Domain Organization of Long Signal Peptides of Single-Pass Integral Membrane Proteins Reveals Multiple Functional Capacity
et al. (2008) Domain Organization of Long Signal Peptides of Single-Pass Integral
Membrane Proteins Reveals Multiple Functional Capacity. PLoS ONE 3(7): e2767. doi:10.1371/journal.pone.0002767
Domain Organization of Long Signal Peptides of Single- Pass Integral Membrane Proteins Reveals Multiple Functional Capacity
Janet Kelso, Max Planck Institute for Evolutionary Anthropology, Germany
0 Centre for Membrane Proteomics, Institute of Cell Biology and Neuroscience, Goethe-University , Frankfurt am Main , Germany
Targeting signals direct proteins to their extra - or intracellular destination such as the plasma membrane or cellular organelles. Here we investigated the structure and function of exceptionally long signal peptides encompassing at least 40 amino acid residues. We discovered a two-domain organization (''NtraC model'') in many long signals from vertebrate precursor proteins. Accordingly, long signal peptides may contain an N-terminal domain (N-domain) and a C-terminal domain (C-domain) with different signal or targeting capabilities, separable by a presumably turn-rich transition area (tra). Individual domain functions were probed by cellular targeting experiments with fusion proteins containing parts of the long signal peptide of human membrane protein shrew-1 and secreted alkaline phosphatase as a reporter protein. As predicted, the N-domain of the fusion protein alone was shown to act as a mitochondrial targeting signal, whereas the C-domain alone functions as an export signal. Selective disruption of the transition area in the signal peptide impairs the export efficiency of the reporter protein. Altogether, the results of cellular targeting studies provide a proof-of-principle for our NtraC model and highlight the particular functional importance of the predicted transition area, which critically affects the rate of protein export. In conclusion, the NtraC approach enables the systematic detection and prediction of cryptic targeting signals present in one coherent sequence, and provides a structurally motivated basis for decoding the functional complexity of long protein targeting signals.
Funding: This research was supported by the Beilstein-Institut zur Fo rderung der Chemischen Wissenschaften, the Centre for Membrane Proteomics at
GoetheUniversity, Frankfurt am Main, the Sonderforschungsbereich 579 (RNA-Ligand Interactions, project A11.2), the Sonderforschungsbereich 628 (Functional
Membrane Proteomics, project p7), and the Deutsche Forschungsgemeinschaft Sta 187/16-1.
Competing Interests: The authors have declared that no competing interests exist.
. These authors contributed equally to this work.
" These authors also contributed equally to this work.
Targeting signals are contiguous stretches of amino acids that
direct proteins to their sub-cellular destinations or the extracellular
space . With few exceptions, the vast majority of extracellular
proteins are exported from mammalian cells via the endoplasmic
reticulum (ER) secretory pathway . While most signal sequences
are N-terminally located, deviant examples have been reported with
internal signals like in human UDP-glucuronosyltransferase , or
bacterial C-terminal secretion signals like in virulence factor from
Mycobacterium tuberculosis , and Escherichia coli (E. coli) haemolysin .
Canonical N-terminal signals are processed by signal peptidases
. The sequence similarity among these cleavable signal peptides
coding for the ER and subsequent protein export is low as they do
not share common residue motifs but rather possess common
physicochemical features coding for the appropriate cellular
compartment [7,8]. Signal recognition by the cellular decoding
machinery may include multiple recognition events [9,10]. This
renders perfect in silico prediction of subcellular locations and the
detection of targeting signals still impossible although many
encouraging attempts have been made . For example, to
counter the dissimilarity in signal peptides for prediction processes,
the amino acid composition has been taken into account resulting in
improved accuracy [8,17,18]. Despite their dissimilarity,
N-terminally located targeting sequences are sometimes interchangeable
between proteins in eukaryotes and even between different
kingdoms. One such example is Escherichia coli (E. coli) beta-lactamase,
which can be exported by Xenopus oocytes . Still, general signal
interchangeability cannot be postulated [20,21]. Public web servers
are available for predicting the subcellular localization of proteins in
various organisms, for example Cell-PLoc (http://chou.med.harvard.
edu/bioinf/Cell-PLoc/)  or the SignalP suite (http://www.cbs.
