Báo cáo khoa học: Insertion of the plant photosystem I subunit G into the thylakoid membrane In vitro and in vivo studies of wild-type and tagged versions of the protein doc

Insertion of the plant photosystem I subunit G into the
thylakoid membrane
In vitro and in vivo studies of wild-type and tagged versions of the
protein
Lisa Rosgaard*, Agnieszka Zygadlo, Henrik Vibe Scheller, Alexandra Mant† and Poul Erik Jensen
Plant Biochemistry Laboratory, Department of Plant Biology, The Royal Veterinary & Agricultural University, Frederiksberg, Denmark
Subunit G of photosystem I (PSI-G) is one of the
polypeptides that form the core of metaphyte PSI
(reviewed in [1]). This intricate complex of more than
100 pigments and 14 polypeptides uses solar energy
to transfer electrons from plastocyanin (PC) in the
thylakoid lumen, across the thylakoid membrane to
Keywords
His-tag; membrane topology; photosystem I;
Strep-tag; transgenic Arabidopsis
Correspondence
P. E. Jensen, Plant Biochemistry Laboratory,
Department of Plant Biology, The Royal
Veterinary & Agricultural University, 40,
Thorvaldsensvej, DK-1871 Frederiksberg C,
Denmark
Fax: +45 35283333
Tel: +45 35283354
E-mail:
*Present address
Novozymes A ⁄ S, Starch R & D, Laurentsvej
55, Bagsværd, Denmark
†Present address
Molecular Immunology Group, Cancer
Sciences Division, Southampton General
Hospital, Mailpoint 824, Tremona Road,

Southampton, SO16 6YD, UK
Note
L. Rosgaard and A. Zygadlo contributed
equally to this work
(Received 9 May 2005, revised 14 June
2005, accepted 17 June 2005)
doi:10.1111/j.1742-4658.2005.04824.x
Subunit G of photosystem I is a nuclear-encoded protein, predicted to
form two transmembrane a-helices separated by a loop region. We use
in vitro import assays to show that the positively charged loop domain
faces the stroma, whilst the N- and C-termini most likely face the lumen.
PSI-G constructs in which a His- or Strep-tag is placed at the C-terminus
or in the loop region insert with the same topology as wild-type photosys-
tem I subunit G (PSI-G). However, the presence of the tags in the loop
make the membrane-inserted protein significantly more sensitive to trypsin,
apparently by disrupting the interaction between the loop and the PSI core.
Knock-out plants lacking PSI-G were transformed with constructs enco-
ding the C-terminal and loop-tagged PSI-G proteins. Experiments on
thylakoids from the transgenic lines show that the C-terminally tagged ver-
sions of PSI-G adopt the same topology as wild-type PSI-G, whereas the
loop-tagged versions affect the sensitivity of the loop region to trypsin, thus
confirming the in vitro observations. Furthermore, purification of PSI
complexes from transgenic plants revealed that all the tagged versions of
PSI-G are incorporated and retained in the PSI complex, although the
C-terminally tagged variants of PSI-G were preferentially retained. This
suggests that the loop region of PSI-G is important for proper integration
into the PSI core. Our experiments demonstrate that it is possible to pro-
duce His- and Strep-tagged PSI in plants, and provide further evidence that
the topology of membrane proteins is dictated by the distribution of posit-
ive charges, which resist translocation across membranes.

Abbreviations
Chl, chlorophyll; Fd, ferredoxin; FNR, ferredoxin-NADP
+
oxidoreductase; His-tag, hexa-histidine tag; LHCI, light harvesting complex
associated with photosystem I; PSI, photosystem I; Lhcb1, major light harvesting complex apoprotein associated with photosystem II;
PC, plastocyanin; Strep-tag, trp-ser-his-pro-gln-phe-glu-lys tag.
4002 FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS
ferredoxin (Fd) on the stromal side. Reduced Fd can
donate electrons via ferredoxin-NADP
+
oxidoreduc-
tase (FNR) to produce NADPH, a central currency of
chemical energy. Light energy is channelled to the PSI
core by a peripheral antenna or light harvesting com-
plex (LHCI), consisting of four members (Lhca1–4) of
the chlorophyll a ⁄ b binding protein family. LHCI is
bound along one side of PSI only, in the vicinity of
the PSI-F and PSI-J subunits [2,3]. Apart from the
asymmetry conferred by LHCI, PSI in both cyanobac-
teria and plants is a relatively rounded, featureless
structure when single particles are viewed by electron
microscopy.
PSI-G, which is absent from cyanobacterial PSI,
shares  30% amino-acid identity with PSI-K in
Arabidopsis [4]. Studies using knockout [5,6] and anti-
sense [7] Arabidopsis lines have proposed that PSI-G
plays a role in stabilizing the PSI core and the peri-
pheral antenna, respectively. Additionally, PSI-G may
be an important regulator of PSI activity [6,7]. PSI-K,
on the other hand, appears to be important for sta-

