Download The use of isotope-coded affinity tags (ICAT)

520
Biochemical Society Transactions (2004) Volume 32, part 3
The use of isotope-coded affinity tags (ICAT) to
study organelle proteomes in Arabidopsis
thaliana
T.P.J. Dunkley*, P. Dupree*1 , R.B. Watson† and K.S. Lilley*
*Department of Biochemistry, Cambridge Centre for Proteomics, University of Cambridge, Building O, Downing Site, Cambridge CB2 1QW, U.K.,
and †Applied Biosystems, Lingley House, 120 Birchwood Boulevard, Birchwood Point, Warrington, Cheshire WA3 7QH, U.K.
Abstract
Organelle proteomics is the analysis of the protein contents of a subcellular compartment. Proteins identified
in subcellular proteomic studies can only be assigned to an organelle if there are no contaminants present in the sample preparation. As a result, the majority of plant organelle proteomic studies have focused
on the chloroplast and mitochondria, which can be isolated relatively easily. However, the isolation of
components of the endomembrane system is far more difficult due to their similar sizes and densities. For
this reason, quantitative proteomics methods are being developed to enable the assignment of proteins to
a specific component of the endomembrane system without the need to obtain pure organelles.
Introduction
Organelle proteomics is the study of the protein complement
of a specific subcellular compartment and involves subcellular
fractionation followed by protein identification, typically by
MS. Within the eukaryotic cell, proteins are spatially organized according to their function, so the assignment of an
uncharacterized protein to an organelle provides the biologist
with an insight into its possible role in the cell [1]. In addition,
the enrichment of an organelle before proteomic analysis
enables the identification of low-abundance proteins that
would not be detected in unfractionated samples.
The biggest challenge in organelle proteomics is the production of pure organelle preparations, which are important
since the presence of contaminating proteins reduces the
confidence with which novel proteins can be assigned to an
organelle of interest. For the plant endomembrane system,
it is technically impossible to obtain absolutely pure preparations of its component organelles. Therefore to assign
specific locations to integral membrane proteins within the
plant endomembrane system, novel methods that are not
dependent on the production of pure organelle preparations
are required. Such a method will be introduced later in this
paper.
Organelle proteomics
In plants, the majority of subcellular proteomic studies
have focused on the mitochondria and chloroplast. This
is due to the relative ease with which highly enriched
preparations of these organelles can be obtained. Peltier
Key words: Golgi, isotope-coded affinity tag (ICAT), membrane protein, organelle, quantitative
proteomics.
Abbreviations used: 2D, two-dimensional; ER, endoplasmic reticulum; ICAT, isotope-coded
affinity tagging; LeuD3, leucine containing three deuteriums; SILAC, stable isotope-labelling with
amino acids in cell culture.
1
To whom correspondence should be addressed (email [email protected]).
C 2004
Biochemical Society
et al. [2], for example, prepared thylakoids from Arabidopsis
leaves and subsequently separated the soluble lumenal
proteins by two-dimensional PAGE. A total of 81 different
proteins were identified by a combination of peptide mass
fingerprinting and peptide sequencing by both MS/MS
(tandem MS) and Edman degradation (see [3] for a review
of MS in proteomics). Of these, 30 lumenal proteins were
identified, as well as 32 proteins that are associated with
the thylakoid membrane on the stromal side. Also, 12
proteins that were of unknown location were identified,
as well as six non-chloroplast contaminants from the ER
(endoplasmic reticulum), mitochondria and cytosol. These
results demonstrate that highly enriched chloroplasts can be
obtained. However, as the isolation of a pure thylakoid lumen
fraction was not achieved, the 12 identified proteins with no
known location could not be localized to the thylakoid lumen
with confidence.
Two-dimensional PAGE has also been used to study the
plant mitochondrial proteome. Kruft et al. [4] and Millar
et al. [5] both purified mitochondria from Arabidopsis darkgrown suspension cell culture. Both projects reported no, or
very little, contamination from other organelles. However,
mitochondria prepared from Arabidopsis leaves and stems
did contain chloroplast proteins, indicating that the success
of a subcellular proteomics experiment is dependent on the
choice of starting material [4].