In eukaryotes, a canonical N-terminally located protein export
signal typically contains three distinguishable parts: a positively
charged N-terminal section (n-region), a hydrophobic core (h-region),
and a signal peptidase recognition site (c-region) [8,11]. The
approximate average length of such signal peptides is 22 amino
acid residues . While the c-region typically consists of five
residues, both the h- and the n-region show more variability in length.
This variability has been suggested to enable alternative functions
[10,24]. In fact, much longer examples of signal peptides are known
to exhibit additional functions besides precursor targeting
[10,25,26], for example regulation of the protein export rate as
described for interleukin-15 , or signal peptide accumulation in
the nucleoli in the case of mouse mammary tumor virus Rem protein
after release from the endoplasmic reticulum .
In the present study, we introduce a structurally motivated
modularization of long signal peptides into separate functional
modules, and demonstrate the actual functional relevance of this
concept for the long signal peptide of the integral membrane protein
shrew-1 (SH) as an example. Shrew-1 was originally isolated from an
epithelial-like cell line obtained from an endometriosis biopsy . It
contains a cleavable N-terminal signal peptide of 43 residues , an
extracellular domain (residues 44282), a transmembrane segment
(residues 283303) and a cytoplasmic domain (residues 304411).
Shrew-1 is transported to the basolateral part of the plasma
membrane in polarized epithelial cells and interacts with the
Ecadherin mediated adherens junction complex [29,31]. In
nonpolarised cells, like transformed epithelial cells, shrew-1 also displays
plasma membrane localization, though apparently less polarized.
Shrew-1 appears to be involved in the regulation of cell invasion and
motility and, in line with this, interacts with protein CD147, a known
promoter of invasiveness .
Based on proteome analysis by machine-learning systems, we
propose a bipartite domain model (NtraC model) of long signal
peptides from single-pass integral membrane proteins. According
to this model, such long signal peptides may contain two separate
functional domains: an N-terminal domain (N-domain) and a
C-terminal domain (C-domain) traceable by a turn-rich linker
area connecting both. We denote this linker element transition
area (tra). Proof-of-principle for the validity of the NtraC domain
model is provided by in vitro targeting experiments with shrew-1.
Many single-spanning integral membrane proteins
possess long signal peptides with a bipartite domain
Analysis of long signal peptides was performed in two steps:
First, potential domains were predicted using a novel
machinelearning technique for turn prediction . Potential
turncontaining regions were found to be predominantly located in
the central portion of these long signals. Based on the location of
this transition area, long signal peptides were dissected into two
parts, an N-terminal (N) and a C-terminal (C) fragment. Then,
the resulting sequence fragments were scrutinized for potential
targeting functions. The concept of this NtraC model of signal
peptide organization is based on the hypothesis that the two
functional modules in a long signal peptide may exhibit
individually distinct tasks in the context of protein targeting. This
requires a minimal peptide length, and for the present study we
decided to focus only on signal peptide domains containing
conventional signals with an expected average length of
approximately 20 residues each. This choice is motivated by the observed
average length of targeting signals coding for a single
compartment . Certainly, we cannot exclude the existence of other
targeting signals of hitherto unknown structure (e.g., unusually
short signals) within long signal peptides.
Searching for long signal peptides ($40 residues) in the
UniProtKB database (release 53.2)  yielded 296 vertebrate
proteins, including homologues. All sequences were analyzed with
regard to their potential NtraC organization. Within our NtraC
analysis software, predictions for potential targeting signals were
done using the software SignalP 3.0  (signals coding for protein
transport into the ER, signal peptide and signal anchor prediction)
and TargetP  (signals coding for mitochondrial import).
Potential turn-forming elements were detected using our software
tool SVMTurn (www.modlab.deRSoftwareRSVMTurn) .
SVMTurn uses Support Vector Machine classifiers for recognition
of various turn types in amino acid sequences. Turns with
intramolecular hydrogen bonds encompassing four, five, and six
residues are predicted with approximately 80% accuracy.
According to NtraC (www.modlab.deRSoftwareRNtraC)
analysis, 185 of 296 (62%) long signal peptides obey the NtraC domain
organization with a C-domain coding for an ER targeting signal
(Suppl. Table S1). We found no strict conservation of turn residues in
all 185 sequences. As expected for beta-turns, Gly is overrepresented
at residue position 3 of a regular beta turn . 45 of thee 185
candidate proteins possess both an N-domain coding for a putative
mitochondrial transit peptide and a C-domain coding for an
endoplasmic reticulum (ER) targeting signal (Figure 1). For 13 of
these sequences, signal peptidase cleavage sites were not predicted.
