A comparison of labeling and label-free mass spectrometry-based proteomics approaches.

Abstract

The proteome of the recently discovered bacterium Methylocella silvestris has been characterized using three profiling and comparative proteomics approaches. The organism has been grown on two different substrates enabling variations in protein expression to be identified. The results obtained using the experimental approaches have been compared with respect to number of proteins identified, confidence in identification, sequence coverage and agreement of regulated proteins. The sample preparation, instrumental time and sample loading requirements of the differing experiments are compared and discussed. A preliminary screen of the protein regulation results for biological significance has also been performed.

Full text

Page 1

Subscriber access provided by WATERS CORP
Journal of Proteome Research is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.
Technical Note
A comparison of labelling and label-free mass
spectrometry-based proteomics approaches.
Vibhuti Jitendra Patel, Konstantinos Thalassinos, Susan Slade, Joanne
B Connolly, Andrew Crombie, J. Colin Murrell, and James Scrivens
J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/pr900080y • Publication Date (Web): 12 May 2009
Downloaded from http://pubs.acs.org on May 19, 2009
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Page 2

1
A comparison of labelling and label-free mass spectrometry-based proteomics approaches
Vibhuti J. Patel†§, Konstantinos Thalassinos†§, Susan E. Slade†, Joanne B. Connolly‡, Andrew
Crombie†, J. Colin Murrell†, James H. Scrivens†*
Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK,
Waters Corporation, Atlas Park, Simonsway, Manchester M22 5PP, UK

§ Both authors contributed equally to this work.
*To whom correspondence should be addressed. Tel: +44(0)24-7657-4189. Fax: +44(0)24-7652-3568
E-mail: j.h.scrivens@warwick.ac.uk
† Department of Biological Sciences, University of Warwick, Coventry CV4 7AL, UK
‡ Waters Corporation, Atlas Park, Simonsway, Manchester M22 5PP, UK

The proteome of the recently discovered bacterium Methylocella silvestris has been characterised using
three profiling and comparative proteomics approaches. The organism has been grown on two different
substrates enabling variations in protein expression to be identified. The results obtained using the
experimental approaches have been compared with respect to number of proteins identified, confidence
in identification, sequence coverage and agreement of regulated proteins. The sample preparation,
instrumental time and sample loading requirements of the differing experiments are compared and
discussed. A preliminary screen of the protein regulation results for biological significance has also
been performed.
Keywords: proteomics, quantification, iTRAQ, label-free, methanotroph, mass spectrometry
Page 1 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 3

2
Introduction
Since the mid-1990s, mass spectrometry-based strategies have been the mainstream method for protein
identification [1]. There remain, however, a number of issues to be tackled. Intrinsic characteristics of
proteomes raise a number of experimental challenges. By nature, proteomes are large and complex. A
single gene can often give rise to multiple, distinct proteins due to alternative splicing, sequence
polymorphisms and post-translational modifications. Protein databases generated from the genome of an
organism may, therefore, not be a true reflection of the potential protein complement [2]. There has
been significant progress in the development of new approaches to tackle these issues, but technical
challenges persist.
An ideal approach would enable the comprehensive characterisation of proteomes in a high-throughput
manner. Currently, the techniques involved can be complex, costly and involving time-consuming data
analysis. A low number of replicate experiments conducted – often due to a lack of sample availability
– means that reproducibility is a concern. In addition, any given technique may only yield information
on a fraction of the relevant peptides in any single analytical run [3].
An established proteomics approach is based on the separation of proteins via one- or two-dimensional
polyacrylamide gel electrophoresis (PAGE). Proteins are digested within the gel, and the resulting
peptides extracted for MS analysis. Drawbacks associated with PAGE include dynamic range,
insufficient resolving power to fully separate all proteins within a sample, and restricted sample
throughput [4].
Non gel-based techniques have been developed for the analysis of complex proteomic samples: so-
called ‘shotgun’ experiments, where a whole proteome is digested without prior protein separation.
Typically, the resulting peptides are separated by strong cation exchange chromatography (SCX) before
reversed-phase LC-MS/MS analysis [5], an example of an approach known as multi-dimensional protein
identification techniques (MudPIT). This method has been shown to provide increased proteome
Page 2 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 4