These studies resulted in the identification of soluble or
membrane-associated proteins, but very few integral membrane proteins. Integral membrane proteins are underrepresented on two-dimensional gels as strong ionic detergents,
which are required for membrane protein solubilization, are
incompatible with protein separation by isoelectric focusing
[6]. Therefore the identification of integral membrane proteins requires the use of non-two-dimensional gel-based
protein separation. Ferro et al. [7] prepared chloroplast
Proteomics of Plant Proteins
envelopes from spinach leaves and utilized the solubility
of hydrophobic proteins in chloroform/methanol mixtures
to enrich for integral membrane proteins. These membrane
proteins were then solubilized with SDS and separated using
conventional one-dimensional SDS/PAGE. Bands were cut
from the gel, their protein contents digested with trypsin and
the resulting peptides were sequenced by MS/MS. Using this
approach, 54 proteins were identified including 21 with at
least four predicted transmembrane spans. Ferro et al. [7]
also studied the Arabidopsis chloroplast envelope proteome
using a combination of chloroform/methanol, high pH and
high salt extraction techniques to enrich for membrane
proteins. The extracted proteins were digested with trypsin
and the resulting peptides were separated and analysed by
reverse-phase capillary LC-MS/MS. Using this approach, 112
proteins were identified, of which 79% were envelope proteins, 15% were stromal or thylakoid and 6% were contaminants from other organelles [8]. Millar and Heazlewood [9]
also identified hydrophobic integral membrane proteins from
mitochondria by using a high pH wash to remove membraneassociated and soluble proteins, followed by SDS/PAGE and
MS/MS.
To date, no comprehensive study of any component of the
plant endomembrane system has been reported, reflecting
the difficulty in purifying the component organelles which
have similar sizes and densities. However, proteomic studies
of a component of the animal endomembrane system, the
Golgi apparatus, have been conducted. Taylor et al. [10,11]
prepared intact Golgi stacks from rat liver and separated
their protein contents by two-dimensional PAGE. Spots were
cut from the gel and after tryptic digestion their protein
contents were analysed by MS/MS. Of the 55 proteins
identified with known subcellular locations, approx. 50%
were Golgi localized. As a result of this, the proteins
of unknown location could not be assigned to the Golgi
apparatus based on this study. In addition, the 2D (twodimensional) PAGE approach used in this study was not
suitable for the identification of the highly hydrophobic
integral membrane components of the Golgi. Bell et al.
[13] addressed this problem by enriching Golgi membrane
proteins, prepared from rat liver, by Triton X-114 phase
partitioning. These membrane proteins were then solubilized
with SDS and separated by SDS/PAGE [12,13]. Using MS
and Edman sequencing, 17 Golgi integral membrane proteins
were identified as well as 24 contaminating proteins, most of
them from the ER and mitochondria.
The large number of non-Golgi proteins identified in the
two Golgi studies discussed highlights the difficulty in isolating a specific component from the endomembrane system.
The components of the endomembrane system have similar
sizes and densities and are therefore difficult to separate by
ultracentrifugation. In addition, the endomembrane system
is dynamic, with membranes and proteins continuously
trafficking between different organelles. Contaminating
proteins are even more of a problem for the identification of
organelle proteins that are present at very low copy numbers.
The identification of low-abundance proteins requires either
development of higher sensitivity MS techniques or the use of
larger amounts of starting material. Either approach will lead
to the identification of low levels of contaminants as well as
the proteins of interest, unless the organelle preparations are
completely pure [14].
Quantitative proteomics for assigning
proteins to organelles
Quantitative proteomics can be used to circumvent the
problem of contamination in assigning proteins to subcellular compartments. Several techniques have been developed
to compare proteomes quantitatively; these include difference
gel electrophoresis (DIGE) [15,16], ICAT (isotope-coded
affinity tagging) [17] and SILAC (stable isotope-labelling
with amino acids in cell culture) [18]. Of these, difference
gel electrophoresis is a 2D-PAGE-based technique and
thus is not suitable for the quantitative comparison of
membrane proteins. ICAT and SILAC, on the other hand,
are not based on 2D-PAGE and rely on MS for protein
quantification. Therefore ICAT and SILAC can be used
to compare membrane proteomes. Foster et al. [19] used
SILAC to identify lipid-raft-localized proteins in HeLa cells,
despite the presence of contaminating proteins in the lipid raft
preparation. SILAC is a quantitative proteomics technique
that involves feeding one set of cultured cells with the essential
amino acid leucine containing three deuteriums (LeuD3) in
place of three hydrogens and a second set of cells with nondeuterated leucine (Leu). After several generations of growth
on LeuD3-containing media, cells contain LeuD3 in place
of Leu in all of their proteins. As a consequence of this,
leucine-containing peptides from the LeuD3 grown cells
are 3 Da heavier than the equivalent peptide from the Leu
cells. MS can then be used to distinguish peptides derived
from the LeuD3 and Leu cells and their relative abundance
can be calculated, enabling the quantitative comparison of
protein levels in the two samples. Foster et al. [19] treated
the Leu grown cells with a drug to disrupt lipid rafts. The
drug-treated Leu cells were then pooled with untreated
LeuD3 cells and lipid rafts were prepared. The proteins in
the lipid raft fraction were then solubilized, digested and
analysed by LC-MS/MS. Peptides with higher abundance in
the untreated LeuD3 cells were assigned as lipid raft proteins
because the drug treatment disrupts rafts and therefore should
result in lower levels of lipid raft proteins and consequently
peptides in the lipid raft enriched fraction. Out of a total of
392 quantifiable proteins identified in the lipid raft fraction,
241 proteins were found to be genuinely lipid-raft localized. The use of a quantitative proteomics technique thus
enabled the localization of proteins to a subcellular domain
despite the presence of contaminating proteins.