Thus, they might act as signal anchors. All 32 remaining candidates,
which show a predicted domain combination analogous to shrew-1
(N-Domain: mTP, C-domain: SP) and posses a predicted signal
peptidase cleavage site, are listed in Table 1. The C-domains of the
remaining 140 NtraC-organized sequences code for ER targeting. In
contrast to shrew-1, however, their N-domains may contain an
additional feature or targeting function that is different from
conventional mitochondrial targeting signals.
To check the influence of a potential bias in these results due to
clusters of homologues in the set of 296 candidate genes, we
manually eliminated all orthologues. This procedure did not affect
the ratio of NtraC-organized vs. non-NtraC-organized samples
(Figure 1, values in brackets). In the human genome alone, we
found 105 signal peptides with $40 residues overall, among which
71 (68% of 105) are NtraC-organized.
We provide a public web service for NtraC analysis of amino
acid sequences (www.modlab.deRSoftwareRNtraC) and invite
the scientific community to scrutinize our NtraC domain model
using this prediction server.
Proteins with NtraC-organized signal sequences apparently have
common features. 19 of the 32 candidate sequences are annotated in
UniProt as type-I membrane proteins containing a single potential
transmembrane segment (TMS). Among these, the only
experimentally validated TMS is the one of shrew-1 , which was a clear
motivation for us to use this protein for the cellular proof-of-principle
study. We then performed TMS predictions for the 13 remaining
sequences using the software tools Phobius  and SVMtm ,
which in all cases gave rise to the same results: Two proteins yielded
strong positive scores indicating the likely presence of a TMS, three
received weaker scores favoring TMS presence, and eight are
seemingly devoid of a TMS. These results increase the number of
candidate proteins from 19 to 24 out of 32, corresponding to 75% as
a conservative estimation.
Summarizing, we identified a class of long signal peptides
distinguished by the NtraC domain architecture. This structural
and functional organization is present in signal peptides of many
single-pass membrane proteins. For further study, we selected one
of these proteins, human shrew-1 as an example.
Experimental system for assessment of prediction results:
Shrew-1 signal peptide and SEAP reporter protein
Based on the theoretical analysis described in the previous
paragraph, we used secreted alkaline phosphatase (SEAP) as a
reporter protein in order to probe the targeting capacity of the
predicted domains of shrew-1s signal peptide. The SEAP reporter
Figure 1. Overview of NtraC-organized sequences among long signal sequences found in vertebrate proteins. Set sizes without
orthologues are given in brackets. The numbers represent conservative estimates based on validated prediction tools for targeting signal recognition
and turn structure prediction.
system allows for the exchange of the intrinsic signal peptide by
other potential signal peptide sequences, which can then be tested
for biological activity . SEAP is a glycoprotein which becomes
N-glycosylated by oligosaccharyl transferase located in the ER
. Therefore, its N-glycosylation status is an indication of
translocation into the ER lumen, which in turn is a prerequisite for
SEAP secretion into the supernatant.
The C-domain acts as a secretion signal. According to the
NtraC model, the shrew-1 signal peptide (residues 143 ,
SignalP 3.0 probability = 0.95) is divided into three domains: It
contains an N-domain (residues 119) and a C-domain (residues
2043) connected by the transition area (residues 1624). The
Cdomain is predicted as a standard secretion signal containing an n-,
h-, and c-region (SignalP 3.0 probability = 0.9), whereas the
Ndomain receives a prediction as a mitochondrial transit peptide
(TargetP probability = 0.3).
Within the transition area, three adjacent and partly overlapping
b-turns were predicted (positions 1624). Interestingly, no further
bturns were found in the remainder of the signal peptide. The position
of the turns appears to be evolutionary conserved among different
species, as shown by a multiple sequence alignment of seven
vertebrate shrew-1 homologues, suggesting a fundamental functional
importance of this region (Suppl. Figure S1).
To functionally test the predicted signal peptide domains, six
constructs coding for different SEAP fusion proteins were devised
(Figure 2). They were transfected into HEK 293T cells, and SEAP
activity was determined in both the supernatants and in whole cell
As shown in Figure 3A, the C-domain (SHC-SEAPDSP) alone is
able to direct SEAP fusion protein to the supernatant. The
Ndomain (SHN-SEAPDSP) alone does not have this targeting capacity.
The same holds for the whole cell lysates (Figure 3A, white bars).
