Proteomics 2004, 4, 397–406
DOI 10.1002/pmic.200300607
Wojciech Szponarski1
Nicolas Sommerer2
Jean-Christophe Boyer1
Michel Rossignol2
Rémy Gibrat1
Large-scale characterization of integral proteins
from Arabidopsis vacuolar membrane by
two-dimensional liquid chromatography
1
UMR 5004, Plant Biochemistry &
Molecular Physiology, INRA
2
UR 1199, Proteomics, INRA
Montpellier, France
397
We developed a method to characterize different classes of membrane proteins within
a single experiment and using simple matrix-assisted laser desorption/ionization-time
of flight-mass spectrometry (MALDI-TOF-MS) analysis. After membrane solubilization
with the nondenaturing detergent n-dodecyl-b-D-maltoside, proteins were separated
successively by gel filtration and ion-exchange chromatography and finally by sodium
dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). This procedure
allowed to characterize 70 proteins from a membrane fraction enriched in plant vacuolar membrane (Arabidopsis), including integral proteins like the V0 complex of the H1ATPase, the H1-pyrophosphatase or the glutathione S-conjugate ATPase AtMRP1,
and peripheral proteins like the subunits of the catalytic V1 complex of the H1-ATPase.
Approximately 60% of identified proteins were predicted to possess at least two transmembrane domains. Furthermore, proteins, with molecular masses ranging between
20 and 200 kDa were distributed into two populations with maximum frequencies at
pI 5.3 and 8.9. Finally, this procedure appeared to allow the identification of proteins
known to be minor in whole-cell extracts like signaling or vesicular trafficking proteins.
Almost 50% of the proteins identified were functionally unknown whereas the others
confirmed that the plant vacuole is a multipurpose compartment.
Keywords: Arabidopsis thaliana / Liquid chromatography / Mass spectrometry / Membrane proteins / Vacuole
PRO 0607
1 Introduction
Approximately 20–30% of genes in animal, plant, yeast,
and bacteria genomes encode integral membrane proteins [1, 2] that are involved in fundamental functions like
uptake and compartmentalization of ions and nutriments,
or cellular signaling. Membrane proteins are also of considerable importance for diagnostics and therapeutics:
50% of currently known drug targets are either membrane
receptors or ion channels [3]. Therefore, a large knowledge of membrane proteomes would be desirable. Presently, the most popular proteomic approach involves the
combination of two-dimensional gel electrophoresis
(2-DE) with mass spectrometry (MS), which offers simple
and straightforward access to large protein patterns,
including to their change upon treatments. Unfortunately,
Correspondence: Dr. Rémy Gibrat, UMR 5004 – BPMP, INRA, 2
place Viala, F-34060 Montpellier cedex 1, France
E-mail: [email protected]
Fax: 133-(0)4-6752-5737
Abbreviations: AA, amino acid; BTP, 1,3-bis[tris(hydroxymethyl)-methyl-amino]-propane; DM, n-dodecyl b-D-maltoside;
GRAVY, grand average of hydrophobicity; TMD, transmembrane
domain
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
despite recent improvements [4], this technology remains
poorly suitable to separate highly hydrophobic, basic or
low-abundance proteins [5, 6]. Thus, subcellular membrane proteomes, and specially their integral protein moiety, remain poorly accessible by 2-DE. On the other hand,
protein identification is most routinely performed from the
peptide mass fingerprints generated by MALDI-TOF-MS.
This technique offers the specific advantages to operate
at both quite high throughput and relatively low cost.
However, some classes of peptides, such as hydrophobic
peptides, are not easily observed, and data from complex
protein mixtures are poorly usable to query databases.
Therefore, the most popular large-scale approaches are
not appropriate to investigate hydrophobic proteins.
More recently, alternative approaches, relying on chromatographic protein separation or not-involving prior
separation of proteins were proposed (review [7]). However, all these approaches involve large-scale shotgun
sequence determination, and, therefore, require more
sophisticated MS/MS or coupled LC-MS/MS. On the
other hand, quantitative differential proteomics is usually
not possible unless using specific labeling [8]. As a consequence, proteomic studies on membrane proteins have
lagged behind that of soluble proteins [7], and membrane
proteins remain underrepresented in the databases.
www.proteomics-journal.de
398
W. Szponarski et al.
In plants, early attempts to characterize membrane proteomes concerned the plasma membrane and several organelle membranes. These works demonstrated also that
integral proteins were poorly accessible when resolved by
2-DE [9–11]. More recently, alternative procedures were
proposed, and applied successfully to the chloroplast
envelope, the endoplasmic reticulum, and the mitochondrion [12–14]. Their common strategy was to focus on
some protein subpopulation, by carbonate stripping or
solvent extraction of membrane, and to use a simple
SDS-PAGE step to resolve the proteins. However, quite
complex protein patterns were obtained, and protein
could be identified only by MS/MS. Another consequence
is that separate procedures, like the classical 2-DE, are
required in order to establish comprehensive membrane
proteomes including the peripheral proteins. At the cell
level, as mentioned, only a few membrane types benefited from these efforts. Several membranes, including
the two terminal membranes that delimit the cytoplasm,
the cell membrane (plasma membrane) and the vacuolar
membrane (tonoplast), remained largely uncharacterized,
particularly with respect to integral proteins.
In the present work, we investigated the tonoplast proteome from A. thaliana. For this purpose, we developed
an alternative procedure allowing for the separation of
both integral and peripheral proteins, and identification
by MALDI-TOF-MS consisting of (i) the solubilization of
proteins with detergents, (ii) two successive separation
steps of the population by gel filtration and anionexchange chromatography; and (iii) final separation of
proteins by SDS-PAGE prior to in-gel digestion and protein identification. We report here the characterization of a
first set of 70 tonoplast proteins.
