Biology Reference
In-Depth Information
by amino acids in cell culture (SILAC [
7
]), are often considered to
be more accurate at quantifying protein abundance [
2
]. However,
the high expense, limitation of samples that can be analyzed in a
single experiment, and need for dedicated software are the main
drawbacks of labeling techniques, especially when compared to
label-free quantitation. In addition, reports of underestimation of
fold changes in iTRAQ experiments are not uncommon [
4
], as
well as sequence-dependent O-acylation modification that may
compromise accurate quantification [
8
].
Label-free strategies can be divided into two distinct groups in
bottom-up proteomic experiments: extracted ion current and spec-
tral counting. Spectral counting is a “random sampling” quantita-
tive method based on premise that high-abundance peptides will
likely be more selected for fragmentation and thus result in higher
number of MS/MS acquired spectra [
9
,
10
]. In other words, the
frequency of these MS/MS scans reflects the abundance of this
peptide in a sample (Fig.
1
). Once the data are searched against a
particular database, the protein abundance will be estimated based
on the sum of all MS/MS spectra matching to every peptide
derived from the protein query (i.e., the number of peptide-
spectrum matches (PSMs)). Protein ratios determined by spectral
counting correlate well with other label-free strategies [
11
,
12
].
However, this approach has showed to be weak for low-abundance
proteins, when few spectral counts are acquired for a particular
protein [
9
,
13
] (Table
1
).
Although several concerns have been raised, the main limita-
tions of the spectral counting strategy are the “chimera” MS/MS
spectra and the protein inference problem (i.e., assembling peptide
sequences into proteins). Chimera spectra hamper correct peptide
assignment. They are frequently observed when co-eluted peptides
present similar
m
/
z
, being consequently isolated and fragmented
simultaneously generating a merged fragment spectrum. The pres-
ence of these MS/MS spectra significantly affects protein identifica-
tion and may contribute to the usually low number of spectra that
are successfully matched to a peptide sequence [
14
]. Chimera spec-
tra comprise a significant number of MS/MS spectra acquired in
modern mass spectrometers. Hoopmann and coworkers estimated
that 11 % of all MS/MS spectra acquired in a high-resolution mass
spectrometer are chimera spectra [
15
]. Recently, Houel et al. showed
that chimera spectra acquired in an LTQ-Orbitrap™ mass spectrom-
eter may reach a fraction equal to 50 % in complex samples [
14
].
One way to reduce the peptide co-elution problem is to reduce
sample complexity. Multidimensional protein identification tech-
nology, MudPIT, was presented by Washburn and coworkers,
reporting the identification of 1,484 proteins from
Saccharomyces
cerevisiae
[
23
]. In MudPIT, fractionation is performed with the pro-
teins already digested from a complex sample. Peptides are loaded in
a strong cation exchange (SCX) chromatography column at once.
