Methods and Supplementary Figure 12).
For levels of GSK3beta and beta-catenin in the other cell lines analyzed, see Supplementary
Figure 14.
Discussion
Chromatographic performance
Using 30 minute solvent gradients (for efficient analysis), the solid core particle packed
nano-column provided the highest peak capacity and hence resolution, with the silica
monolith at a clear 2nd place. Similar results has been observed with larger-bore columns
15
(37), but it was not given that this would be the case for nano-scale columns, as successful
packing/polymerization can be dependent on column diameter. For example, solid core
particles packed in 2.1 mm ID columns have previously not been able to match the efficiency
of comparable 4.6 mm ID columns (38), while monolith columns are easier to prepare in
capillary/nano-format (39).
Insignificant differences were observed between the four columns regarding sensitivity.
Others have shown clear increases in sensitivity with reduced flow-rates in nano-LC (40). In
our hands, with the flow-rates and column dimensions used (200-500 nL/min, 50-100 μm ID),
the signal intensity was in effect not dramatically affected, which may be due to recent
improvements in the needle geometries and interface constructions (9, 41).
The column´s retention time robustness (of importance for strict identification and MS/MS
scheduling) differed significantly, with the particle packed columns having the least retention
time shift as function of sample complexity, with the core shell column having the best
within-complex sample repeatability. The retention time robustness of the solid core particle
packed column was somewhat surprising considering its lower surface. We believe this
factor was less important due to the use of a full porous particle packed pre-column. In
summary: there are considerable differences in the performance of commercial nano LC
columns, and the solid core particle packed column was considered to best within the
framework of this study (which was meant to reflect targeted proteomics applications of
complex samples in general).The other morphologies may of course excel under other
circumstances. For example, the PS-DVB material is very suitable for larger molecules, e.g.
intact proteins (42) or perhaps larger peptides from lys C protease. In addition, the PS-DVB
phase is excellent for very narrow nano-LC (open tubular format), with performance
matching or surpassing the columns investigated here (43, 44). Open tubular LC columns are
however not yet commercially available.
Identification and quantification in complex samples
Targeted LC-MS/MS is an alternative to common approaches such as WB and enzyme-linked
immunosorbant assays (ELISA) (2) , which are far more used in e.g. clinical settings. It is
claimed that selectivity is higher with LC-MS/MS-based measurements, but this relies on
three major factors; 1) high retention time repeatability, 2) use of proper internal standards
and 3) good MS/MS-spectral quality (45, 46). Our results show that 1) is obtainable with
nano LC columns (and not just larger ID columns), but differs between products. Regarding
2), using a SILAC mixture as a “universal” internal standard is advantageous as it corrects for
variations in all steps of sample treatment (e.g. trypsination and ion suppression). Although
this is a relatively simple approach, it adds to complexity. As a result, faster scan times were
necessary, requiring a reduction of MS resolution. This points to 1) being of added
importance, as retention time repeatability thus becomes increasingly important for
compound identification (to compensate for lowered mass accuracy). Regarding 3), it was
clear that using just two MS/MS transitions to attempt identifications could result in false
16
positives. Hence, we believe that three MS/MS transitions is an absolute minimum for
identification, which also has been used in the study by Abbatiello et al. (47).
Beta-catenin and GSK3beta could be quantified with corresponding ISprot normalization, with
variation less than 10 % (well within the FDA guidelines limit of 20 % RSD (48)), which points
to nano LC-MS/MS as a potentially high precision tool for fast analysis of complex samples.
However, the difficulty of measuring lower abundant analytes call for increased sensitivity.
One potential path is further reducing the ID of the nano LC column. For example, 10 μm ID
open tubular columns have recently been demonstrated to identify axin1, in a hyphenated
set-up including on-line enzymatic digestion (49). However, such equipment is not
commercially available, and is rather complex. Perhaps compromises would be possible for
commercial implementation, e.g. 30 μm packed columns (50), and use of new core-shell
particles (http://www.sciencedirect.com/science/article/pii/S0021967314007304).