Compared to full length shrew-1 signal peptide (SH-SEAPDSP),
SEAP activity in both the supernatant and whole cell lysates of
SHC-SEAPDSP transfected cells was decreased to about one third.
This implies that the full-length signal peptide is required for full
export efficiency, but basic targeting information is encoded in the
C-domain of the long signal peptide.
Notably, both fusion proteins were detectable by Western blotting
(Figure 3B). This raises the question for the reason of inactivity of the
N-domain containing protein. One explanation would be impaired
translocation from the cytosol into the ER, which in turn should have
resulted in lacking N-glycosylation of SEAP. To check this
hypothesis, we subjected the lysates to PNGase F treatment, which
removes N-linked glycans that are selectively found on
ERtranslocated active protein. Figure 3B shows that the SHN-SEAPDSP
protein is not N-glycosylated (lanes 7 and 8), whereas SHC-SEAPDSP
and SH-SESPDSP contain an N-glycosylated SEAP population (lanes
3 and 5, band marked by an asterisk). We conclude that
SHNSEAPDSP was not transported into the ER. It is noteworthy that
SHN-SEAPDSP was found in two non-glycosylated bands (lanes 7 and
8), indicating the existence of two populations with different
molecular mass. The position of the bands is in line with the idea
that the upper band contains the N-domain of the signal peptide,
which might have been cleaved off in the faster migrating protein
(lower band) by some non-ER protease activity.
The N-domain directs the reporter protein to
mitochondria. The observation of two non-glycosylated
bands in the Western blot analysis raised the question, whether
the SHN-SEAPDSP fusion protein is able to target to mitochondria,
as predicted by our sequence analysis (vide supra). Therefore, we
analyzed mitochondrial localization of SHN-SEAPDSP. HEK 293T
cells were transfected with either SHN-SEAPDSP or SHC-SEAPDSP,
and mitochondria were isolated by differential centrifugation
followed by density gradient centrifugation. Cytosolic (cyto) and
ER fractions obtained by differential centrifugation were positive
for GAPDH as a cytosolic marker protein, or grp94 as an ER
marker, and negative for cytochrome C as a mitochondrial marker
(Figure 4, lanes 14). Mitochondria obtained by density
centrifugation were completely negative for GAPDH, only a
weak band corresponding to grp94 was detectable, and
cytochrome C was prominently detected, indicating efficient
purification of mitochondria (Figure 4, lanes 5 and 6).
SHC-SEAPDSP was detectable in an unglycosylated state in the
cytosolic fraction (Figure 4, lane 1) and in an N-glycosylated state
in the ER fraction (Figure 4, lane 2). In contrast, it was barely
detectable in the mitochondrial fraction (Figure 4, lane 5). A
different distribution was found for SHN-SEAPDSP, which was
present in the cytosolic fraction, but not in the ER fraction
(Figure 4, lanes 3 and 4). This observation is in line with the
Table 1. 32 Vertebrate signal peptides .40 amino acids, which are predicted to be NtraC organized and are similar in their
domain capacity to shrew-1.
NCBI Accession Number
Signal peptide sequence
Underlined residues are predicted turns belonging to the transition area.
absence of SEAP activity in the supernatant and whole cell lysates
extracted from cells transfected with this fusion protein (Figure 3).
Most importantly, SHN-SEAPDSP was prominently detected in the
mitochondrial fraction, which received further confirmation by
immunofluorescence studies in HEK 293T cells (not shown). This
experimental observation is in perfect agreement with the
Deletion of the transition area decreases secretion. The
results presented so far show that the C-domain is sufficient for
secretion of SEAP fusion protein, whereas the N-domain has no ER
translocation capacity, but rather accommodates a mitochondrial
targeting activity. However, when compared to the full length signal
sequence the C-domain exhibits a decreased secretion activity. This
observation gave rise to the question whether the transition area
(residues 1624) influences the efficiency of ER translocation.
To test this hypothesis, we generated constructs coding for three
different SEAP fusion proteins, containing mutations and deletions
in the transition area of the otherwise wild-type shrew-1 signal
peptide. One contains a GlyRIle substitution at position 18
(SHG18I-SEAPDSP) which was predicted to prevent the formation
of the first turn in the transition domain. In the second construct,
we deleted the first four amino acids with the highest turn forming
potential (SHDWPGR-SEAPDSP) of the predicted transition domain.