2 Materials and methods
Proteomics 2004, 4, 397–406
at 95006g. The supernatant was centrifuged for 15 min
at 25 0006g and the resulting supernatant was centrifuged once more for 35 min at 80 0006g. The pellet consisting in microsomal membranes was resuspended in a
gradient buffer (5 mM BTP-MES, pH 7.5, 5 mM DTT, 1 mM
Na2EDTA, 300 mM mannitol, 1 mM leupeptin, 1 mM benzamidine), loaded onto a discontinuous sucrose gradient
(45–38–30–25%) and centrifuged for 4 h at 80 0006g.
The membrane fraction recovered from the top of the gradient was further diluted 4-times in gradient buffer, centrifuged for 45 min at 160 0006g, resuspended in the gradient buffer complemented with 20% glycerol, and stored in
liquid nitrogen. According to classical markers, this fraction was highly enriched in vacuolar membrane (bafilomycin-sensitive ATPase) over plasma membrane (vanadatesensitive ATPase), whereas the presence of mitochondria
(azide-sensitive ATPase) was not detected.
2.2 Protein separation
Membranes were solubilized by n-dodecyl b-D-maltoside
(DM) at 207C and at a detergent/protein ratio of 25. Nonsolubilized proteins were removed by centrifugation
(60 min at 190 0006g). Proteins from supernatant were
first chromatographied by gel filtration at 207C either on a
Superose 6 HR 10/30, or on a HiLoad Superdex 200 16/60
columns (Amersham Biosciences, Orsay, France) in elution buffer composed of 5 mM BTP-Cl (pH 7.5), 1.5 mM
DM and 20% glycerol. The second chromatography by
anion exchange was performed on a MonoQ HR 1 mL
column (Amersham Biosciences) using a linear 0–1 M
NaCl gradient. Protein fractions were either concentrated
using Microcon YM10 (Millipore, St Quentin, France) or
precipitated by 10% TCA, then solubilized by SDS in the
presence of b-mercaptoethanol [16]. Gels were stained
with Coomassie Brilliant Blue R-250.
2.1 Cell culture and membrane preparation
2.3 In-gel digestion and MS analysis
Arabidopsis thaliana suspension cells, ecotype Columbia, were cultured at 247C under continuous light
(30 mE?cm22 ?s21) [15]. Cells were subcultured with a 1:9
dilution factor every 7 days, and were used for membrane
isolation at a density of about 150 g FW?L21, 7 days after
transfer. Cells were filtered, washed extensively in 20 mM
KCl, 5 mM Na2EDTA (pH 5.7), and incubated for 10 min in a
grinding buffer (20 mM 1,3-bis[tris(hydroxymethyl)methyl-amino]-propane (BTP)-ascorbate pH 7.8, containing 300 mM mannitol, 0.5% PVP, 2 mM Na2EDTA, pH 7.8,
2.5 mM MgCl2, 5 mM DTT, 1 mM PMSF, 1 mM leupeptin,
and 1 mM benzamidine). Cells were then disrupted at 47C
in a Cell Disrupter (Constant System, Warwick, England)
at constant pressure of 700 bar and centrifuged for 20 min
Bands were excised from the gel for in-gel trypsin digestion. The protocol was adapted from [17]. The excised
fragments were extensively washed in microcentrifuge
tubes with 25 mM ammonium bicarbonate, 50% HPLCgrade acetonitrile, and were further dehydrated in pure
acetonitrile. Gel fragments were dried under vacuum on
a centrifugal evaporator. Digestion was carried out overnight at 377C with 10–15 mL of 0.25 mg?mL21 trypsin (sequencing grade, modified; Promega, Charbonières,
France) in 25 mM ammonium bicarbonate (pH 7.8). The
resulting tryptic fragments were extracted twice with
100 mL of acetonitrile/water (3:2 v/v) containing 0.1% trifluoroacetic acid in an ultrasonic bath for 15 min. The
pooled supernatants were concentrated to a final volume
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
Proteomics 2004, 4, 397–406
of ca. 10 mL in a centrifugal evaporator. For MS analysis,
a-cyano-4-hydroxycinnamic acid matrix was prepared at
half saturation in acetonitrile/water (1:1 v/v) acidified with
0.1% trifluoroacetic acid. Then, 0.8 mL of each sample
was mixed with 0.8 mL of the matrix and the mixture was
immediately spotted on the MALDI target and allowed to
dry and crystallize. The target was then rinsed with ultrapure water. Mass spectra were recorded on a BiFlex III
MALDI-TOF mass spectrometer (Bruker Daltonics, Bremen, Germany). Reflector spectra were obtained over a
mass range of 600–3500 Da in the short-pulsed ion
extraction mode using an accelerating voltage of 19 kV.
Spectra from 100 to 200 laser shots were summed to
generate a peptide mass fingerprint for each protein
digest. At least two peptide ions generated by the autolysis of trypsin were used as internal standards for calibrating the mass spectra. Manual monoisotopic mass assignment was performed using Bruker’s SNAP procedure.
2.4 Database search
Vacuolar membrane proteome
399
were selected in order to avoid any effect on the ionic
properties. In preliminary experiments, DM was shown to
be the most efficient in solubilizing tonoplast proteins.
As a first chromatographic step, gel filtration was used.
Preliminary runs using a wide range Superose 6 column
indicated that the protein-detergent complexes displayed
apparent molecular masses ranging from 100 to
2000 kDa. Accordingly, a high-resolution Superdex 200
column was selected. As shown in Fig. 1A, proteins were
eluted within five overlapping peaks. However, their polypeptide composition was markedly different as revealed
by the SDS-PAGE patterns obtained for different fractions
(Fig. 1B). This indicated that size separation of proteindetergent complexes allowed for the separation of membrane proteins.