bilizing antenna proteins Lhca2 and -a3 [5,8]. Both
proteins are encoded by the nuclear genome with
N-terminal chloroplast transit (targeting) peptides,
and are predicted to form two transmembrane a-heli-
ces, separated by a charged loop region (Arabidopsis
PSI-G: 6 positive and 7 negative charges; Arabidopsis
PSI-K: 4 positive and 3 negative charges). Topology
studies of barley PSI-K showed that the protein con-
forms to the ‘positive-inside rule’ [9], by having the
positively charged loop in the stroma, with N- and
C-termini in the thylakoid lumen [10]. This finding
agreed with the topology of cyanobacterial PSI-K, as
determined by X-ray crystallography of Synechococcus
elongatus PSI [11,12]. The crystal structure placed
cyanobacterial PSI-K at the outside edge of the com-
plex, a position that has recently been confirmed for
metaphytes, with the publication of a 4.4 A
˚
crystal
structure for Pisum sativum photosystem I [3]. In this
structural model, PSI-G is located on the opposite
edge of the PSI complex from PSI-K [3], which is in
good agreement with biochemical evidence [5,7,13].
The homology between PSI-G and PSI-K, from
which PSI-G probably arose by gene duplication [4],
would suggest a ‘horseshoe’-like topology, with the
loop facing the stroma and the N- and C-termini in
the thylakoid lumen. This topology is also suggested
by the structural model of PSI based on the 4.4 A
˚

crystal structure [3]. However, a resolution of 4.4 A
˚
does not reveal enough structural detail to determine
the actual topology of a membrane protein and so
far there is no biochemical evidence to support the
proposition.
We sought to determine the topology of PSI-G and
to test the feasibility of rescuing Arabidopsis PSI com-
plexes lacking PSI-G with tagged constructs. Successful
introduction of a hexa-histidine (His)- or Strep-tagged
PSI-G into PSI will pave the way for determining the
polypeptide’s location within the complex by means of
immunogold electron microscopy and single-particle
analyses. It will also provide a useful orientation mar-
ker for the otherwise very rounded PSI complex, and
potentially act as an affinity tag for preparation of
ultra-pure PSI particles. His-tags have already been
used to purify active PSII both from Synechocystis
6803 [14] and Chlamydomonas reinhardtii [15] and to
determine the location of PsbH in photosystem II
(PSII) of C. reinhardtii [16]. We now report the expres-
sion of His- and Strep-tagged plant PSI.
Results
A series of cassettes containing tagged variants of
the full-length precursor of Arabidopsis PSI-G (acces-
sion AJ245630) were generated by PCR. Strep-
(WSHPQFEK) or His-tags were inserted in the loop
region, between nucleotide positions 386 and 387, or
at the extreme C-terminus of PSI-G (Fig. 1). These
cassettes were both cloned into binary vectors under

the control of the 35S promoter and terminator, and
into vectors for in vitro transcription and translation.
An Arabidopsis line (psag-1.4 [5], termed DG in this
report), lacking PSI-G due to a transposon footprint
in exon 1, was transformed with the different con-
structs, but to test the ability of the tagged PSI-G to
be correctly targeted and inserted into the thylakoid
membrane, initial analyses were carried out in vitro.
Fig. 1. Schematic representation of PSI-G and the recombinant ver-
sions of PSI-G used in this study. The two transmembrane span-
ning helices are indicated as filled boxes. The positively charged
amino acids in PSI-G are indicated by +. Position of the His-tag
(HHHHHH) and the Strep-tag (WSHPQFEK) in either the loop region
or the C-terminus of PSI-G is indicated. HisT, PSI-G-HisTerm;
StrepT, PSI-G-StrepTerm; HisL, PSI-G-HisLoop; StrepL, PSI-G-Strep-
Loop.
L. Rosgaard et al. Tagged photosystem I
FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS 4003
Targeting of His- and Strep-tagged PSI-G
constructs in vitro
The loop region of PSI-G contains 6 positively charged
amino acids; in agreement with the ‘positive inside
rule’ [9], topology prediction servers such as TMHMM
( [17]) predict PSI-G to have
two transmembrane regions connected by a stromal
loop, with the N- and C-termini in the thylakoid
lumen. In order to test this topology prediction, wild-
type and tagged PSI-G were transcribed and translated
in vitro, then incubated with isolated, intact pea chlo-
roplasts, as described in Experimental procedures.

Analysis of chloroplast fractions postimport (Fig. 2)
shows that all constructs are processed and imported
by isolated chloroplasts. In each case, the full-length
precursor protein (Fig. 2, lanes Tr) is processed to a
smaller polypeptide, corresponding in size to mature
PSI-G (Fig. 2, panels i, and iii–vi, lanes C). This pro-
tein is inside the chloroplasts, because it is protected
from thermolysin digestion (Fig. 2, lanes C+). As a
control, full-length precursor proteins were digested
with thermolysin, to ensure that the mature PSI-G
seen in lanes C+ does not derive from PSI-G bound
to the outside of the chloroplast. None of the precur-
sor proteins yielded a mature-sized degradation prod-
uct when incubated with thermolysin (Fig. 2, lanes
Tr+). wild-type and all tagged PSI-G constructs frac-
tionate with the thylakoid membrane (Fig. 2, lanes T),
but differences become apparent when the thylakoid
membranes are digested with trypsin, a protease that
cleaves after arginine and lysine residues (6 of which
are present in the loop of wild-type PSI-G). Wild-type
PSI-G resists digestion (Fig. 2, panel i, lane T+),
while a control protein, Lhcb1, is digested to a charac-
teristic degradation product, DP (Fig. 2, panel ii, lane
T+). Resistance to trypsin digestion suggests either
that PSI-G’s positive charges (Fig. 1) are on the trans-
side of the thylakoid, or that they are shielded from
trypsin digestion on the cis-side of the membrane. The
A
B
Fig. 2. Determination of the topology of