In the third construct, we deleted the first four amino acids of the
transition area and introduced additional substitutions in the
remaining four amino acids in order to completely disrupt the
transition area (SHDWPGR/mut-SEAPDSP) (for a schematic of all
constructs, see Figure 2 B).
Each of these constructs was transfected into HEK 293T cells,
and again SEAP activity was determined in the supernatants as
well as in whole cell lysates. As shown in Figure 5A, SEAP activity
decreases with increasing disruption of the transition area.
SHDWPGR/mut-SEAPDSP showed the lowest activity which is
similar to the activity of SHC-SEAPDSP. This is consistent with
the assumption that the transition area may be needed for the
overall secretion activity of the shrew-1 signal sequence.
The dependency of secretion efficiency on the integrity of the
transition area should be mirrored in the presence of
Nglycosylated SEAP. This was tested by Western blotting
(Figure 5B). With increasing impairment of the transition area
the ratio of N-glycosylated (upper band, c) to non-glycosylated
SEAP fusion protein (lower band, arrow) species decreased by one
order of magnitude from 1.94 to 0.17 (Figure 5B). We conclude
that protein export efficiency appears to be correlated with the
existence and integrity of the transition area separating N- and
Cdomains of the shrew-1 signal peptide.
Here we report the first systematic approach for predicting
structure and function of long signal peptides of single-pass integral
membrane proteins. Sequence analysis tools suggest a general
organization model for these sequences, which was validated in a
proof-of-principle study using the type I membrane protein
1. Most importantly, according to our NtraC model a structural
feature of the transition area is a crucial determinant of long signal
peptide modularization: A potentially turn- or loop-forming
central element (transition area) acts as some kind of separation
unit between two sequence domains with different targeting
capacity. Results of cellular targeting studies highlight the
functional importance of the transition area. A minimal
interpretation is that it affects ER translocation of the reporter protein.
The N-domain (residues 119) was able to act as a
mitochondrial targeting signal in our experiments. Similar
observations have been made for other proteins containing
consecutive tandem signals rather than cryptic signals as
described by the NtraC model. The transmembrane glycoprotein
nicastrin, which is an essential component of gamma-secretase
, is such an example. Gamma-secretase was found to
translocate into mitochondria in Alzheimer patients, potentially
inducing apoptosis . Transport into the organelle is mediated
by a mitochondrial transit signal following the N-terminal
cleavable signal peptide of nicastrin. Notably, in contrast to the
shrew-1 example and the NtracC domain model, the sequential
order of the targeting signals is inverted in nicastrin and other
proteins containing such a tandem signal, e.g. microsomal
CYP2E1 . This demonstrates that the prediction and
discovery of proteins with multiplex locations is important for an
understanding of the regulation of cell process such as apoptosis.
Mitochondrial targeting of shrew-1 and other proteins
containing NtraC-organized long signals may not occur constitutively but
in a regulated manner or only under cellular stress, and our results
indicate that the mitochondrial targeting signal (N-domain) and
the ER targeting signal (C-domain) are not sequentially processed.
The N-domain of shrew-1 harbors no ER translocation activity,
but is able to mediate mitochondrial targeting. We wish to stress
that this activity has been proven for the isolated N-domain in the
context of the experimental setup used in the present study, and it
needs further investigation to determine the conditions under
which this activity is found in the context of the full-length signal
peptide. Possibly this cryptic activity is revealed under certain
physiological situations only.
As an extension to the already known tandem signals like in the
nicastrin or CYP2E1 precursors [41,43], our NtraC model
provides a framework for cryptic signals. The domain model is
of general relevance, as at least 62% of the known vertebrata
proteins with a signal peptide exceeding 40 residues show an
NtraC-organization. Although it remains unclear if and under
which conditions or regulatory control mitochondrial targeting of
these proteins occurs, we were able show that NtraC-organized
signal peptides can exhibit additional functions besides ER
targeting or protein export. Prediction of such important structural
elements has now become feasible.
Due to its amphipathic nature, we further speculate that the
Ndomain might be involved in dimerization or stabilization of shrew-1
in the plasma membrane or interaction with other proteins [29,32].
Positively charged arginine residues in the N-domain could help the
signal peptide to adopt its native conformation in the plasma
membrane. It would thereby follow the positive inside rule 
and arrest the C-terminal part inside the membrane while being
available for protein-protein interactions on the cytoplasmic side.