3
coverage compared to gels, although it still suffers from problems with reproducibility and dynamic
range. This approach has gained popularity within proteomic studies in preference to gels [2].
In addition to providing a profile of what proteins are present within a system at a given time,
information on the expression levels of these proteins is increasingly required. Techniques in
comparative and quantitative proteomics have, therefore, also developed significantly in recent years.
Relative quantification can be performed on proteins separated by two-dimensional PAGE, using image
analysis software, sometimes incorporating selective labelling approaches such as difference gel
electrophoresis (DiGE) [6]. This approach is subject to the restrictions imposed by the gel methods.
A number of labelling approaches can also be incorporated into ‘shotgun’ type experiments. These
include stable isotope labelling by amino acids in cell culture (SILAC) [7], isotope dilution [8], stable
isotope labelled peptides [9], radiolabelled amino acid incorporation [10], chemically synthesised
peptide standards [11], tandem mass tags (TMT) [12], isotope-coded affinity tags (ICAT) [13], and more
recently, isobaric tags for relative and absolute quantification (iTRAQ) [14]. The iTRAQ system is now
commercially available with eight isobaric tags [15], having only initially been available with four tags,
and has been widely used in proteomic studies [16].
Most label-based quantification approaches have potential limitations: complex sample preparation,
the requirement for increased sample concentration, and incomplete labelling. There has, therefore,
recently been a focus in the area of non-labelled quantification in order to address some of these issues
[17].
Non-labelled techniques which have been developed include peptide match score summation (PMSS)
[18] and spectrum sampling (SpS) [19], both of which can be combined with statistical evaluation to
detect differentially expressed proteins [20]. Another approach utilises a protein abundance index
(PAI), [21] which can be converted to exponentially modified PAI (emPAI) for absolute protein
quantification [22].
Page 3 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 5

4
It has been observed that electrospray ionisation (ESI) provides signal responses that correlate linearly
with increasing concentration [23], but there have been concerns regarding the nonlinearity of signal
response [24]. Previous works have introduced quantitative, label-free LC-MS-based strategies for
global profiling of complex protein mixtures [25] [26]. More recently, a simple LC-MS-based
methodology was published which relies on changes in signal response from each accurate mass
measurement and corresponding retention time (AMRT) to directly reflect concentrations in one sample
relative to another [27], which has since been developed into a label-free system capable of relative and
absolute quantification [28] [29]. All detectable, eluting peptides and their corresponding fragments are
observed via rapid switching between high and low collision energy during the LC-MS/MS experiment,
giving a comprehensive list of all ions that can subsequently be searched [30].
In this work three proteomics approaches have been used to identify and relatively quantify the
proteins within a bacterium when grown under different substrates. Samples have been analysed both
qualitatively and quantitatively by: (i) one-dimensional PAGE, (ii) MudPIT incorporating iTRAQ tags,
and (iii) a data-independent, alternate scanning LC-MS method enabling label-free quantification.
Comparisons have been made regarding experimental considerations such as ease of use, amount of
biological sample required, time required to prepare samples for analysis and total instrument time. The
data obtained have been evaluated with respect to number of protein identifications, confidence of the
assignments, sequence coverage and agreement of regulated proteins. All approaches have been carried
out using equivalent instrumentation, enabling the results to be more directly compared. The organism
used in this study is the methanotrophic bacterium Methylocella silvestris, an environmentally important
organism involved in global methane cycling. Unlike other methanotrophs, M. silvestris is able to grow
on multi-carbon compounds as well as on methane [31]. In this work, cultures of M. silvestris were
grown on acetate and methane.