PSI-G in the thylakoid membrane using
in vitro import experiments. (A) Insertion of
Arabidopsis thaliana wild-type PSI-G or
tagged PSI-G into thylakoids. Shown are
fluorograms of the fractions obtained from
import of radioactive precursors into
isolated, intact pea chloroplasts. The lanes
correspond to: in vitro-translated precursor
(Tr), thermolysin-treated precursor (Tr+),
total, washed chloroplasts immediately post-
import (C), thermolysin-treated chloroplasts
(C+), stromal extract (S), thylakoids (T), and
trypsin-treated thylakoids (T+). Panel i,
wild-type PSI-G; panel ii Lhcb1; panel iii,
PSI-G-HisTerm; panel iv, PSI-G-StrepTerm;
panel v, PSI-G-HisLoop; panel vi,
PSI-G-StrepLoop. DP indicates the charac-
teristic degradation product yielded when
membrane-inserted Lhcb1 is digested by
trypsin. A set of degradation products
yielded by trypsin digestion of thylakoidal
PSI-G-HisLoop is denoted by an asterisk.
(B) Insertion of Chlamydomonas reinhardtii
PSI-G into thylakoids (Lanes as in panel A).
Alignment of the loop region of Arabidopsis
and Chlamydomonas PSI-G. Positively
charged amino acids are shown in italics.
Tagged photosystem I L. Rosgaard et al.
4004 FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS
first scenario is consistent with a topology where the

loop is in the thylakoid lumen, and the N-and C-ter-
mini face the stroma, opposite to that of PSI-K. On
the other hand, the second scenario implies a stromal
loop, like that of PSI-K, but unusally inaccessible to
trypsin – unlike PSI-K [10]. The two constructs tagged
at the C-terminus, PSI-G-HisTerm and PSI-G-Strep-
Term, are also resistant to trypsin digestion of the
thylakoids (Fig. 2, panels iii and iv, lanes T+), but
those tagged in the loop domain, PSI-G-HisLoop and
PSI-G-StrepLoop, are degraded by trypsin (Fig. 2,
panels v and vi, lanes T+). This strongly suggests that
the positively charged loop domain in the latter two
constructs is accessible to trypsin on the stromal side
of the membrane and that placing the tags in the loop
prevents PSI-G from adopting its normal conforma-
tion in the membrane.
Chlamydomonas PSI-G was imported into pea chloro-
plasts to assess whether other PSI-G molecules show
similar topological characteristics (Fig. 2B). Post-
import analysis indicates a trypsin-sensitive loop facing
the stroma, and therefore supports a topology in which
PSI-G has a stromal loop (Fig. 2B, lane T+). An
alignment of the Arabidopsis and Chlamydomonas
PSI-G loop regions is also shown in Fig. 2B, from
which it is evident that the algal loop contains a simi-
lar distribution of positive charges, although fewer
than Arabidopsis (4 instead of 6). Aside from the
charge distributions, the loops exhibit enough variation
to leave room for altered protein–protein interactions,
which may explain why the Chlamydomonas loop

region is exposed to trypsin in the in vitro assay.
The in vitro import experiments suggested that the
loop of Arabidopsis PSI-G faces the stroma. However,
it was also clear that this loop must adopt an unusu-
ally stable structure that is resistant to trypsin diges-
tion under the standard conditions commonly used to
determine the topology of thylakoid membrane pro-
teins. To find out whether the in vitro behaviour was
anomalous, a range of experiments was performed
using thylakoid membranes isolated from wild-type
Arabidopsis (Fig. 3A). wild-type thylakoids were incu-
bated on ice with trypsin for defined periods, then the
thylakoid proteins were separated by SDS ⁄ PAGE and
analysed by immunoblotting using antibodies recogni-
zing PSI subunits with known location and topology.
Quantification of the signal showed that even after a
60 min incubation with trypsin, 60–70% of the PSI-G
protein remains. Yet in the same sample, PSI-K, which
is known to have a stromal loop, is completely diges-
ted. On the other hand, the loop of PSI-O, which is
protected in the thylakoid lumen, is unaffected by the
trypsin treatment, while PSI-D, which is a stromal,
extrinsic PSI subunit, is degraded to a smaller peptide
by the trypsin treatment. In Fig. 3B, a similar experi-
ment has been performed upon PSI complexes purified
from sucrose gradients after solubilization of the
thylakoid membrane using the detergent dodecyl-
b-d-maltoside. Here, the solubilization of the complex
from the membrane clearly renders the PSI-O subunit
accessible to the protease but the PSI-G subunit still