The C-domain is sufficient for protein export via the ER, but not
as effective as the full-length signal peptide. Most strikingly, the
transition area which was first predicted to only link the N- to
Cdomain, turned out to be essential for the full ER translocation
activity of the C-domain. It is noteworthy that the transition area is
the only part of the long signal peptides predicted to predominantly
contain b-turns. Thus, turn formation seems to be not only a
structural element separating the N- and C-domains, but a decisive
feature of long signal peptides supporting the ER translocation
activity of the C-domain. The NtraC model thereby explains earlier
observations made for interleukin-15, which is subjected to different
export rates depending on the length of its signal peptide .
Our model also provides a rational explanation for membrane
targeting of bacterial autotransporters, which possess long signal
peptides: These are in accordance with our NtraC model, where
the C-domain alone is sufficient for transport to the inner
membrane but for proper processing the complete signal peptide is
required . In the present study, we restricted our analysis to
single-spanning integral membrane proteins with signals that have
a similar organization as the long signal peptide of shrew-1. The
role of the transition area besides making the N- and C-domain
distinguishable is subject to further research.
Materials and Methods
Oligonucleotides used for cloning of SEAP fusion
Constructs were generated by PCR (Suppl. Text S1).
Cell lines, cell culture and transfection
HEK 293T (CRL-11268; ATCC, Manassas, USA) were
cultured in Dulbeccos Modified Eagle Medium (DMEM;
(Invitrogen GmbH, Karlsruhe, Germany) with 10% fetal calf
serum (FCS; PAA LABORATORIES, Co lbe, Germany) and 1%
penicillin/streptomycin (Invitrogen GmbH, Karlsruhe, Germany).
66105 cells were seeded per 12 cm2 of culture dish and
transfected with 3 mg DNA 24 h later by using Magnet Assisted
Transfection (MATra, IBA GmbH, Go ttingen, Germany)
according to the manufacturers instructions.
Immunoblotting and antibodies
After collection of supernatant for SEAP assays, cells were
washed with PBS and lysed with 100 ml RIPA buffer (150 mM
NaCl, 50 mM Tris-HCl, pH 7.5, 0.5% sodium deoxycholate, 1%
Nonidet P-40, 0.1% SDS) containing proteinase inhibitor cocktail
Complete (Roche Diagnostics GmbH, Mannheim, Germany) at
4uC for 30 min. Lysates were cleared by centrifugation in a
microcentrifuge at 4uC for 5 min. Where indicated, cell lysates
were treated with PNGase F which removes N-glycans according
to the manufacturers instructions (New England Biolabs,
Frankfurt, Germany). For immunoblotting, 20 mg of protein from
each cell lysate was separated in a 6% SDS PAA-gel. Protein blots
were incubated with rabbit polyclonal anti-myc antibody (0.5 mg/
ml; Sigma-Aldrich Chemie GmbH, Mu nchen, Germany) diluted
in TBST (10 mmol/L Tris-HCl, pH 7.4, 150 mmol/L NaCl;
0.05% Tween 20). Glyceraldehyde-3-phosphate dehydrogenase
(GAPDH) was probed with mouse monoclonal anti-GAPDH
antibody (1 mg/ml, Ambion/Applied Biosystems, Darmstadt,
Germany), cytochrome c with mouse monoclonal anti-cytochrome
c antibody (0.4 mg/ml; medac, Wedel, Germany) and Grp94 with
rat monoclonal anti-Grp94 antibody (2 mg/ml; medac, Wedel,
Germany). Secondary alkaline phosphatase-conjugated goat
antirabbit antibody, horseradish peroxidase-conjugated goat
antirabbit, horseradish peroxidase conjugated goat anti-mouse
antibody and horseradish peroxidase conjugated goat anti-rat
antibody (all Jackson ImmunoResearch, Dianova GmbH,
Hamburg, Germany) were used for detection of first antibodies.
Enzyme substrates were NBT/BCIP (Roche Diagnostics GmbH,
Mannheim, Germany) for alkaline phosphatase or a solution of
luminol (2.5 mM), p-coumaric acid (0.4 mM), Tris-HCl, pH 8.5
(100 mM) and 0.009% H2O2 for horseradish peroxidase.
The densitometric analysis of the Western blots was performed
with Image J (Scion). The densities of the corresponding bands on
the blot were measured and the ratio of the upper band to the
lower band of each construct was calculated.
