Biology Reference
In-Depth Information
plastid development, researchers suggested correcting for the
actually decreased total mass ratio between chloroplast and other
proteins by normalizing exclusively to the TTP of the measured
chloroplast proteome. In other words, these approaches consider
the chloroplast protein dataset as independent from the other pro-
teins and perform normalization only on the set of chloroplast pro-
teins. The argument that, e.g., chloroplast proteins in albino
mutant make up for only a fragment of the total mass compared to
wild-type plastids that contain large amounts of photosynthetic
proteins is valid, but the differences are small. We have calculated
the total mass of plastid proteins in wild type and compared it to
the data obtained from three albino mutants apg1, apg2, and apg3
[
18
]. While 73 % of the total spectra identify plastid proteins in
wild type, this is the case for 54 % of the spectra in apg1, 48 % in
apg2, and 45 % in apg3 (data not shown). Because the calculation
of statistical significance is usually more reliable when fewer data
points are expected to vary, we normally refrain from this type of
normalization. However, we recommend checking the PSM distri-
bution between the organelles for all mutants individually.
3.10 Assigning
Proteins to Organelles
In order to assign proteins to organelles, we recommend using
existing organellar proteome maps for
Arabidopsis thaliana
. These
can be generated conveniently using the SUBA database (
http://
suba.plantenergy.uwa.edu.au/
)
[
19
] and the interested researcher
is provided with a number of different selection criteria for the
localization of a protein to a certain organelle. In case of chloro-
plasts, we have assembled such a reference table based on a number
of different criteria. This table is available from our recent publica-
tions [
5
].
3.11 Statistical
Data Handling
With the above-described normalization procedure, a numerical
value that resembles protein abundance can be assigned to every
identified protein (
see
Note 12
). Proteins with detection gaps in
mutant or wild type or those that were not identified in every rep-
licate receive a zero as abundance value. This way undetected pro-
teins are treated the same way as low-abundance proteins, i.e., they
are represented by a low nSpC value. Because lack of protein detec-
tion in complex mixtures is usually a result of low protein abun-
dance, this procedure reflects the experimental reality. The
numerical values assigned to proteins can now be used for data
evaluation to determine which proteins were up- or down-regulated
in the experiment. This mostly entails the calculation of fold
changes between the protein abundance values in different sam-
ples. If sufficient biological replicates were included in the experi-
ment, a statistical test such as a Student's
t
test (Welch test) or an
ANOVA analysis can be performed as well, followed by applying a
maximum p-value cutoff. We usually reject the null hypothesis (no
change between mutant and wild type) at p-values below 0.05