In order to further resolve the still complex fractions
recovered, a second chromatographic step was introduced. Figure 2A illustrates, for some fractions from the
previous gel filtration chromatography, the chromatographic profiles obtained using an anion exchange
For protein identification, the latest versions of the
National Center for Biotechnology Information nonredundant database were manually searched with the monoisotopic peptide mass lists, using the ProFound search
engine [18]. Search parameters were as following: first all
taxa and then Arabidopsis thaliana as taxonomic category, mass tolerance of 0.1 Da, 1 missed cleavage site
allowed, protein mass range of 0–1000 kDa, pI range of
0–14, no chemical modification of amino acids. To take
into account the possible presence of protein mixtures,
up to four different runs were performed by changing the
supposed number of proteins present in excised band.
The run giving the best score was then conserved and,
for each identified protein, the list of experimental matching masses was used for a final run in the version “single
protein”. Only proteins presenting both high probability
and Z-score higher than 1.65 were considered as identified with a sufficient confidence.
3 Results
3.1 Separation of tonoplast proteins
The present work aimed at solubilizing tonoplast proteins
with detergents, for subsequent screening by liquid chromatography, and final separation by SDS-PAGE. In this
kind of procedure, the experimental conditions have to
be selected in such a way that the charge and size of protein-detergent complexes are governed by their protein
moiety. In this view, neutral or zwitterionic detergents
(CHAPS, DM, n-octyl glucoside, and zwittergent 3.14)
Figure 1. First chromatographic separation of DM-solubilized vacuolar membrane proteins: gel filtration. (A) Elution of proteins, layered onto a Superdex 200 column and
monitored at 280 nm. (B) SDS-PAGE: lanes 1–8 correspond to the fractions indicated in the chromatogram
above; lane T : total vacuolar membrane proteins.
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
400
W. Szponarski et al.
Proteomics 2004, 4, 397–406
MonoQ column. For all fractions, a part of proteins was
not retained at pH 7.4, and retained proteins could be
eluted using a saline gradient. The SDS-PAGE polypeptide patterns of both retained and nonretained fractions
were quite simple, in terms of number of protein bands,
and markedly different (Fig. 2B). This indicated that the
separation of protein-detergent complexes according to
their electrical charge allowed also to separate further
the tonoplast proteins. Combination of the two steps
resulted in a rich pattern including more than 200 SDSPAGE bands according to Mr and elution volume in gel
filtration or ionic strength for elution in ion exchange chromatography.
3.2 MS characterization of proteins
As the procedure appeared to result in quite simple protein patterns, as judged from final SDS-PAGE patterns, a
large characterization of the bands was undertaken. For
this purpose, a number of bands visible after Coomassie
blue staining were excised, in-gel digested using trypsin,
and the digests were analyzed by MALDI-TOF-MS.
Nearly 70 proteins were identified by accessions in databases (Table 1), including ca. 15% of proteins coresolved
in the same bands. On the whole, according to the TargetP [19] algorithm, less than 10% of the proteins displayed a predicted targeting signal suggesting a possible
mitochondrial or chloroplastic origine.
Figure 2. Second chromatographic separation of DMsolubilized vacuolar membrane proteins: anion exchange.
(A) Chromatogram of protein fraction No. 2 of gel filtration
(Fig. 1), layered onto a MonoQ HR exchanger and eluted
according to a linear NaCl gradient (dashed line). (B) SDSPAGE lanes: lane 1, protein fraction non retained onto the
MonoQ exchanger; lanes 2–9, protein fractions eluted by
NaCl as indicated in (A).
Table 1. Features of the proteins identified in the vacuolar membrane fraction from Arabidopsis thaliana
Definition
% Cov. N
Mr
pI
GRAVY
TMD AA/TMD
Accession
No.
Locus
NP_178011
At1g78900
ATPase 70 kDa subunit, putative
67
39
68.8
63
5.1
20.172
2
NP_192853
At4g11150
H1-transporting ATPase chain E,
vacuolar
70
20
26.0
23
6
20.487
0
NP_563916
At1g12840
Vacuolar ATP synthase subunit C,
putative
64
25
42.6
45a
5.4
20.284
1
V-ATPase subunit D
34
8
29.1
32b
9.5
20.243
0
calc.
CAB46439
meas.
TargetP
Loc.
312
375
55
R
other
1
other
4
mito
5
mito
1
BAA32210
At1g15690
Vacuolar proton pyrophosphatase
21
15
80.8
72
5.1
0.615 14
NP_568051
At4g39080
Putative proton pump
40
32
92.8
88
5.6
0.033
7
117
secpath 1
other
3
NP_174329
At1g30400
Glutathione S-conjugate transporting
ATPase (AtMRP1)
26
27
181.9 170
5.9
0.029 17
95
other
1
MRP-like ABC transporter
28
30
169.1 155c
5.9
0.178 15
101
other
3
NP_182301
At2g47800
Glutathione-conjugate transporter
AtMRP4
15
18
169.1 155c
7.8
0.178 17
89
other
4
NP_191292
At3g57330
Ca21-transporting ATPase-like
protein
29
28
111.9 103
6
0.143 10
103
chloro
4
AAC49791
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
Proteomics 2004, 4, 397–406
Vacuolar membrane proteome
401
Table 1. Continued
% Cov. N
pI
GRAVY
Locus
NP_567258
At4g03560
Putative calcium channel
21
13
84.9
78
4.9
NP_181428
At2g38940
Phosphate transporter (AtPT2)
20
7
58.6
55
8.7
T49233
At3g43190
Sucrose synthase-like protein
13
9
93.0
97
6.1
20.3
NP_182287
At2g47650
Putative dTDP-glucose
4–6-dehydratase
21
7
49.9
53
9.1
NP_200261
At5g54500
1,4-Benzoquinone reductase-like;
Trp repressor binding protein-like
50
6
21.8
23
NP_195226
At4g35000
L-Ascorbate
42
11
31.6
NP_567350
At4g10050
Lipase-like protein
22
8
38.5
NP_187214
At3g05630
Putative phospholipase D
17
13
NP_197756
At5g23670
Serine palmitoyltransferase
17
10
54.3
50
NP_176653
At1g64720
Membrane-related protein CP5,
putative
17
7
43.7
NP_193872
At4g21410
Serine threonine kinase-like
protein
16
9
76.0
NP_187455
At3g07980
Putative MAP3K epsilon protein
kinase
13
AAB33487
At4g21380
ARK3 product/receptor-like serine/
threonine protein kinase ARK3
AAD23040
At2g46370
Auxin-responsive GH3-like protein
NP_199387
Definition
Mr
Accession
No.