We have applied a quantitative proteomics approach to
study the plant endomembrane system. First, membrane
fractions enriched in ER and Golgi were prepared from
Arabidopsis callus. Soluble and membrane associated proteins
were then removed from the membranes using a high pH carbonate wash [20]. To assess the degree of cross-contamination
C 2004
Biochemical Society
521
522
Biochemical Society Transactions (2004) Volume 32, part 3
Figure 1 Standard cleavable ICAT workflow used for the
quantitative comparison of membrane protein levels in an ER-rich
Figure 2 Membrane proteins from an ER-rich and a Golgi-rich
fraction were labelled with light and heavy ICAT respectively and
and a Golgi-rich fraction prepared from Arabidopsis callus
subjected to the standard cleavable ICAT procedure as shown in
Figure 1
MS spectra for peptides derived from two ER proteins and a Golgi protein
are shown. The relative abundance of each protein in the two fractions
can be determined by comparing the intensity of the ICAT light (ER)- and
ICAT heavy (Golgi)-labelled peptides.
between the two organelles, proteomic analysis of the Golgirich fraction was performed. Golgi membrane proteins were
identified, as well as numerous uncharacterized proteins.
However, the presence of contaminating ER proteins
in the Golgi-rich fraction prevented the assignment of
these uncharacterized proteins to the Golgi membrane. To
address this problem, a quantitative comparison of an ERrich fraction and a Golgi-rich fraction was performed to
discriminate between proteins resident in the two organelles,
since Golgi proteins should be more abundant in the Golgirich fraction and vice versa. The quantitative technique
applied centred around the use of the cleavable ICAT
reagents from Applied Biosystems (Warrington, Cheshire,
U.K.). Cleavable ICAT reagents consist of an iodoacetamide
group, which reacts specifically with cysteine thiol functional groups, connected to biotin by a linker that contains
either nine 13 C or nine 12 C. The mass difference between these
two tags is therefore 9 Da. The biotin group enables ICATlabelled peptides to be avidin affinity-purified to remove
unlabelled peptides and hence simplify the peptide mixture.
The cleavable ICAT reagent also contains an acid-cleavable
group, so that the biotin can then be removed by treating
the labelled peptides with trifluoroacetic acid, resulting in
higher quality MS/MS data. In this study, the ER-rich fraction
was labelled with the light version of the cleavable ICAT
reagent and the Golgi-rich fraction with the heavy 13 C
version. The two samples were then pooled and digested.
The resulting peptides were fractionated by cation-exchange
chromatography and the ICAT-labelled cysteine-containing
peptides were isolated by running each fraction through an
avidin affinity column. After removal of the biotin group
C 2004
Biochemical Society
with trifluoroacetic acid, the peptides were analysed by LCMS/MS (see Figure 1). Peptides derived from the ER-rich
and Golgi-rich fractions were separated by 9 Da due to the
mass difference between the heavy and light tags. The relative
abundance of a protein in the two fractions could thus be
determined by comparing the intensities of the ICAT lightand ICAT heavy-labelled versions of each peptide. Spectra
obtained for peptides from two known ER proteins and a
Golgi protein are shown in Figure 2. The peptides shown
are doubly charged, so the light- and heavy-labelled versions
Proteomics of Plant Proteins
of each peptide are separated by m/z 4.5. Each peptide is
represented by a series of peaks due to the natural abundance
of 13 C. As shown in Figure 2, the ER proteins calnexin
and cytochrome b5 and the Golgi protein gtl-6 are present
in both the ER-rich and Golgi-rich fractions, due to the
cross-contamination already discussed. However, the use of
ICAT labelling enables us to compare the amount of each
protein in the two fractions. Calnexin and cytochrome b5
are clearly more abundant in the ER-rich fraction and gtl-6
is more abundant in the Golgi-rich fraction, thus enabling
us to distinguish between Golgi-localized and ER-localized
proteins without the need to obtain pure Golgi and ER preparations. A comprehensive analysis of the membrane proteins present in the ER-rich and Golgi-rich membrane
fractions, and the quantification of their relative abundances,
is currently in progress. This approach should enable the
localization of previously uncharacterized endomembrane
proteins by comparing their relative abundances with those
of proteins of known locations.