Isolation of mitochondria
24 hours after transfection of HEK 293T cells mitochondria
were isolated with the Qproteome Mitochondria Isolation Kit
(Qiagen, Hilden, Germany) according to the manufacturers
instructions. Briefly, after removal of nuclei, cell debris, cytosolic
and microsomal cell fractions, the mitochondria pellet was
resuspended in 0.5 M sucrose buffer (1 mM EDTA, 0.1% BSA,
10 mM Tris-HCl, pH 7.5), layered on a 12 M sucrose gradient
(1 mM EDTA, 0.1% BSA, 10 mM Tris-HCl, pH 7.5) and
centrifuged for 2 h at 25,000 rpm. The mitochondrial band was
collected, diluted with 2 volumes of 1 mM EDTA, 10 mM
TrisHCl, pH 7.4 buffer and pelleted by centrifugation at 20,0006g for
15 min. 20 mg of protein of each fraction was loaded on a 10%
PAA-gel and separated by SDS-PAGE.
Table S1 Vertebrate signal peptides .40 amino acids, which
are predicted to be NtraC organized but differ in their domain
capacity from shrew-1. Underlined residues are predicted turns
belonging to the T-domain
Found at: doi:10.1371/journal.pone.0002767.s001 (0.17 MB
Oligonucleotides used for cloning of SEAP fusion
We thank Matthias Schmidt and Monika Kamprad for technical support,
and Bernhard Dobberstein, Katja Kapp, and Paul Wrede for fruitful
discussion. Norbert Dichter helped us set up the web interface.
Conceived and designed the experiments: JAH ER AS ASP. Performed the
experiments: JAH ER. Analyzed the data: JAH ER AS ASP GS.
Contributed reagents/materials/analysis tools: MM. Wrote the paper:
JAH GS. Performed the biological experiments: ER. Designed and
supervised the biological experiments and analyzed the biological data:
ASP AS. Analyzed the bioinformatical data and developed the NtraC model:
1. Blobel G ( 2000 ) Protein targeting . Biosci Rep 20 : 303 - 344 .
2. Nickel W ( 2005 ) Unconventional secretory routes: direct protein export across the plasma membrane of mammalian cells . Traffic 6 : 607 - 614 .
3. Ouzzine M , Magdalou J , Burchell B , Fournel-Gigleux S ( 1999 ) An internal signal sequence mediates the targeting and retention of the human UDPglucuronosyltransferase 1A6 to the endoplasmic reticulum . J Biol Chem 274 : 31401 - 31409 .
4. Champion PA , Stanley SA , Champion MM , Brown EJ , Cox JS ( 2006 ) Cterminal signal sequence promotes virulence factor secretion in Mycobacterium tuberculosis . Science 313 : 1632 - 1636 .
5. Gray L , et al. ( 1989 ) A novel C-terminal signal sequence targets Escherichia coli haemolysin directly to the medium . J Cell Sci Suppl 11 : 45 - 57 .
6. von Heijne G, ed ( 1994 ) Signal Peptidases . Austin: R.G. Landes .
7. Watson ME ( 1984 ) Compilation of published signal sequences . Nucl Acids Res 12 : 5145 - 5164 .
8. Izard JW , Kendall DA ( 1994 ) Signal peptides: exquisitely designed transport promoters . Mol Microbiol 13 : 765 - 773 .
9. Jungnickel B , Rapoport TA ( 1996 ) A posttargeting signal sequence recognition event in the endoplasmic reticulum membrane . Cell 82 : 261 - 270 .
10. Martoglio B ( 2003 ) Intramembrane proteolysis and post-targeting functions of signal peptides . Biochem Soc Trans 31 : 1243 - 1247 .
11. von Heijne G ( 1990 ) The signal peptide . J Membr Biol 115 : 195 - 201 .
12. Feng ZP ( 2001 ) Prediction of the subcellular location of prokaryotic proteins based on a new representation of the amino acid composition . Biopolymers 58 : 491 - 499 .
13. Schneider G , Fechner U ( 2004 ) Advances in the prediction of protein targeting signals . Proteomics 4 : 1571 - 1580 .
14. Emanuelsson O , Brunak S , von Heijne G , Nielsen H ( 2007 ) Locating proteins in the cell using TargetP, SignalP and related tools . Nat Protoc 2 : 953 - 971 .
15. Shen H-B , Chou K-C ( 2007 ) Signal-3L: A 3-layer approach for predicting signal peptides . Biochem Biophys Res Comm 363 : 297 - 303 .