At5g45750
E71440
TargetP
Loc.
R
61
other
2
0.323 12
45
other
4
2
404
mito
5
20.337
2
222
other
1
6
20.108
1
204
other
5
32b
6.5
20.365
1
287
other
3
41
5.7
20.246
3
117
other
3
117.9 128
6.1
20.391
1
1039
other
4
9.4
20.064
3
163
secpath 3
45a
9.6
20.214
2
193
secpath 2
81
8.6
20.202
3
228
secpath 1
12
151.1 162
5.9
20.331
4
342
other
17
13
96.6 100
8.2
20.286
3
284
secpath 2
18
8
5.6
20.141
2
288
other
1
5.6
20.287
1
216
other
3
104
other
5
calc.
peroxidase
TMD AA/TMD
64.5
meas.
70
d
Rab-type small GTP-binding
protein-like
41
9
23.9
25
GTP-binding protein RAB2A
43
8
23.2
25d
0.244 12
4
7
20.256
2
NP_200410
At5g55990
Calcineurin B-like protein 2
40
9
25.8
28e
4.9
20.22
0
NP_173136
At1g16920
Guanine nucleotide regulatory
protein, putative
35
8
24.0
28e
5.6
20.341
1
216
other
NP_197628
At5g22360
Synaptobrevin-like protein
39
6
25.3
23
9.6
0.072
2
111
secpath 5
NP_187555
At3g09440
Heat-shock protein (At-hsc70–3)
20
8
71.1
76
5
20.394
1
649
other
NP_187724
At3g11130
Putative clathrin heavy chain
21
23
5.2
20.152
2
853
secpath 4
NP_200987
At5g61790
Calnexin-like protein
26
12
2
265
secpath 2
193.2 180
60.5
65
4.8
20.758
f
secpath 5
3
2
AAD02498
At1g75780
b-Tubulin 1
21
7
50.2
53
4.7
20.38
0
other
2
NP_193232
At4g14960
Tubulin a-6 chain (TUA6)
31
11
49.5
53f
4.9
20.2
1
450
other
5
NP_180515
At2g29550
Tubulin b-7 chain
31
9
50.7
53f
4.7
20.413
1
449
other
2
NP_177360
At1g72150
Cytosolic factor, putative
28
13
64.0
67
4.8
20.535
0
other
3
AAF02837
At1g56075
Elongation factor EF-2
21
14
94.2
90
5.9
20.228
2
other
3
NP_564623
At1g53210
Expressed protein
28
14
53.6
47
5.2
0.465
9
55
NP_568671
At5g46860
Expressed protein
44
7
29.5
26
6
20.373
1
268
mito
5
g
6.3
20.328
1
382
other
2
9.8
20.426
2
184
secpath 1
NP_566556
At3g16580
Expressed protein
18
7
44.4
46
NP_565975
At2g42570
Expressed protein
19
7
43.2
46g
NP_195548
At4g38350
Putative protein
11
10
NP_193616
At4g18810
Putative protein
28
15
116.0 102
68.4
5.3
secpath 5
0.259 10
105
other
2
9
20.294
2
311
chloro
3
h
9.5
20.331
0
other
2
8.6
20.641
1
other
3
71
NP_191816
At3g62570
Putative protein
20
10
62.0
67
NP_192879
At4g11400
Putative protein
15
8
65.7
67h
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
423
573
www.proteomics-journal.de
402
W. Szponarski et al.
Proteomics 2004, 4, 397–406
Table 1. Continued
Definition
% Cov. N
Accession
No.
Locus
NP_197958
At5g25860
Putative protein
23
NP_193002
At4g12650
Putative protein
NP_195600
At4g38890
Putative protein
NP_197638
At5g22460
NP_191202
At3g56430
NP_197853
NP_196676
Mr
calc.
pI
51.9
54
9.1
21
8
59.7
55
27
12
78.6
83
Putative protein
28
7
39.1
Putative protein
16
9
48.5
At5g24650
Putative protein
37
8
At5g11150
Putative protein
43
7
NP_180257
At2g26890
Unknown protein
14
NP_174433
At1g31480
Unknown protein
Unknown protein
AAK59445
TMD AA/TMD
meas.
11
AAL32694
GRAVY
TargetP
Loc.