Several other nano-LC columns are available, and due to practical reasons all of these were
not investigated in this study. However, our results may also serve as a source for
comparison to other columns in “real sample” analysis (and not just assessment with clean
standards).
References
1. Michalski A, Cox J, Mann M. More than 100,000 detectable peptide species elute in single shotgun proteomics runs but the majority is inaccessible to data-dependent LC-MS/MS. J Proteome Res. 2011;10(4):1785-93. 2. Picotti P, Aebersold R. Selected reaction monitoring-based proteomics: workflows, potential, pitfalls and future directions. Nat Meth. 2012;9(6):555-66. 3. Liebler DC, Zimmerman LJ. Targeted Quantitation of Proteins by Mass Spectrometry. Biochemistry. 2013;52(22):3797-806. 4. Chen Y, Gruidl M, Remily-Wood E, Liu RZ, Eschrich S, Lloyd M, et al. Quantification of beta-catenin signaling components in colon cancer cell lines, tissue sections, and microdissected tumor cells using reaction monitoring mass spectrometry. J Proteome Res. 2010;9(8):4215-27. 5. Neue UD. Theory of peak capacity in gradient elution. J Chromatogr A. 2005;1079(1–2):153-61. 6. Wang X, Stoll DR, Schellinger AP, Carr PW. Peak Capacity Optimization of Peptide Separations in Reversed-Phase Gradient Elution Chromatography: Fixed Column Format. Anal Chem. 2006;78(10):3406-16. 7. Kocher T, Swart R, Mechtler K. Ultra-high-pressure RPLC hyphenated to an LTQ-Orbitrap Velos reveals a linear relation between peak capacity and number of identified peptides. Anal Chem. 2011;83(7):2699-704. 8. Hsieh EJ, Bereman MS, Durand S, Valaskovic GA, MacCoss MJ. Effects of Column and Gradient Lengths on Peak Capacity and Peptide Identification in nanoflow LC-MS/MS of Complex Proteomic Samples. J Am Soc Mass Spectrom. 2013;24(1):148-53. 9. Reschke BR, Timperman AT. A study of electrospray ionization emitters with differing geometries with respect to flow rate and electrospray voltage. J Am Soc Mass Spectrom. 2011;22(12):2115-24. 10. Schmidt A, Karas M, Dulcks T. Effect of different solution flow rates on analyte ion signals in nano-ESI MS, or: when does ESI turn into nano-ESI? J Am Soc Mass Spectrom. 2003;14(5):492-500. 11. Marchetti N, Cavazzini A, Gritti F, Guiochon G. Gradient elution separation and peak capacity of columns packed with porous shell particles. J Chromatogr A. 2007;1163(1–2):203-11. 12. Unger KK, Skudas R, Schulte MM. Particle packed columns and monolithic columns in high-performance liquid chromatography-comparison and critical appraisal. J Chromatogr A. 2008;1184(1–2):393-415. 13. Eeltink S, Dolman S, Detobel F, Swart R, Ursem M, Schoenmakers PJ. High-efficiency liquid chromatography–mass spectrometry separations with 50 mm, 250 mm, and 1 m long polymer-based
17
monolithic capillary columns for the characterization of complex proteolytic digests. J Chromatogr A. 2010;1217(43):6610-5. 14. Novakova L, Vaast A, Stassen C, Broeckhoven K, De Pra M, Swart R, et al. High-resolution peptide separations using nano-LC at ultra-high pressure. J Sep Sci. 2013;36(7):1192-9. 15. Fekete S, Fekete J. The Potential of Shell Particles in Fast Liquid Chromatography. Ultra-High Performance Liquid Chromatography and its Applications: John Wiley & Sons, Inc.; 2013. p. 133-67. 16. Rozenbrand J, van Bennekom WP. Silica-based and organic monolithic capillary columns for LC: Recent trends in proteomics. J Sep Sci. 2011;34(16-17):1934-44. 