resists complete protease digestion. In fact, only when
the highly active, nonspecific, proteinase K is used, is
it possible to degrade PSI-G significantly (Fig. 3C).
A
C
B
Fig. 3. Protease treatment of thylakoid membranes and PSI com-
plexes isolated from wild-type Arabidopsis. (A) Digestion of thyla-
koid membranes over time. Immunoblot of thylakoid samples after
0, 30 and 60 min digestion with trypsin, probed with antibodies
directed against PSI-G, PSI-K (intrinsic membrane protein, stromal
loop), PSI-O (intrinsic membrane protein, luminal loop) and PSI-D
(extrinsic membrane protein, stromal side). Each lane corresponds
to 2 lg Chl. (B) Digestion of PSI complexes purified from wild-type
Arabidopsis thylakoids after detergent solubilization and sucrose
gradient centrifugation. Immunoblot of undigested (–) PSI com-
plexes and complexes digested with trypsin for 30 min (+). Anti-
bodies used as in part A. Each lane corresponds to 1 lg Chl. (C)
Digestion of thylakoids (2 lg Chl) and PSI complexes (1 lg Chl)
using trypsin and proteinase K. Immunoblots probed with antibod-
ies directed against PSI-G and PSI-D.
L. Rosgaard et al. Tagged photosystem I
FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS 4005
These results support the notion that PSI-G adopts an
unusually protease-resistant structure when associated
with PSI in the thylakoid membrane. Importantly
however, the observation that PSI-G in thylakoid
membranes is partially degraded under conditions
where a protease-susceptible subunit such as PSI-O,
with a known lumenal loop resists degradation,

strongly indicates that the PSI-G loop is stromal.
Expression of Strep- and His-tagged PSI-G
constructs in vivo
To extend the in vitro import experiments to in vivo
conditions, an Arabidopsis mutant in which the psaG
gene has been disrupted by transposon insertion [5]
was transformed with constructs encoding the tagged
versions of PSI-G. DG plants germinate normally, but
grow slightly smaller, paler and flower slightly later
than wild type [5]. However, in our growth chambers,
DG plants display a less pronounced phenotype. In
total, c. 620 T
1
and T
2
plants from two separate trans-
formation experiments were screened by immunoblot-
ting, and approximately 50% of those transformed
with a PSI-G construct (as opposed to an empty vec-
tor), expressed PSI-G or its tagged counterpart.
Thylakoids were prepared from pools of plants
expressing similar levels of PSI-G, and analysed by
immunoblotting, using antibodies to PSI-G and PSI-F,
as an indicator of the relative content of PSI in the
samples. Plants lacking PSI-G have 40% less PSI com-
pared to wild-type [5,7]. Representative thylakoid pre-
parations are shown in Fig. 4A. From the immunoblot
it is clear that the steady state level of PSI-G, expressed
as the PSI-G to PSI-F ratio, in all the transformed lines
is lower than in the true wild type. This suggests that

the expression of PSI-G and steady state level of PSI-G
in all the transformed lines is suboptimal.
High resolution SDS ⁄ PAGE shows that PSI-G bear-
ing a His- or Strep-tag migrates more slowly than
wild-type PSI-G, with all tagged constructs behaving
similarly, typified in Fig. 4B. That these bands repre-
sent PSI-G carrying the appropriate tag was confirmed
by probing gel blots with an anti-hexa-His antibody or
StrepTactin-HRP conjugated with horseradish peroxi-
dase. The results for the tagged constructs are shown
in Fig. 4C. Interestingly, both tags are recognized
more efficiently in the context of the PSI-G loop
region than at the C-terminus.
Topology of PSI-G in vivo
In order to examine the topology of PSI-G in vivo,
samples of thylakoids from wild-type (empty vector)
and transformed plants were treated separately with
thermolysin and trypsin as described for the in vitro
import analyses. PSI-G, and a control protein, Lhcb1,
were then detected by immunoblotting (Fig. 5). Wild-
type and C-terminally tagged PSI-G resist digestion
by either protease whereas PSI-G-HisLoop and PSI-
G-StrepLoop are sensitive to both proteases. Lhcb2
is, as expected, clipped by both proteases. The beha-
viour of the wild-type and tagged constructs in vivo
exactly parallels the in vitro results, and suggest that
PSI-G carrying either a C-terminal or a loop His- or
Strep-tag is able to insert into the thylakoid mem-
brane with the same topology as wild-type PSI-G.
The versions of PSI-G that carry a His- or a Strep-

tag in the loop are sensitive to digestion by the pro-
teases, whereas the C-terminally tagged versions
remain as protease-resistant as the wild-type protein.
Thus, the loop-tags disrupt the structure of the loop
A
B
C
Fig. 4. Tagged PSI-G is present in thylakoids of DG plants trans-
formed with the various PSAG constructs. (A) Immunoblot of
pooled thylakoids (0.5 lg Chl per lane) from lines expressing wild-
type or tagged PSI-G, or lacking PSI-G (DG). PSI-F is detected as an
indicator of PSI content. The ratio PSI-G ⁄ PSI-F is indicated under
each lane (nd, not determined). (B) Comparison of the migration of
wild-type and His-tagged PSI-G by high-resolution SDS ⁄ PAGE.
(C) Immunodetection of His and Strep tags in PSI-G constructs.
Tagged photosystem I L. Rosgaard et al.
4006 FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS
and ⁄ or its interaction with other proteins such as the
PSI core.
Incorporation of tagged PSI-G into PSI
Whilst it is evident that both C-terminally and loop-
tagged PSI-G can insert into the thylakoid in vivo
(Figs 4 and 5), it is possible that the tags hinder assem-
bly into PSI complexes. Therefore, it was of interest to
find out whether any of the tagged proteins are assem-
bled into PSI complexes. We solubilized thylakoid
membranes using dodecyl-b-d-maltoside and purified
PSI complexes using sucrose gradient centrifugation.
The purified PSI complexes were then analysed by
immunoblotting, using antibodies against PSI-G and