Conclusions
Subcellular proteomic analysis has typically involved the
cataloguing of proteins present in a single organelle-enriched
fraction. This approach has been reasonably successful for
easily purified organelles, such as the mitochondria and
chloroplast. However, the plant endomembrane system has so
far proved refractory to subcellular proteomic analysis, due
to the difficulty in isolating its component organelles. The
use of quantitative proteomics to analyse the relative levels of
proteins in different organelle-enriched fractions provides a
solution to this problem and enables us to distinguish between
proteins from different subcellular compartments without the
need to obtain pure organelles.
We thank the Biotechnology and Biological Sciences Research
Council and Applied Biosystems for financial support and also Applied
Biosystems for the gift of the ICAT reagents. We also thank the
members of Cambridge Centre for Proteomics for their technical
assistance and Zhinong Zhang for maintaining the Arabidopsis callus.
References
1 Dreger, M. (2003) Mass Spectrom. Rev. 22, 27–56
2 Peltier, J.B., Emanuelsson, O., Kalume, D.E., Ytterberg, J., Friso, G.,
Rudella, A., Liberles, D.A., Soderberg, L., Roepstorff, P., von Heijne, G.
et al. (2002) Plant Cell 14, 211–236
3 Aebersold, R. and Mann, M. (2003) Nature (London) 422, 198–207
4 Kruft, V., Eubel, H., Jansch, L., Werhahn, W. and Braun, H.P. (2001)
Plant Physiol. 127, 1694–1710
5 Millar, A.H., Sweetlove, L.J., Giege, P. and Leaver, C.J. (2001)
Plant Physiol. 127, 1711–1727
6 Santoni, V., Molloy, M. and Rabilloud, T. (2000) Electrophoresis 21,
1054–1070
7 Ferro, M., Salvi, D., Riviere-Rolland, H., Vermat, T., Seigneurin-Berny, D.,
Grunwald, D., Garin, J., Joyard, J. and Rolland, N. (2002) Proc. Natl.
Acad. Sci. U.S.A. 99, 11487–11492
8 Ferro, M., Salvi, D., Brugiere, S., Miras, S., Kowalski, S., Louwagie, M.,
Garin, J., Joyard, J. and Rolland, N. (2003) Mol. Cell. Proteomics 2,
325–345
9 Millar, A.H. and Heazlewood, J.L. (2003) Plant Physiol. 131, 443–453
10 Taylor, R.S., Wu, C.C., Hays, L.G., Eng, J.K., Yates, J.R. and Howell, K.E.
(2000) Electrophoresis 21, 3441–3459
11 Taylor, R.S., Jones, S.M., Dahl, R.H., Nordeen, M.H. and Howell, K.E.
(1997) Mol. Biol. Cell 8, 1911–1931
12 Dominguez, M., Fazel, A., Dahan, S., Lovell, J., Hermo, L., Claude, A.,
Melancon, P. and Bergeron, J.J.M. (1999) J. Cell Biol. 145, 673–688
13 Bell, A.W., Ward, M.A., Blackstock, W.P., Freeman, H.N.M.,
Choudhary, J.S., Lewis, A.P., Chotai, D., Fazel, A., Gushue, J.N.,
Paiement, J. et al. (2001) J. Biol. Chem. 276, 5152–5165
14 van Wijk, K.J. (2001) Plant Physiol. 126, 501–508
15 Unlu, M., Morgan, M.E. and Minden, J.S. (1997) Electrophoresis 18,
2071–2077
16 Lilley, K.S., Razzaq, A. and Dupree, P. (2002) Curr. Opin. Chem. Biol. 6,
46–50
17 Gygi, S.P., Rist, B., Gerber, S.A., Turecek, F., Gelb, M.H. and Aebersold, R.
(1999) Nat. Biotechnol. 17, 994–999
18 Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H.,
Pandey, A. and Mann, M. (2002) Mol. Cell. Proteomics 1, 376–386
19 Foster, L.J., de Hoog, C.L. and Mann, M. (2003) Proc. Natl. Acad.
Sci. U.S.A. 100, 5813–5818
20 Fujiki, Y., Hubbard, A.L., Fowler, S. and Lazarow, P.B. (1982) J. Cell Biol.
93, 97–102
Received 25 November 2003
C 2004
Biochemical Society
523