R
other
3
other
4
other
1
20.198
0
5.9
0.532
9
6.2
20.507
0
43
7.2
20.271
4
85
52
9.7
20.226
2
217
other
5
27.8
31
9.6
20.125
4
65
other
1
25.3
23
9.4
20.164
2
25
277.1 240
5.7
20.072
8
634
other
3
16
11
99.4 106
5.7
20.705
2
435
other
1
39
15
54.3
5
20.25
0
other
2
57
59
secpath 4
secpath 4
Unknown protein
36
8
36.5
34
8.8
0.03
2
169
secpath 5
NP_180318
At2g27490
Unknown protein
18
7
25.7
23
9.5
0.016
1
232
other
4
NP_196906
At5g14020
Unknown protein
19
8
47.9
50
6.7
0.041
2
216
other
4
AAK82537
At1g07930
At1g07930/T6D22_3
33
10
49.4
53
9.4
20.325
0
other
2
AAK59761
AT3g42050
50.3
48
6.6
20.063
0
other
2
108.4 115
9.7
20.306
1
977
other
2
4
AT3g42050/F4M19_10
51
20
AAF79278
F14D16.2
19
16
AAD55276
F25A4.3
18
11
78.0
84
5.5
20.386
4
175
chloro
AAF87033
T24P13.18
23
9
58.6
63
9.7
20.288
2
253
secpath 1
NP_182060
At2g45350
Hypothetical protein
18
10
68.4
71i
6.7
0.038
3
202
other
3
NP_189566
At3g29210
Hypothetical protein
17
9
67.9
71i
5.4
20.782
1
594
other
1
NP_176767
At1g65920
Hypothetical protein
17
16
111.1 120
9.4
20.433
3
335
other
2
NP_180894
At2g33360
Hypothetical protein
17
8
66.5
68
9.2
20.474
0
other
2
NP_179704
At2g21080
Hypothetical protein
20
7
46.6
43
9.3
0.415
6
69
other
2
NP_173079
At1g16290
Hypothetical protein
17
8
52.3
55
5.6
20.466
4
114
other
2
Hypothetical protein T20N10.250
21
10
59.7
65
9
20.122
1
517
secpath 5
T49173
Vacuolar membrane proteins were analyzed by MALDI-TOF-MS. Columns correspond to: accession, accession No. in
NCBI; locus, locus according to AGI annotation; definition, protein definition for the NCBI accession; % Cov., percentage
of coverage of the database sequence by the experimental peptide mass fingerprint; N, number of peptides matching
specifically the accession; Mr calc., molecular weight calculated from the sequence; Mr meas., molecular weight measured
on SDS-PAGE; pI, isoelectric point calculated from the sequence; GRAVY, grand average of hydrophobicity; TMD, number
of transmembrane domains expected from the sequence according to the TMPred software [21]; AA/TMD, ratio of the
number of AA to the number of TMDs; TargetP, subcellular location (Loc.) predicted by the TargetP algorithm [19]: chloro,
occurrence of a chloroplast transit peptide; mito, occurrence of a mitochondrial targeting peptide; secpath, occurrence of a
secretion signal peptide; other, no transit or secretion peptide found in the 130 N-ter AA of the protein sequence; R, reliability class of the TargetP prediction (the class 1 being the most reliable one). Letters in superscript in the Mr meas. column
refer to proteins identified in a single band excised from a same gel.
Identified proteins included proteins previously characterized at the tonoplast, together with a large number of
putative or hypothetical proteins deduced from systematic annotation of Arabidopsis genome. Well-known tonoplast proteins included the two proton pumps (the PPase
and the V-ATPase) and one ABC transporter, AtMRP1.
Other transporters, previously not identified at the vacuo-
lar membrane, were two other ATPases of the ABC transporter family, two putative Ca21-ATPases and channels,
and one high-affinity phosphate carrier. The other identified proteins covered a range of functions including lipid
synthesis or degradation, protein degradation, kinases
and phosphatases, proteins involved in membrane trafficking, and cytoskeleton proteins.
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
Proteomics 2004, 4, 397–406
Vacuolar membrane proteome
403
3.3 Features of proteins characterized in the
tonoplast-enriched fraction
Most proteins ranged between 20 and 120 kDa, with ca.
10% of proteins of higher molecular mass (Fig. 3A). In
terms of pI, proteins distributed into two main populations
with maximum frequencies observed at pI 5.3 and 8.9
(Fig. 3B). On the other hand, as in this work tonoplastenriched fractions were directly submitted to proteomic
analysis, the resulting population was expected to include
both peripheral and integral proteins. As a first insight, the
calculation of grand average of hydrophobicity (GRAVY)
scores [20] showed that 25% of proteins displayed positive scores, with values up to 0.615 (Table 1). Further, the
occurrence of transmembrane domains (TMDs) was predicted using the TMPred algorithm [21]. Approximately
25% of the identified proteins were expected to possess
at least 4 TMDs, including proteins with up to 17 TMDs,
and this proportion dropped to 60% for proteins with at
least 2 TMDs (Fig. 4A). Interestingly, for these 60% proteins, the observed distribution according to the number
of TMDs (Fig. 4A) was close to that computed using
another algorithm, HMMTOP [22] in the Arabidopsis
Membrane Protein Library [23]. In order to get more
insights into the relative contribution of putative TMDs to
Figure 4. Transmembrane domains predicted from the
sequence of vacuolar membrane proteins. Seventy proteins were identified in the vacuolar membrane fraction
(Table 1). (A) Distribution of TMDs according to the
TMPred algorithm; the continuous line gives the distribution of proteins which, detected in the whole translated
genome of A. thaliana, exhibit at least 2 TMDs [23]; (B)
distribution of the ratio AA/TMD; open symbols refer to
the known transport proteins.
the sequence, the ratio of the total number of amino acids
(AAs) over the number of predicted TMDs was related to
the number of TMDs (Fig. 4B). For proteins showing at
least 6 TMDs, an average AA/TMD ratio of 79 6 25 was
observed, whereas large variations in this ratio were calculated for less hydrophobic proteins. In addition, all
those proteins that were identified at the function level
corresponded to transporters.