17. Dolman S, Eeltink S, Vaast A, Pelzing M. Investigation of carryover of peptides in nano-liquid chromatography/mass spectrometry using packed and monolithic capillary columns. J Chromatogr B Analyt Technol Biomed Life Sci. 2013;912:56-63. 18. Guiochon G. Monolithic columns in high-performance liquid chromatography. J Chromatogr A. 2007;1168(1-2):101-68 19. Nunez O, Nakanishi K, Tanaka N. Preparation of monolithic silica columns for high-performance liquid chromatography. J Chromatogr A. 2008;1191(1-2):231-52. 20. Zhou F, Lu Y, Ficarro SB, Webber JT, Marto JA. Nanoflow Low Pressure High Peak Capacity Single Dimension LC-MS/MS Platform for High-Throughput, In-Depth Analysis of Mammalian Proteomes. Anal Chem. 2012;84(11):5133-9. 21. Rogeberg M, Wilson SR, Malerod H, Lundanes E, Tanaka N, Greibrokk T. High efficiency, high temperature separations on silica based monolithic columns. J Chromatogr A. 2011;1218(41):7281-8. 22. MacDonald BT, Tamai K, He X. Wnt/beta-catenin signaling: components, mechanisms, and diseases. Dev Cell. 2009;17(1):9-26. 23. Voronkov A, Holsworth DD, Waaler J, Wilson SR, Ekblad B, Perdreau-Dahl H, et al. Structural basis and SAR for G007-LK, a lead stage 1,2,4-triazole based specific tankyrase 1/2 inhibitor. J Med Chem. 2013;56(7):3012-23. 24. Wisniewski JR, Zougman A, Nagaraj N, Mann M. Universal sample preparation method for proteome analysis. Nat Meth. 2009;6(5):359-62. 25. Ong SE, Mann M. A practical recipe for stable isotope labeling by amino acids in cell culture (SILAC). Nat Protoc. 2006;1(6):2650-60. 26. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. 27. Consortium TU. Activities at the Universal Protein Resource (UniProt). Nucleic Acids Res. 2014;42(D1):D191-D8. 28. Wang M, Weiss M, Simonovic M, Haertinger G, Schrimpf SP, Hengartner MO, et al. PaxDb, a Database of Protein Abundance Averages Across All Three Domains of Life. Mol Cell Proteomics. 2012;11(8):492-500. 29. Carr SA, Abbatiello SE, Ackermann BL, Borchers C, Domon B, Deutsch EW, et al. Targeted Peptide Measurements in Biology and Medicine: Best Practices for Mass Spectrometry-based Assay Development Using a Fit-for-Purpose Approach. Mol Cell Proteomics. 2014;13(3):907-17. 30. Peterson AC, Russell JD, Bailey DJ, Westphall MS, Coon JJ. Parallel reaction monitoring for high resolution and high mass accuracy quantitative, targeted proteomics. Mol Cell Proteomics. 2012;11(11):1475-88. 31. Sun L, Zhu G, Dovichi NJ. Comparison of the LTQ-Orbitrap Velos and the Q-Exactive for proteomic analysis of 1-1000 ng RAW 264.7 cell lysate digests. Rapid Commun Mass Spectrom. 2013;27(1):157-62. 32. Ikediobi ON, Davies H, Bignell G, Edkins S, Stevens C, O'Meara S, et al. Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Mol Cancer Ther. 2006;5(11):2606-12. 33. Gerber SA, Rush J, Stemman O, Kirschner MW, Gygi SP. Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. PNAS. 2003;100(12):6940-5. 34. Edfors F, Bostrom T, Forsstrom B, Zeiler M, Johansson H, Lundberg E, et al. Immunoproteomics using polyclonal antibodies and stable isotope-labeled affinity-purified recombinant proteins. Mol Cell Proteomics. 2014;13(6):1611-24. 35. Lange V, Picotti P, Domon B, Aebersold R. Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol. 2008;4:222-35. 36. Lau T, Chan E, Callow M, Waaler J, Boggs J, Blake RA, et al. A Novel Tankyrase Small-Molecule Inhibitor Suppresses APC Mutation–Driven Colorectal Tumor Growth. Cancer Res. 2013;73(10):3132-44.