PSI-F (Fig. 6A). That PSI-G contained the appropri-
ate tag was confirmed by probing gel blots with an
antihexa-His antibody or StrepTactin-HRP conjugated
with horseradish peroxidase (Fig. 6B). The results
clearly indicate that both the C-terminally and
loop-tagged versions of PSI-G are present in PSI, sug-
gesting that all versions of PSI-G can be incorporated
into the PSI complex. However, C-terminally His- and
Strep-tagged PSI-G seem to incorporate to a higher
degree than the loop-tagged versions of PSI-G. This
may suggest that the loop of PSI-G is important for
stable integration of the subunit into the PSI complex.
A protein band smaller that the tagged versions of
PSI-G is present in the lanes with the loop-tagged PSI,
but is also seen faintly in the lanes with the terminally
tagged PSI. The respective tags were only present in
the upper band (Fig. 6B) and we have no reason to
believe that the lower band is a degradation product
of the tagged versions of PSI-G. Cross-contamination
with wild-type PSI-G during the process of preparing
thylakoids or PSI particles can also be ruled out as the
double band could be detected in at least two inde-
pendent preparations of both thylakoids and PSI parti-
cles from the four lines carrying the tagged versions of
PSI-G. The most likely explanation is therefore that
the psaG transposon knock-out line used for the trans-
formation experiments is unstable and apparently a
fraction of the cells within the plant revert to wild
type, giving rise to the wild-type-sized immuno-detect-
able band. It seems that transformation of the trans-

poson-tagged psaG knock-out line somehow increases
this reversion rate.
Fig. 5. Tagged PSI-G inserts into the thylakoid membrane in vivo.
Immunoblot of thylakoids (2 lg Chl per lane) isolated from plant
lines transformed with the various constructs and subsequently
subjected to protease treatment. The lanes correspond to:
untreated thylakoids (T), thylakoids treated with thermolysin (P1)
and thylakoids treated with trypsin (P2). DP denotes the character-
istic degradation product of membrane-inserted Lhcb1 digested by
thermolysin or trypsin.
A
B
Fig. 6. The tagged PSI-G subunit is incorporated into photosystem I.
(A) Immunoblot of PSI complexes (1 lg Chl per lane) purified from
thylakoids shown in Fig. 5. The ratio of PSI-G : PSI-F is shown
beneath the lanes (nd, not determined). B: Immunoblot of PSI com-
plexes (1 lg Chl per lane) purified from thylakoids and probed with
Anti-Hexa-His tag or Anti-Strep tag Ig, as indicated alongside the
panels.
L. Rosgaard et al. Tagged photosystem I
FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS 4007
Discussion
We have successfully incorporated His- and Strep-
tagged PSI-G into Arabidopsis PSI complexes in vivo.In
achieving this, we have been able to collect multiple lines
of evidence that PSI-G inserts into the thylakoid mem-
brane with a stromal loop and its N- and C-termini
facing the lumen. PSI-G is markedly protected from
trypsin degradation, both in vitro and in vivo, which
means that the positively charged loop region is not eas-

ily accessible to protease from the stromal face of the
membrane. By contrast, PSI-K is cleaved by trypsin into
two fragments, representing the two transmembrane
spans [10]. However, PSI-G constructs with a His- or
Strep-tag in the loop region are sensitive to proteases,
indicating that the loop region becomes more accessible,
presumably by disrupting protein–protein interactions
within PSI, or by altering the conformation of the loop.
His-tags were employed in a topological study of the
major light harvesting chlorophyll a ⁄ b binding apopro-
tein, Lhcb1 [18]. Interestingly, fusion of a His-tag to the
C-terminus, which must cross the thylakoid membrane
during insertion, did not prevent Lhcb1 adopting its cor-
rect topology. The most likely explanation of our own
experimental observations is that the C-terminal His-
and Strep-tags are also translocated across the thylakoid
membrane, such that PSI-G has the same topology as
PSI-K. This means that PSI-G obeys the ‘positive-inside
rule’ [9], which states that the topology of membrane
proteins is dictated by the distribution of positive
charges, which resist translocation across membranes.
Our findings provide the first biochemical evidence
for the predictions made by Ben-Shem et al. [3] in the
interpretation of their crystal structure of Pisum
sativum PSI, in which the loop of PSI-G is suggested
to face the stroma.
The amount of PSI-G in the transformed knock-out
line does not reach the level of the wild type. The lines
expressing the tagged versions of PSI-G all accumulate
significantly lower amounts of PSI-G than wild-type;