4 Discussion
4.1 Membrane protein separation
Figure 3. Biochemical characteristics deduced from the
sequence of vacuolar membrane proteins. Seventy proteins were identified in the vacuolar membrane fraction
(Table 1). (A) molecular weight; (B) pI.
It is now well recognized that 2-DE is very efficient to
separate soluble proteins but poorly suitable for wide
analysis of subcellular membrane proteomes [7].
Whereas the second migration step (SDS-PAGE) is
known to work efficiently with integral proteins, the aggregation of hydrophobic proteins that occurs upon reaching
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
404
W. Szponarski et al.
their pI during the first migration step constitutes a major
bottleneck. On the other hand, ion-exchange chromatography can be formally considered as an alternative way to
isoelectric focusing, since it separates proteins according
to local and outer arrays of ionized residues. Therefore, in
order to overcome the limitation of 2-DE for hydrophobic
proteins, in this work, we hypothesized that chromatographic procedures could be helpful, provided that proteins were maintained in solution. For this purpose, we
deviced a procedure aiming at solubilizing the proteins
using neutral detergents. By this way, we expected that
the surface charge of protein-detergent complexes would
be mainly contributed by proteins, thus allowing for their
separation. Indeed, although overlapping elution profiles
were obtained, both ion-exchange chromatography and
gel filtration were shown to allow the separation of fractions differing by their protein pattern. Furthermore, in gel
filtration experiments, no clear correlation was found between protein Mr and elution order, and proteins of different Mr in SDS-PAGE were resolved in chromatographic
fractions displaying apparent Mr up to 2000 kDa. This
suggests that the procedure did not result in the solubilization of single proteins, but rather in the production of
micelles sufficiently small so that their size and charge
was determined by the protein moiety. By combining the
two chromatographic steps, and by taking advantage of
the capacity of ion exchange to reconcentrate proteins
after the dilution caused by gel filtration, fractions displaying relatively simple patterns were finally obtained.
In this work, only one-third of the most abundant proteins
resolved by the described procedure were characterized
by MS. However, several arguments demonstrate that the
procedure is actually able to give a large picture of the
proteins constituting a membrane. First, the 70 identified
proteins covered a large range of both molecular mass
(including proteins of nearly 300 kDa) and pI (including
basic proteins with pI of nearly 10). Furthermore, the calculated pI distribution was bimodal, in agreement with
calculations made at the genome scale [24]. These results
indicate that the use of micelles did not introduce bias
according to protein size and charge. On the another
hand, both the coarse calculation of GRAVY scores and
the prediction of TMDs suggest that at least one quarter
of the identified proteins would be very hydrophobic and
display four predicted TMDs or more. Furthermore, the
identification of both true tonoplast integral proteins with
numerous TMDs, such as ABC transporters, and peripheral proteins, such as subunits of the catalytic part of the
vacuolar proton pump, confirms that the screening of proteins in micelles has the capacity to give a large access to
the different protein types constituting the membrane proteome. Taken together, these features show that the present procedure overcomes simultaneously some of the
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2004, 4, 397–406
main bottlenecks of classical 2-DE, with respect to high
Mr and/or basic and/or hydrophobic proteins. In this
respect, this procedure compares with published
approaches aiming at analyzing hydrophobic proteomes.
In Deinococcus radiodurans, combination of initial stripping of peripheral proteins with solvent extraction was
recently shown to result in the characterization of a membrane subproteome encompassing 27% hydrophobic
proteins, as judged by their GRAVY score [26]. On the other
hand, using highly purified preparations of chloroplasts,
direct solvent extraction was proven to allow the identification of 39% proteins with at least four TMDs, most of
them being basic proteins [12]. Therefore, both solventand micelles-based approaches appear to give access
to hydrophobic and basic proteomes with similar efficiency. However, as solubilization of proteins in micelles
enables chromatographic fractionation prior to final
SDS-PAGE, protein identification becomes efficient by
simple MALDI-TOF-MS, and does not require LC-MS/
MS sequencing. In addition, for the same reason, quantitative proteomics can be expected from the comparison
of relatively simple SDS-PAGE patterns. Therefore, owing
to the increasing amount of knowledge on numerous genomes, the present procedure is likely to offer advantages
in terms of throughput and cost for a wide number of both
analytical and comparative applications. Finally, it should
be emphasized that, as this procedure is able to work with
relatively high initial protein loading and concentrates proteins in a quite large number of fractions according to their
physicochemical properties, it gives access also to minor
membrane proteins in final SDS-PAGE lanes, such as,
here, signaling or trafficking proteins.
4.2 Vacuolar membrane proteome
The vacuole is the largest compartment in mature plant
cells, accounting for more than 80% of the total cell volume. Several small vacuoles may also co-exist with the
central vacuole, distinct either by their soluble or by their
membrane protein content [27], and the plant cell
vacuome is now considered as a multipurpose compartment. In this view, a large diversity of proteins can be
expected, in agreement, here, with both the protein patterns observed showing more than 200 different SDSPAGE bands and the nature of identified proteins.
4.2.1 General features
Several of the identified proteins show positive GRAVY
scores and numerous predicted TMDs, and can be therefore classified as integral proteins. At the same time,
these proteins do not display targeting signals suggesting
any organelle location, but exhibit a large bimodal pI diswww.proteomics-journal.de
Proteomics 2004, 4, 397–406
tribution. Recently, similar analysis resulted in the proposition that the simultaneous display of at least four TMDs,
a chloroplast targeting sequence, an AA over TMD ratio
lower than 100 and a pI higher then 8.8 could constitute
general features of the inner membrane proteins from the
chloroplast envelope in Arabidopsis [12]. On this basis,
Arabidopsis tonoplast integral proteins appear to resemble envelope proteins by their AA/TMD ratio, but differ by
a broader pI distribution. Owing to the present low molecular knowledge of plant membrane proteins, it can be
speculated that generalized knowledge of such properties could provide specific signatures of well-defined subcellular membranes allowing bioinformatical search for
predicting the location of novel membrane proteins in
Arabidopsis.