Materials and methods
Page 4 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 6

5
Bacterial growth and sample preparation
Methylocella silvestris was grown in fermenter cultures on diluted nitrate mineral salts (NMS)
medium with methane or acetate (5 mM) as previously described [31]. Cells, grown to late exponential
phase (OD540 ~1.0), were harvested by centrifugation (17,700 × g, 20 min, 4°C), washed in growth
medium, resuspended in 0.1 M PIPES buffer (piperazine-N,N′-bis[2-ethanesulfonic acid], pH 7.0), and
frozen in liquid nitrogen. Subsequently, frozen cells were thawed and resuspended in PIPES buffer
containing 1 mM benzamidine and broken by four passes through a French pressure cell at 125 MPa
(4°C) (American Instrument Co., Silver Spring, MD). Cell debris and membranes were removed by two
centrifugation steps (13,000 × g, 30 min, 4°C, followed by 140,000 × g, 90 min, 4°C ), and the
supernatant, containing soluble cytoplasmic proteins, used for analysis. A protein assay was conducted
on the soluble extract, using a Micro BCA Protein Assay Kit (Pierce Protein Research Products, Thermo
Scientific, Cramlington, UK) according to the manufacturer's protocol.

Protein separation by gel electrophoresis
Proteins were resolved by 1D SDS-PAGE (14 µg per lane) and stained with Coomassie Blue. 30 to 40
slices were excised from each lane, and subjected to tryptic digestion. All processing of the gel plugs
was performed by a MassPrep robotic protein handling system (Waters Corporation, Manchester, UK)
using the manufacturer’s protocol. In brief, the gel plugs were destained, the disulfide bonds were
reduced by the addition of dithiothreitol and the free cysteine residues were alkylated with
iodoacetamide. The gel plugs were washed prior to a dehydration step, followed by the addition of
trypsin (Promega, Southampton, UK), and incubated for 4.5 h. The resultant tryptic peptides were
extracted twice and transferred to a cooled 96-well microtitre plate; if necessary, they were stored at –20
°C.

Page 5 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 7

6
iTRAQ labelling and strong cation exchange chromatography (SCX)
Labelled quantification was carried out using the iTRAQ 4-plex labelling kit (Applied Biosystems,
Warrington, UK). Protein extracts from the two growth conditions were digested and labelled according
to the manufacturer’s standard protocol, and the samples pooled and lyophilised. A total of 400 µg
protein from each growth condition was labelled, giving a total protein loading of 800 µg. As SCX was
carried out offline, the potential for sample losses is higher. A larger initial protein loading was
therefore used in order to minimise such losses and optimise the number of proteins identified by this
approach. 200 µg of acetate-grown sample was labelled with the 114 reporter tag, and 200 µg with the
116 reporter tag. 200 µg of the methane-grown sample was labelled with the 115 tag and 200 µg with
the 117 tag. As per the manufacturer’s protocol, a maximum of 100 µg of protein was labelled per vial
of iTRAQ label, i.e. two vials were used per label. The labelling of one growth condition with two
different iTRAQ tags provides the means for an internal control to monitor labelling efficiency. The
labelled tryptic peptides were partially resolved using a PolySULFOETHYL A SCX column, 2.1 mm ×
20 cm, 5 µm particles, 300 Å pore size (PolyLC, Columbia, USA), using a stepwise gradient of KCl,
adapted from Link et al. [32], from 2.5–50% salt solution over a period of 75 minutes. In total, 64
fractions were collected.

In-solution tryptic digestion
100 µg of soluble protein extract was resuspended in 1 mL of 0.1% Rapigest (Waters Corporation,
Milford, MA) and concentrated using a 5 kDa cut-off spin column. The solution was then heated at
80°C for 15 minutes, reduced with DTT at 60°C for 15 minutes, alkylated in the dark with
iodoacetamide at ambient temperature for 30 minutes, and digested with 1:50 (w/w) sequencing grade
trypsin (Promega, Southampton, UK) for 16 hours. RapiGest was hydrolysed by the addition of 2 µL 15
M HCl, centrifuged, and each sample diluted 1:1 with a 50 fmol/µl glycogen phosphorylase B standard
Page 6 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 8

7
tryptic digest to give a final protein concentration of 500 ng/µl per sample and 25 fmol/µl phosphorylase
B.