however, the data do not allow us to conclude that the
loop-tagged versions of PSI-G accumulate to a lesser
extent than the C-terminally tagged versions, although
there is a weak tendency. Even the line transformed
with the wild-type version of PSI accumulates less
PSI-G than wild type. This is a somewhat surprising
result as the 35S promoter used in this study should
ensure strong constitutive expression of the gene. A
likely explanation for this is that the transposon-tagged
psaG gene in the knock-out line used for the transforma-
tion experiments still produces a psaG transcript, and
together with the strong constitutive expression of
tagged or wild-type psaG transcripts, causes cosuppres-
sion and subsequent accumulation of less PSI-G
protein. Preliminary experiments in which individual
transformants were found to accumulate near-wild-
type levels of PSI-G at the age of 7–8 weeks after ger-
mination, show that the transformants lost expression
4–5 weeks later (results not shown).
In conclusion, we have shown using both in vitro and
in vivo methods, that PSI-G adopts a topology in the
thylakoid membrane with the loop facing the stroma
and its N- and C-termini facing the lumen. We have also
shown that plants lacking PSI-G can be transformed
with tagged versions of PSI-G, albeit limited by the
expression level of PSI-G in the transformants. Finally,
we have demonstrated that His- or Strep-tagged PSI can
be made in planta. In future experiments, we will evalu-
ate its use for quick purification of PSI and alignment of
PSI particles during structural determination of PSI

using electron microscopy and single particle analysis.
Experimental procedures
PSAG constructs
Constructs encoding tagged variants of Arabidopsis PSAG
(accession number AJ245630) were prepared by polymerase
chain reaction, using the EST 279G1T7 (obtained from the
ABRC, Ohio, USA and described in [7]) as a template.
Primers were designed as follows: PSAG with a C-terminal
hexa-histidine tag (PSI-G-HisTerm), 5¢-GCGGAGCTCAT
GGCCACAAGCGCATCAGC-3¢ and 5¢-GCGGCATGCT
CA
GTGGTGGTGGTGGTGGTGTCCAAAGAAGCTTG
GGTCGTAT-3¢ (His-tag underlined); PSAG with a C-ter-
minal Strep-tag (PSI-G-StrepTerm), 5¢-GCGGAGCTCAT
GGCCACAAGCGCATCAGC-3¢ and 5¢-GCGGCATGCT
CA
TTTTTCGAACTGCGGGTGGCTCCATCCAAAGAA
GCTTGGGTCGTAT-3¢ (region encoding the Strep-tag,
WSHPQFEK [19], underlined). Constructs containing a
His- or Strep-tag in the loop region (PSI-G-HisLoop and
PSI-G-StrepLoop) were prepared in three stages: a primary
amplification with primers 5¢-GCGGAGCTCATGGCCAC
AAGCGCATCAGC-3¢ and 5¢-
GTGGTGGTGGTGGTGG
TGGAAATGGGTTTTTCCGTTCTGC-3¢ (His-tag under-
lined) or 5¢-
TGGAGCCACCCGCAGTTCGAAAAAGAA
GCTGGAGATGATCGTGCT-3¢ (Strep-tag underlined),
then a secondary amplification with primers 5¢-
CACCAC

CACCACCACCACGAAGCTGGAGATGATCGTGCT-3¢
(His-tag underlined) or 5¢-
TGGAGCCACCCGCAGTTCG
AAAAAGAAGCTGGAGATGATCGTGCT-3¢ (Strep-tag
underlined) and 5¢-GCGGCATGCTCATCCAAAGAAGC
TTGGGTCG-3¢. The two amplified fragments were used
as a combined template for the tertiary amplification,
using primers 5¢-GCGGAGCTCATGGCCACAAGCGCA
Tagged photosystem I L. Rosgaard et al.
4008 FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS
TCAGC-3¢ and 5¢-GCGGCATGCTCATCCAAAGAAGC
TTGGGTCG-3¢ for both PSI-G-HisLoop and PSI-G-Strep-
Loop.
Tagged PSAG constructs were cloned either into
pGEM
Ò
4Z (Promega GmbH, Germany) under the SP6 pro-
moter, for in vitro transcription and translation, or into
pPS48 [20] under the control of the 35S promoter and termi-
nator. For each tagged variant and the wild-type PSAG,a
cassette containing the PSAG construct flanked by the 35S
promoter and terminator was excised and ligated into the
binary vector pPZP111 [21], ready for transformation of
Agrobacterium tumefaciens. All DNA constructs were fully
sequenced to confirm their identities before experimental use.
In vitro import assays
PSAG constructs were transcribed in vitro using SP6 RNA
polymerase, then translated in a Wheat Germ Lysate system
(Promega GmbH, Germany), in the presence of [
3

