Biology Reference
In-Depth Information
Box 1.1 Quantification in MS-Based Proteomics
Mass spectrometric approaches providing relative and absolute
quantification have been a focus of many recent developments
in the field. MS-based quantification is non-trivial because for
different peptide species there is no proportionality between
their respective amounts and the signal intensities they generate
in the mass spectrometer. This is due to the very diverse
chemicophysical properties of peptides with different
sequences, resulting in widely varying ionization efficiencies.
For chemically identical peptides, however, signal intensity is
proportional to the amount
stage of sample preparation. All variations in sample prepara-
tion are then experienced by both samples equally, leading to
very high quantitative accuracy. Chemical methods are usually
applied at a later stage, by which time quantitative differences
due to separate sample preparation may already be established.
Furthermore, care has to be taken that the chemical labeling
procedure proceed to the same degrees of completeness in the
different
samples and that chemical
side reactions are
minimized.
Metabolic methods almost always quantify the peptide in
the intact form in the MS spectrum, whereas some of the
chemical methods use differentially isotope labeled fragment
ions ('reporter ions') to determine the relative ratios from the
MS/MS spectra. A disadvantage of the latter methods is that, in
complex mixtures, peptides apart from the intended one are co-
fragmented. These also contribute their identical reporter ions,
distorting measured ratios
[77]
.
Targeted approaches (SRM or MRM) are also fragmentation-
based quantification methods but they aim to monitor only
transitions from precursor to sequence-specific fragments.
Several such transitions are monitored in rapid succession for
a single peptide and several peptides can be targeted at any
given elution time. This ensures that the recorded signal is due
to the intended peptide. For quantification, an isotope-labeled,
synthetic peptide standard for each peptide of interest needs to
be introduced. However, since synthesis, purification and
storage of many labeled peptides are resource-intensive, the
label-free transition data is often used for approximate quanti-
fication. In general, MRM-based quantification methods
require extensive method development because the most
sensitive and specific transitions need to be determined for each
peptide separately. There are therefore a number of large-scale
projects to construct such data on a global, organism-wide
scale
[187
e
within the linear range of the
instrument
and this is the basis of all isotope labeling methods
as well as of many label-free quantification approaches. In
addition, it is often assumed that the most readily detected
peptides of each protein have roughly similar ionization effi-
ciencies across all proteins, and that their signal is therefore
a proxy for the protein amount.
Label-free approaches are appealing because they can be
used on any sample and do not require any additional experi-
mental steps. A basic version of label-free quantification is
called spectral counting, and simply compares the number of
times a peptide has been fragmented. Since there is a stochastic
tendency of shotgun proteomics to fragment more abundant
peptides more often, this provides a rough measure of peptide
abundance
[180,181]
. In a more accurate version of label-free
quantification, the MS signals of each peptide identifying
a protein are added and this protein intensity value is compared
between the different experiments
[75]
. Ideally, the intensities
of the same peptide species are directly compared across
experiment. Challenges in label-free workflows are day-to-day
variations in instrument performance or slight variations in
sample preparation, which can reduce accuracy. Nevertheless,
they are gaining ground owing to the increasing availability of
high-resolution mass spectrometers and the development of
sophisticated algorithms. They are best suited to cases where at
least several-fold changes in protein or peptide intensities are
expected.
The most accurate methods for quantification by MS make
use of the fact that the MS response of the same compound in
different isotopic states is the same. This principle has been
employed for decades in the small molecule field, where it is
sometimes called isotope dilution MS, and it has also been used
for many years with peptides. In proteomics, the peptide pop-
ulations from two different samples are labeled by the intro-
duction of light or heavy stable isotopes such as
12
C vs.
13
C and
14
N vs.
15
N, mixed and analyzed together. The mass spec-
trometer easily distinguishes heavy peptides from light peptides
by their mass shift, but since they are chemically equivalent
they behave the same during chromatographic separation and
ionization. The ratio of the heavy and light peak intensities
therefore represents the relative amounts of the corresponding
proteins in the samples to be compared. There are many
different methods of introducing labels, for example metabolic
labeling methods such as SILAC
[182]
, or chemical ones such
as TMT
[183]
, iTRAQ
[184]
and di-methyl labeling
[185,186]
.
The metabolic methods have the principal advantage that the
two populations to be compared can be mixed at a very early
e
189]
.
Apart from relative quantification of two or more proteomes,
it is in many instances necessary to estimate the absolute
amounts of proteins. If a known amount of a synthetic, labeled
peptide is added to the sample, the ratio of the heavy to the light
version of the peptide immediately yields the absolute amount
of the endogenous peptide present (absolute quantification or
'AQUA' method
[190]
). If the extraction and digestion effi-
ciency of the protein in the sample is also known, this further-
more yields the absolute amount of protein in the sample. The
same principle also applies to spiking in known amounts of
proteins, except that this automatically controls for digestion
efficiency, including the tendency of the enzyme to produce
peptides with missed cleavages
[191]
.
Absolute protein amounts can be converted into copy
numbers per cell, an important parameter for modeling.
Evidently, it is impractical to spike in reference peptides or
proteins for an entire proteome. Therefore, in the simplest case,
the MS signals of peptides identifying a given protein are
summed up and divided by the total MS signal of all proteins.
This procedure can be calibrated by the estimated total protein
amount or with the help of reference peptides or proteins for
a select subset of proteins across the dynamic range
[2,86,91]
.
e