LC-MS/MS acquisition for gel-separated samples
Peptides extracted from the digested gel were transferred to a nanoACQUITY system (Waters
Corporation). A 6.4 µl aliquot of extract was mixed with 13.6 µl of 0.1% formic acid and loaded onto a
0.5 cm LC Packings C18 5 µm 100Å 300 µm i.d µ-precolumn cartridge. Flushing the column with
0.1% formic acid desalted the bound peptides before a linear gradient of solvent B (0.1% formic acid in
acetonitrile) at a flow rate of approximately 200 nl/min eluted the peptides for further resolution on a 15
cm LC Packings C18 5 µm 5Å 75 µm i.d. PepMap analytical column. The eluted peptides were
analysed on a Micromass Q-Tof Global Ultima (Waters Corporation) mass spectrometer fitted with a
nano-LC sprayer with an applied capillary voltage of 3.5 kV. The spectral acquisition scan rate was 1.0
s with a 0.1 s interscan delay. The instrument was calibrated against a collisionally induced dissociation
(CID) spectrum of the doubly charged precursor ion of [Glu1]-fibrinopeptide B (GFP – Sigma Aldrich,
St. Louis, USA), and fitted with a GFP lockspray line. The instrument was operated in data dependent
acquisition (DDA) mode over the mass/charge (m/z) range of 50-2000. During the DDA analysis, CID
experiments were performed on the three most intense, multiply charged peptides as they eluted from
the column at any given time. Once these data have been collected, the next three most intense peptides
are selected, and this process repeated.

LC-MS/MS acquisition for iTRAQ samples
Fractions collected from the SCX separation of iTRAQ-labelled peptides were snap-frozen on dry ice
and lyophilised to dryness. The samples were resuspended in 20 µl 0.1% formic acid and transferred to
a CapLC system (Waters Corporation). A 6.4 µl aliquot of extract was mixed with 13.6 µl of 0.1%
Page 7 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 9

8
formic acid and loaded onto a 0.5 cm LC Packings C18 5 µm 100Å 300 µm i.d precolumn cartridge.
Flushing the column with 0.1% formic acid desalted the bound peptides before a linear gradient of
solvent B (0.1% formic acid in acetonitrile) at a flow rate of approximately 200 nl/min eluted the
peptides for further resolution on a 15 cm LC Packings C18 5 µm 5Å 75 µm i.d. PepMap analytical
column. The eluted peptides were analysed on a Micromass Q-Tof Global Ultima (Waters Corporation)
mass spectrometer fitted with a nano-LC sprayer with an applied capillary voltage of 3.5 kV. The
spectral acquisition scan rate was 1.0 s with a 0.1 s interscan delay. The instrument was calibrated
against a CID spectrum of the doubly charged precursor ion of GFP, and fitted with a GFP lockspray
line. The instrument was operated in data dependent acquisition (DDA) mode as described above.

LC-MS configurations for label-free analysis
Nanoscale LC separations of tryptic peptides for qualitative and quantitative multiplexed LC-MS
analysis were performed with a nanoACQUITY system (Waters Corporation) using a Symmetry C18
trapping column (180 µm x 20 mm 5 µm) and a BEH C18 analytical column (75 µm x 250 mm 1.7 µm).
The composition of solvent A was 0.1% formic acid in water, and solvent B, 0.1% formic acid in
acetonitrile. Each sample (total protein 0.5 µg) was applied to the trapping column and flushed with 1%
solvent B for 5 minutes at a flow rate of 15 µL/min. Sample elution was performed at a flow rate of 300
nL/min by increasing the organic solvent concentration from 3 to 40% B over 90 min. All analyses were
conducted in triplicate. The precursor ion masses and associated fragment ion spectra of the tryptic
peptides were mass measured with a Q-ToF Premier mass spectrometer (Waters Corporation) directly
coupled to the chromatographic system.
The time-of-flight analyzer of the mass spectrometer was externally calibrated with NaI from m/z 50 to
1990, with the data post-acquisition lockmass-corrected using the monoisotopic mass of the doubly
charged precursor of GFP, fragmented with a collision energy of 25V. The GFP was delivered at 500
Page 8 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 10