H]leucine
(Amersham Biosciences, Denmark). Intact chloroplasts were
isolated from pea seedlings, and in vitro import assays were
carried out as described in [22]. Samples were analysed by
Tricine-SDS ⁄ PAGE [23] and fluorography.
Transformation of Arabidopsis with tagged
PSAG constructs
Wild-type plants were Arabidopsis thaliana, ecotype Colum-
bia 0. The Arabidopsis PSI-G knock-out line (DG [5],
Columbia 0 background) was generously provided by
Dr. D. Leister, Max Planck Institute, Cologne, Germany.
Prior to transformation, plants were screened for the pres-
ence or absence of PSI-G by immunoblotting. Five DG
plants per construct were subjected to Agrobacterium-medi-
ated transformation, using the floral dip method [24]. Five
wild-type plants were transformed with an empty pPZP111
vector. Seeds from transformed plants were surface-steril-
ized in 5% (v ⁄ v) sodium hypochlorite, 0.02% (v ⁄ v) Triton
X-100, washed with sterile water and plated on MS medium
supplemented with 50 lgÆmL
)1
kanamycin. Kanamycin-
resistant seedlings were transferred to soil and subsequently
analysed for the expression of PSAG and the tagged con-
structs by immunoblotting. That individual transformants
contained the correct construct was confirmed by PCR
amplification of genomic DNA using primers complement-
ary to the 35S promoter and terminator, followed by DNA
sequencing of the amplicons.
Immunoblot analysis of transgenic Arabidopsis

For screening of the transformants, one mature leaf was
excised from each individual plant, placed in an Eppendorf
tube and frozen in liquid nitrogen. Frozen tissue was
pulverized in 200 lL protein extraction buffer [PEB: 100 mm
Tris ⁄ HCl, pH 8.0, 50 mm EDTA, pH 8.0, 250 mm NaCl,
0.7% SDS, 1 mm dithiothreitol, 1· Complete Protease
Inhibitor Cocktail (Roche)], using a pestle. The sample was
then incubated at 68 °C for 10 min, followed by centrifuga-
tion at 15 000 g for 10 min at 4 °C. The supernatant was
removed, transferred to a fresh tube, and its chlorophyll con-
tent estimated by measuring light absorbance at 652 nm [25].
Protein equivalent to 2 lg chlorophyll was acetone-precipita-
ted before being separated by SDS ⁄ PAGE and transferred to
nitrocellulose. Antibodies employed were rabbit polyclonals
against PSI-G, PSI-F and a monoclonal anti-His tag anti-
body (Novagen, Merck Biosciences GmbH, Germany). The
Strep tag was detected on blots using StrepTactin coupled to
Horse Radish Peroxidase (Bio-Rad, Herlev, Denmark). Sam-
ples equivalent to 0.125, 0.250 and 0.500 lg chlorophyll were
analysed for quantitative immunoblot analysis of thylakoid
PSI proteins. For immunoblotting of purified PSI particles,
samples containing 0.5 and 1 lg chlorophyll were analysed.
Isolation of thylakoid membranes and PSI
particles from Arabidopsis
Healthy leaves from typically 5–10 Arabidopsis plants were
pooled. Thylakoid membranes were isolated according to
[26]. PSI particles were isolated from thylakoids after solu-
bilization with dodecyl-b-d-maltoside and sucrose density
ultracentrifugation, as described in [8].
Trypsin treatment of thylakoids and PSI particles

Thylakoid membranes equivalent to 2 lg Chl and PSI com-
plexes equivalent to 1 lg Chl were treated with trypsin (Sig-
ma, type XIII) on ice at 0.25 mgÆmL
)1
final concentration.
Thylakoid digestions were carried out in 10 mm
Hepes ⁄ KOH, pH 8.0, 5 mm MgCl
2
(HM) and PSI diges-
tions were carried out in 20 mm Tricine ⁄ NaOH, pH 7.5,
0.06% dodecyl-b-maltoside. PSI incubations were stopped
by addition of soybean trypsin inhibitor (Sigma type I-S) to
a final concentration of 1 mgÆmL
)1
and boiling loading buf-
fer. In the case of thylakoids, the samples were washed in
400 lL HM and the pellets resuspended in trypsin inhibitor
and boiling loading buffer. Samples were loaded on
SDS ⁄ PAGE gels for immunoblot analysis.
Acknowledgements
We thank the ABRC at Ohio State University for pro-
viding ESTs and Lis Drayton Hansen for excellent
technical assistance. We are grateful to Dr D. Leister
for the gift of the Arabidopsis PSI-G knock-out line,
Prof J D. Rochaix for the gift of the cDNA clone
encoding Chlamydomonas PSI-G and to Dr A. Ben-
Shem and Prof N. Nelson for sharing unpublished
L. Rosgaard et al. Tagged photosystem I
FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS 4009
data with us. The Danish National Research Founda-

tion, the Danish Veterinary and Agricultural Research
Council (23-03-0105) and the EU (Contract No
HPRN-CT-2002–00248) are gratefully acknowledged.
References
1 Jensen PE, Haldrup A, Rosgaard L & Scheller HV
(2003) Molecular dissection of photosystem I in higher
plants: topology, structure and function. Physiol Plant
119, 313–321.
2 Boekema EJ, Jensen PE, Schlodder E, van Breemen JFL,
van Roon H, Scheller HV & Dekker JP (2001) Green
plant photosystem I binds light harvesting complex I on
one side of the complex. Biochemistry 40, 1029–1036.
3 Ben-Shem A, Frolow F & Nelson N (2003) Crystal
structure of plant photosystem I. Nature 426, 630–635.
4 Kjærulff S, Andersen B, Skovgaard Nielsen V, Lindberg
Møller B & Okkels JS (1993) The PSI-K subunit of
photosystem I from barley (Hordeum vulgare L.): Evi-
dence for a gene duplication of an ancestral PSI-G ⁄ K
gene. J Biol Chem 268, 18912–18916.
5 Varotto C, Pesaresi P, Jahns P, Lessnick A, Tizzano M,
Schiavon F, Salamini F & Leister D (2002) Single and
double knock-outs of the genes for photosystem I subu-
nits G, K and H of Arabidopsis. Effects on photosystem
I composition, photosynthetic electron flow and state
transitions. Plant Physiol 129, 616–624.
6 Zygadlo A, Jensen PE, Leister D & Scheller HV (2005)
Photosystem I lacking the PSI-G subunit has a higher
affinity for plastocyanin and is sensitive to photo-
damage. Biochim Biophys Acta 1708, 154–163.
7 Jensen PE, Rosgaard L, Knoetzel J & Scheller HV