Vacuolar membrane proteome
405
4.2.3 Calcium homeostasis
Two putative Ca21-ATPases and channels were found here
in the vacuolar membrane fraction, as well as a putative
calcineurin (calmodulin-activated Ser-Thr protein phosphatase, essential for the transduction of Ca21 signals,
described as cytosolic). The spatial and temporal regulation of the cytoplasmic Ca21 concentration, primarily
involved in signal transduction, is controlled by coordinated activities of Ca21-ATPase and channels which
have been recorded in various membranes like the
plasma membrane, the tonoplast, the endoplasmic reticulum, chloroplasts, and the nuclear membranes of plant
cells [31, 32]. The observation of these various proteins
at the tonoplast reinforces the role of this membrane in
the regulation of calcium homeostasis.
4.2.2 Transport proteins
Plant vacuoles are known to carry out storage or mobilization of a variety of solutes (from mineral ions to sugars,
organic or amino acids), detoxification of xenobiotics, and
regulation of turgor pressure by coupling water to solute
transports. Transport proteins found here encompass the
two vacuolar proton pumps (the PPase and the hetero-oligomeric V-ATPase) which energize most (secondary)
vacuolar transport of solutes via the electrochemical gradient generated. In addition, three highly hydrophobic ABC
transport ATPases of high Mr were also found. ABC is the
largest known transport family, ranging from bacteria to
humans. At the vacuolar membrane level, ABC transporters function to sequester xenobiotics as well as endogenous compounds, which often require glutathione conjugation in the cytoplasm prior to their transport. To our
knowledge, only one of the three ABC carriers found here,
the glutathione S-conjugate ATPase AtMRP1, has been
functionally described at the vacuolar level previously
[28]. Therefore, this work could offer the first experimental
evidence for the presence of AtMRP4 and the protein
encoded by the gene AAC49791 at the tonoplast.
4.2.4 Vesicular trafficking and signaling
The concentration of cytosolic phosphate is regulated by
resupply from the vacuole [29]. The high affinity phosphate carrier found here in the vacuolar fraction (AtPT2)
was previously reported to be expressed in roots upon P
starvation [30]. Therefore, AtPT2 appears to be also
expressed in suspension cells which, when cultivated for
7 days in a standard medium, are P-starved. However,
AtPT2 is likely not located at the vacuolar membrane
since it complements a yeast strain (pho84) defective in
the high-affinity phosphate transport function. Accordingly, this could reveal either a plasma membrane contamination of the vacuolar fraction used in this work or
an alternative minor location of AtPT2.
Several proteins observed here are known to participate
to membrane trafficking in the cell (known or putative
Rab proteins, clathrin, Hsc70, synaptobrevin). Firstly,
Rab proteins are involved in vesicle targeting [33]. In
addition, in yeast, clathrin-coated vesicles mediate the
transport of the soluble vacuolar proteins from the
trans-Golgi network (TGN) to the endosomal/prevacuolar compartment [34]. Hsc70 is an uncoating ATPase,
stripping clathrin-coated vesicles [35]. Synaptobrevin is
a v-SNARE involved in the process of vesicles docking
and fusion, especially important for the inheritance of
vacuoles during the cell cycle [36]. It is well-established
that endoplasmic reticulum (ER)-derived transport-vesicles travel to and are incorporated into vacuoles [37],
and the role of vesicle-mediated solute transport between the vacuole and the plasma membrane has
been recently addressed [38]. The simultaneous finding
of the proteins above at the tonoplast might reflect,
therefore, the tight connections to vacuolar compartment that exist within vesicular transport from plasma
membrane to endosomal system. In the same way, it
should be pointed out that microtubules are required
for membrane trafficking, and are also tightly involved
in the dynamic organization of the vacuole during cell
cycle [39]. Interestingly, three tubulin gene products,
both from a- and b-types, were observed here. Finally,
three putative protein kinases, and the protein phosphatase calcineurin were found at the tonoplast. It could be
speculated whether their occurrence highlights potential
involvement in the perception and transduction of signals, like Ca21 signal, or simply reveals the highly regulated functioning of the various activities located at this
membrane.
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.de
406
W. Szponarski et al.
4.2.5 Metabolic activities
Beside transporters, several putative or known enzymes
involved in miscellaneous pathways, such as lipid synthesis or degradation (lipase, phospholipase D, palmitoyl
transferase), protein degradation (cystein proteinase),
sugar supply and metabolism (sucrose synthase, glucose
dehydratase) and peroxidation, have been found. For
most of them, their presence at the tonoplast is not unexpected. For instance, a phospholipase D activity has
been shown in the tonoplast of suspension cells of Acer
pseudoplatanus [40], as well as cysteine proteinases in
vacuoles of plant tissues [41]. Similarly, the ascorbate-dependent destruction of hydrogen peroxide, was suggested to occur also in the vacuole [42], and we detected
an ascorbate-peroxidase. Very recently, a tonoplastassociated sucrose synthase activity, that would be
involved in sucrose remobilization from the vacuole, was
demonstrated [43]. Accordingly, the present work suggests that the product of the At3g43190 gene, annotated
as sucrose-synthase-like protein, might constitute the
support for this activity previously unknown at this location.
In conclusion, the 2-D liquid chromatrograhy of proteins in
micelles described here appears to constitute an alternative way to investigate membrane proteomes. Owing to
its capacity to analyze simultaneously integral and peripheral proteins by simple MALDI-TOF-MS, this
approach can be speculated to be convenient both for
the characterization of various membrane proteomes
from sequenced organisms and for the investigation of
membrane-located responses by comparative proteomics.