9
fmol/µL to the mass spectrometer via a NanoLockSpray interface using the auxiliary pump of a
nanoACQUITY system at a flow rate of 500 nL/min. The reference sprayer was sampled every 60
seconds.
Accurate mass data were collected in data independent mode of acquisition (LC-MSE) by alternating
the energy applied to the collision cell between a low energy and elevated energy state. The spectral
acquisition scan rate was 0.6 s with a 0.1 s interscan delay. In the low energy MS mode, data were
collected at constant collision energy of 4 eV. In elevated energy MS mode, the collision energy was
ramped from 15 eV to 35 eV during each integration.

Data processing for DDA acquisitions
The uninterpreted MS/MS data from the gel-separated and iTRAQ-labelled samples were processed
using ProteinLynx Global Server (PLGS) v2.3. The data were smoothed, background subtracted,
centred and deisotoped. All data were lockspray calibrated against GFP using data collected from the
reference line during acquisition.

Data processing for label-free acquisitions
The LC-MSE data were processed using PLGS v2.3. The ion detection, data clustering and
normalisation of the data independent, alternate scanning LC-MSE data has been explained in detail
elsewhere [33]. In brief, lockmass-corrected spectra are centroided, deisotoped, and charge-state-
reduced to produce a single accurately mass measured monoisotopic mass for each peptide and the
associated fragment ion. The initial correlation of a precursor and a potential fragment ion is achieved
by means of time alignment.

Page 9 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 11

10
Database searches
All data were searched using PLGS v2.3 against a Methylocella silvestris database
(http://genome.ornl.gov/microbial/msil). Fixed modification of carbamidomethyl-C was specified, and
variable modifications included were acetyl N-terminus, deamidation N, deamidation Q and oxidation
M. For the iTRAQ data, variable modifications for the isobaric tags were specified. One missed
cleavage site was allowed. Search parameters specified were a 50 ppm tolerance against the database-
generated theoretical peptide ion masses and a minimum of one matched peptide.
For the LC-MSE data, the time-based correlation applied in data processing was followed by a further
correlation process during the database search that is based on the physicochemical properties of
peptides when they undergo collision induced fragmentation [34]. The precursor and fragment ion
tolerances were determined automatically. The protein identification criteria also included the detection
of at least three fragment ions per peptide, at least one peptide determined per protein and the
identification of the protein in at least two out of three technical replicates. By using protein
identification replication as a filter, the false positive rate is minimised, as false positive protein
identifications, i.e. chemical noise, have a random nature and as such do not tend to replicate across
injections. This approach rules out systematic search events errors due to the repeated ambiguity of a
particular spectrum and the subsequent sequence assignment by a search algorithm, as could be the case
with peptide-centric searches.

Protein quantification using iTRAQ labelling
PLGS was also used for quantitative evaluation of MS/MS data generated from the analysis of the
iTRAQ-labelled peptides. A relative quantification was conducted using a merged dataset comprising
the results from the database search. Concentration ratios of iTRAQ-labelled proteins were calculated
Page 10 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 12

11
based on signal intensities of reporter ions observed in peptide fragmentation spectra, with the relative
areas of the peaks corresponding to proportions of the labelled peptides [14].

Protein quantification using label-free system
Relative quantitative analysis across conditions was performed by comparing normalised peak
area/intensity of each identified peptide. Normalisation of the data was conducted by the use of an
internal protein digest standard. In brief, peak areas/intensities are corrected using those of the internal
protein digest. Intensity measurements are typically further adjusted on those components, that is de-
isotoped and charge-state reduced accurate mass retention time pairs, that replicate throughout the
complete experiment. Next, the redundant, proteotypic quantitative measurements provided by the
multiple tryptic peptide identification from each protein were used to determine an average, relative
protein fold-change. The algorithm performs binary comparisons for each of the conditions to generate
an average normalised intensity ratio for all matched proteins. Proteins with a likelihood of
quantification smaller than 0.05 were considered to be significantly regulated. The entire data set of
differentially expressed proteins was further filtered by considering only the identified peptides that
replicated two out of three technical instrument replicates. A likelihood of regulation higher than 95%,
as reported by the quantification algorithm, was considered.
Page 11 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 13