(2002) Photosystem I activity is increased in the absence
of the PSI-G. J Biol Chem 277, 2789–2803.
8 Jensen PE, Gilpin M, Knoetzel J & Scheller HV (2000)
The PSI-K subunit of photosystem I is involved in the
interaction between light-harvesting complex I and the
photosystem I reaction center core. J Biol Chem 275,
24701–24708.
9 Gavel Y, Steppuhn J, Herrmann R & von Heijne G
(1991) The ‘positive-inside rule’ applies to thylakoid
membrane proteins. FEBS Letts 282, 41–46.
10 Mant A, Woolhead CA, Moore M, Henry R & Robinson
C (2001) Insertion of PsaK into the thylakoid membrane
in a ‘horseshoe’ conformation occurs in the absence of
signal recognition particle, nucleoside triphosphates, or
functional Albino3. J Biol Chem 276, 36200–36206.
11 Fromme P, Jordan P & Krauss N (2001) Structure of
photosystem I. Biochim Biophys Acta 1507, 5–31.
12 Jordan P, Fromme P, Witt HT, Klukas O, Saenger W
& Krauss N (2001) Three-dimensional structure of
cyanobacterial photosystem I at 2.5 A
˚
resolution.
Nature 411, 909–917.
13 Jansson S, Andersen B & Scheller HV (1996) Nearest-
neighbor analysis of higher-plant photosystem I holo-
complex. Plant Physiol 12, 409–420.
14 Bricker TM, Morvant J, Masri N, Sutton HM & Fran-
kel LK (1998) Isolation of a highly active photosystem
II preparation from Synechocystis 6803 using a histi-
dine-tagged mutant of CP47. Biochim Biophys Acta

1409, 50–57.
15 Sugiura M, Inoue Y & Minagawa J (1998) Rapid and
discrete isolation of oxygen-evolving His-tagged photo-
system II core complex from Chlamydomonas reinhardtii
by Ni
2+
affinity column chromatography. FEBS Lett
426, 140–144.
16 Bu
¨
chel C, Morris E, Orlova E & Barber J (2001) Locali-
sation of the PsbH subunit in photosystem II: a new
approach using labelling of His-tags with a Ni(
2+
)-NTA
gold cluster and single particle analysis. J Mol Biol 312,
371–379.
17 Kosemund K, Geiger I & Paulsen H (2000) Insertion of
light-harvesting chlorophyll a ⁄ b protein into the thyla-
koid. Topographical studies. Eur J Biochem 267, 1138–
1145.
18 Krogh A, Larsson B, von Heijne G & Sonnhammer
ELL (2001) Predicting transmembrane protein topology
with a hidden Markov model: Application to complete
genomes. J Mol Biol 305, 567–580.
19 Skerra A & Schmidt TGM (2000) Use of the Strep-tag
and streptavidin for recombinant protein purification
and detection. Methods Enzymol 326, 271–304.
20 Kay R, Shan A, Daly M & McPherson J (1987) Duplica-
tion of CaMV 35S promoter sequences creates a strong

enhancer for plant genes. Science 236, 1299–1302.
21 Hajdukiewicz P, Svab Z & Maliga P (1994) The small,
versatile pPZP family of Agrobacterium binary vectors
for plant transformation. Plant Mol Biol 25, 989–994.
22 Robinson C & Mant A (2002) Import of proteins into
isolated chloroplasts and thylakoid membranes. In
Molecular Plant Biology Volume Two: a Practical
Approach (Gilmartin, P & Bowler, C, eds), pp. 123–146.
Oxford University Press, Oxford, UK.
23 Scha
¨
gger & Von Jagow (1987) Tricine-sodium dodecyl
sulfate-polyacrylamide gel electrophoresis for the
separation of proteins in the range from 1 to 100 kDa.
Anal Biochem 166, 368–379.
24 Clough JC & Bent AF (1998) Floral dip: a simplified
method for Agrobacterium-mediated transformation of
Arabidopsis thaliana . Plant J 16, 735–743.
25 Lichtenthaler HK (1987) Chlorophylls and carotenoids:
Pigments of photosynthetic biomembranes. Methods
Enzymol 148, 350–382.
26 Haldrup A, Naver H & Scheller HV (1999) The inter-
action between plastocyanin and photosystem I is
inefficient in transgenic Arabidopsis plants lacking the
PSI-N subunit of photosystem I. Plant J 17, 689–698.
Tagged photosystem I L. Rosgaard et al.
4010 FEBS Journal 272 (2005) 4002–4010 ª 2005 FEBS