This work was supported by the program “Novel tools”
from Genoplante (No. 19993663) and by the Proteome
Platform of the Montpellier-LR Génopole. The authors
are grateful to Cécile Fizames for helpful discussions.
Received May 13, 2003
Revised July 9, 2003
Accepted July 31, 2003
5 References
[1]
[2]
[3]
[4]
Wallin, E., von Heijne, G., Prot. Sci. 1998, 7, 1029–1038.
Stevens, T. J., Arkin, I. T., Proteins 2000, 39, 417–420.
Reiss, T., Trends Biotechnol. 2001, 19, 496–499.
Görg, A., Obermaier, C., Boguth, G., Harder, A. et al., Electrophoresis 2000, 21, 1037–1053.
[5] Santoni, V., Malloy, M., Rabilloud, T., Electrophoresis 2000,
21, 1054–1070.
[6] Gygi, S., Corthals, G. L., Zhang, Y., Rochon, Y., Aebersold, R.,
Proc. Natl. Acad. Sci. USA 2000, 97, 9390–9395.
 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2004, 4, 397–406
[7] Wu, C. C., Yates, J. R., Nat. Biotechnol. 2003 21, 262–267.
[8] Han, D. K., Eng, J., Zhou, H., Aebersold, R., Nature Biotechnol. 2001, 19, 946–951.
[9] Santoni, V., Rouquié, D., Doumas, P., Mansion, M. et al.,
Plant J. 1998, 16, 633–641.
[10] Prime, T. A., Sherrier, D. J., Mahon, P., Packman, L. C.,
Dupree, P., Electrophoresis 2000, 21, 3488–3499.
[11] Kruft, V., Eubel, H., Jänsch, L., Werhahn, W., Braun, H.-P.,
Plant Physiol. 2001, 127, 1694–1710.
[12] Ferro, M., Salvi, D., Rivière-Rolland, H., Vermat, T. et al.,
Proc. Natl. Acad. Sci. USA 2002, 99, 11487–11492.
[13] Maltman, D. J., Simon, W. J., Wheeler, C. H., Dunn, M. J.
et al., Electrophoresis 2002, 23, 626–639.
[14] Millar, A. H., Heazlewood, J. L., Plant Physiol. 2003, 131,
443–453.
[15] Axelos, M., Curie, C., Mazzolini, L., Bardet, C., Lescure, B.,
Plant Physiol. Biochem. 1992, 30, 123–128.
[16] Laemmli, U. K., Nature 1970, 222, 680–685.
[17] Jensen, O. N., Wilm, M., Shevchenko, A., Mann, M., Methods Mol. Biol. 1999, 112, 513–530.
[18] http://prowl.rockefeller.edu/cgi-bin/ProFound
[19] http://www.cbs.dtu.dk/services/TargetP/
[20] http://www.expasy.org/tools/protparam.html
[21] http://www.ch.embnet.org/software/TMPRED_form.html
[22] http://www.enzim.hu/hmmtop/
[23] Ward, J. M., Bioinformatics 2001, 17, 560–563.
[24] Luche, S., Santoni, V., Rabilloud, T., Proteomics 2003, 3,
249–253.
[25] Van Bogelen, R. A., Schiller, E. E., Thomas, J. D., Neidhardt,
F. C., Electrophoresis 1999, 20, 49–59.
[26] Blonder, J., Goshe, M. B., Moore, R. J., Pasa-Tolic, L. et al.,
J. Proteome Res. 2002, 1, 351–360.
[27] Martinoia, E., Massonneau, A., Frangne, N., Plant Cell Physiol. 2000, 41, 1175–1186.
[28] Lu, Y. P., Li, Z. S., Rea, P. A., Proc. Natl. Acad. Sci. USA
1997, 94, 8243–8248.
[29] Sakano, K., Yazaki, Y., Okihara, K., Mimura, T., Kiyota, S.,
Plant Physiol. 1995, 108, 295–302.
[30] Muchhal, U. S., Pardo, J. M., Raghothama, K. G., Proc. Natl.
Acad. Sci. USA 1996, 93, 10519–10523.
[31] Sze, H., Liang, F., Hwang, I., Curran, A. C., Harper, J. F., Ann.
Rev. Plant Physiol. Plant Mol. Biol. 2000, 51, 433–462.
[32] White, P. J., Biochim. Biophys. Acta 2000, 1465, 171–189.
[33] Hama, H., Tall, G. G., Horazdovsky, B. F., J. Biol. Chem.
1999, 274, 15284–15291.
[34] Deloche, O., Schekman, R. W., Mol. Biol. Cell 2002, 13,
4296–4307.
[35] Cremona, O., Dev. Cell 2001, 1, 592–604.
[36] Nichols, B. J., Ungermann, C., Pelham, H. R., Wickner,
W. T., Haas, A., Nature 1997, 387, 199–202.
[37] Chrispeels, M. J., Herman, E. M., Plant Physiol. 2000, 123,
1227–1234.
[38] Ed Echeverría, E., Plant Physiol. 2000, 123, 1217–1226.
[39] Kutsuna, N., Hasezawa, S., Plant Cell Physiol. 2002, 43,
965–973.
[40] Tavernier, E., Pugin, A., Biochim. Biophys. Acta 1995, 1233,
118–122.
[41] Muntz, K., Shutov, A. D., Trends Plant Sci. 2002, 7, 340–344.
[42] Mehlhorn, H., Lelandais, M., Korth, H. G., Foyer, C. H.,
FEBS Lett. 1996, 378, 203–206.
[43] Etxeberria, E., Gonzalez, P., J. Exp. Bot. 2003, 54, 1407–
1414.
www.proteomics-journal.de