18
37. Beneito-Cambra M, Herrero-Martínez JM, Ramis-Ramos G, Lindner W, Lämmerhofer M. Comparison of monolithic and microparticulate columns for reversed-phase liquid chromatography of tryptic digests of industrial enzymes in cleaning products. J Chromatogr A. 2011;1218(41):7275-80. 38. Gritti F, Guiochon G. Repeatability of the efficiency of columns packed with sub-3μm core-shell particles: Part II. 2.7 μm Halo-ES-Peptide-C18 particles in 4.6mm and 2.1mm×100mm column formats. J Chromatogr A. 2012;1252(0):31-44. 39. Ishizuka N, Kobayashi H, Minakuchi H, Nakanishi K, Hirao K, Hosoya K, et al. Monolithic silica columns for high-efficiency separations by high-performance liquid chromatography. J Chromatogr A. 2002;960(1–2):85-96. 40. Shen Y, Zhao R, Berger SJ, Anderson GA, Rodriguez N, Smith RD. High-Efficiency Nanoscale Liquid Chromatography Coupled On-Line with Mass Spectrometry Using Nanoelectrospray Ionization for Proteomics. Anal Chem. 2002;74(16):4235-49. 41. Liu Q, Cobb JS, Johnson JL, Wang Q, Agar JN. Performance comparisons of nano-LC systems, electrospray sources and LC-MS-MS platforms. J Chromatogr Sci. 2014;52(2):120-7. 42. Rogeberg M, Wilson SR, Greibrokk T, Lundanes E. Separation of intact proteins on porous layer open tubular (PLOT) columns. J Chromatogr A. 2010;1217(17):2782-6. 43. Rogeberg M, Vehus T, Grutle L, Greibrokk T, Wilson SR, Lundanes E. Separation optimization of long porous-layer open-tubular columns for nano-LC-MS of limited proteomic samples. J Sep Sci. 2013;36(17):2838-47. 44. Luo Q, Yue G, Valaskovic GA, Gu Y, Wu SL, Karger BL. On-line 1D and 2D porous layer open tubular/LC-ESI-MS using 10-microm-i.d. poly(styrene-divinylbenzene) columns for ultrasensitive proteomic analysis. Anal Chem. 2007;79(16):6174-81. 45. van den Broek I, Romijn FPHTM, Smit NPM, van der Laarse A, Drijfhout JW, van der Burgt YEM, et al. Quantifying Protein Measurands by Peptide Measurements: Where Do Errors Arise? J Proteome Res. 2014. 46. Gallien S, Duriez E, Demeure K, Domon B. Selectivity of LC-MS/MS analysis: Implication for proteomics experiments. J Proteomics. 2013;81(0):148-58. 47. Abbatiello SE, Mani DR, Schilling B, Maclean B, Zimmerman LJ, Feng X, et al. Design, implementation and multisite evaluation of a system suitability protocol for the quantitative assessment of instrument performance in liquid chromatography-multiple reaction monitoring-MS (LC-MRM-MS). Mol Cell Proteomics. 2013;12(9):2623-39. 48. Administration Fad. Guidance for Industry Bioanalytical Method Validation. Rockville, MD: US Deparment of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Re. 2001:25. 49. Hustoft HK, Vehus T, Brandtzaeg OK, Krauss S, Greibrokk T, Wilson SR, et al. Open Tubular Lab-On-Column/Mass Spectrometry for Targeted Proteomics of Nanogram Sample Amounts. PLoS ONE. 2014;9(9):e106881. 50. Kocher T, Pichler P, Pra MD, Rieux L, Swart R, Mechtler K. Development and performance evaluation of an ultra-low flow nano liquid chromatography-tandem mass spectrometry set-up. Proteomics. 2014.