12
Results and discussion
Protein identifications
Three distinct experimental approaches have been employed in order to provide profiling and
quantitative information regarding the proteome of M. silvestris. The numbers of proteins identified via
each approach are summarised in Table 1. The total number of non-redundant proteins identified, when
single peptide identifications are included, is comparable for all three techniques at 389, 384 and 425
proteins respectively.

Gel-based iTRAQ Label-free
Total number of protein identifications 389 384 425
Single-peptide identifications 154 206 4
Proportion of identifications which were
from a single peptide
40 % 54 % 0.9 %
Identifications with more than one peptide 235 178 421

Table 1: Total protein identifications for the three experimental approaches
Differences arise, however, when looking at the number of peptides per protein identification. There
have been questions raised in the literature regarding the validity of identifications performed using a
single peptide, so-called ‘one-hit wonders’, and whether they should be included in the list of proteins
identified [35]. Of the gel separation identifications, 154 were from a single peptide as are 206 in the
iTRAQ experiment. As a proportion of the total proteins identified, these values are 40% and 54%
respectively. This is typical of many results in the literature [36]. In the label-free results, of the 425
identifications, only 4 are from a single peptide: proportionally less than 1%. As the label-free analysis
is performed in triplicate, only an identification observed in at least two of the three replicates was taken
to be valid; therefore, single-peptide identification in label-free data means that a single peptide was
found in at least two of three data sets. Proteins identified by each experimental setup are listed in
Page 12 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 14

13
Supplementary Data, Tables 1-5. Figure 1 shows the overlap of protein identifications between the
three approaches; 1a uses all data, including single-peptide identifications, 1b illustrates filtered data,
with only identifications obtained with two or more peptides. All proteins identified are listed in
Supplementary Tables 1 to 5, giving information on the molecular weight and pI of the identifications,
and also the number of peptides identified. When including single peptide based identifications, there
are a total of 699 proteins identified. Each of the techniques uniquely provides approximately 17% of
those identifications. The remaining 49% of the identifications overlapped as shown. To overcome the
uncertainty involved in the inclusion of single peptide-based identification, Figure 2b shows the data
presented only including identifications made using a minimum of two peptides. This gives a total of
509 protein identifications, of which 9% were unique to the gel-based approach, 6% to iTRAQ, and
38% to label-free. This shows a significant increase in the proportion of unique identifications by the
label-free method, a reduction in gel-based unique identifications and a considerable decrease in those
uniquely identified by iTRAQ.

Figure 1a: Number of proteins identified by the various experimental approaches, including single-
peptide identifications.
Page 13 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Page 15

14

Figure 1b: Proteins identified by the various experimental approaches, identifications based on a
minimum of two peptides.

A closer inspection of the number of proteins identified with and without the inclusion of single-
peptide identifications reveals some interesting observations. As one would expect, the total number of
proteins identified is lower when single-peptide identifications are excluded (509 when excluded,
compared to 699 when included), including those identifications common to all three methods (89 when
excluded, compared to 152 when included). In contrast, the number of proteins unique to the label-free
method, and those common to the label-free and gel methods, has increased. This is because all but four
of the proteins identified by the label-free method were done so with two or more peptides, whereas the
gel and iTRAQ methods generated a large number of single-peptide identifications. The fact that 152
proteins were independently identified by all three methods provides strong evidence that although some
of these (63 in total) were identified with a single-peptide by one or more technique, they should
possibly not be discarded as false-positive identifications. This raises the questions as to what should be
done with protein identifications based on a single-peptide. While the majority of these are likely to
Page 14 of 32
ACS Paragon Plus Environment
Journal of Proteome Research
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60