16. Chou K-C , Shen H-B ( 2007 ) Signal-CF: A subsite-coupled and window-fusing approach for predicting signal peptides . Biochem Biophys Res Comm 357 : 633 - 640 .
17. Horton P , et al. ( 2007 ) WoLF PSORT: protein localization predictor . Nucl Acids Res 35 : W585 - W587 .
18. Tamura T , Akutsu T ( 2007 ) Subcellular location prediction of proteins using support vector machines with alignment of block sequences utilizing amino acid composition . BMC Bioinformatics 8 : 466 .
19. Wiedmann M , Huth A , Rapoport TA ( 1984 ) Xenopus oocytes can secrete bacterial beta-lactamase . Nature 309 : 637 - 639 .
20. Al-Qahtani A , Teilhet M , Mensa-Wilmot K ( 1998 ) Species-specificity in endoplasmic reticulum signal peptide utilization revealed by proteins from Trypanosoma brucei and Leishmania . Biochem J 331 : 521 - 529 .
21. Hegde RS , Bernstein HD ( 2006 ) The surprising complexity of signal sequences . Trends Biochem Sci 31 : 563 - 571 .
22. Chou K-C , Shen H-B ( 2008 ) Cell-PLoc: A package of web-servers for predicting subcellular localization of proteins in various organisms . Nature Protocols 3 : 153 - 162 .
23. Bendtsen JD , Nielsen H , von Heijne G , Brunak S ( 2004 ) Improved prediction of signal peptides: SignalP 3.0 . J Mol Biol 340 : 783 - 795 .
24. Froeschke M , Basler M , Groettrup M , Dobberstein B ( 2003 ) Long-lived signal peptide of lymphocytic choriomeningitis virus glycoprotein pGP-C . J Biol Chem 278 : 41914 - 41920 .
25. Ramanujan S , Bernstein HD ( 2006 ) The surprising complexity of signal sequences . Biochem Sci 31 : 563 - 571 .
26. Martoglio B , Dobberstein B ( 1998 ) Signal sequences: more than just greasy peptides . Trends Cell Biol 8 : 410 - 415 .
27. Kurys G , Tagaya Y , Bamford R , Hanover JA , Waldmann TA ( 2000 ) The long signal peptide isoform and its alternative processing direct the intracellular trafficking of interleukin-15 . J Biol Chem 275 : 30653 - 30659 .
28. Dultz E , Hildenbeutel M , Martoglio B , Hochman J , Dobberstein B , et al. ( 2008 ) The signal peptide of the mouse mammary tumor virus Rem protein is released from the endoplasmic reticulum membrane and accumulates in nucleoli . J Biol Chem 283 : 9966 - 9976 .
29. Bharti S , Handrow-Metzmacher H , Zickenheiner S , Zeitvogel A , Baumann R , et al. ( 2004 ) Novel membrane protein shrew-1 targets to cadherin-mediated junctions in polarized epithelial cells . Mol Biol Cell 15 : 397 - 406 .
30. Resch E , Quaiser S , Quaiser T , Schneider G , Starzinski-Powitz A , et al. ( 2008 ) Synergism of shrew-1's signal peptide and transmembrane segment required for plasma membrane localization . Traffic , in press.
31. Jakob V , Schreiner A , Tikkanen R , Starzinski-Powitz A ( 2006 ) Targeting of transmembrane protein shrew-1 to adherens junctions is controlled by cytoplasmic sorting motifs . Mol Biol Cell 17 : 3397 - 3408 .
32. Schreiner A , Ruonala M , Jakob V , Suthaus J , Boles E , et al. ( 2007 ) Junction protein shrew-1 influences cell invasion and interacts with invasion-promoting protein CD147 . Mol Biol Cell 18 : 1272 - 1281 .
33. Meissner M , Koch O , Klebe G , Schneider G ( 2008 ) Prediction of turns types in protein structure by machine-learning classifiers . Proteins , in press.
34. Wu CH , Apweiler R , Bairoch A , Natale DA , Barker WC , et al. ( 2006 ) The Universal Protein Resource (UniProt): an expanding universe of protein information . Nucl Acids Res 34 : D187 - 191 .
35. Emanuelsson O , Nielsen H , Brunak S , von Heijne G ( 2000 ) Predicting subcellular localization of proteins based on their N-terminal amino acid sequence . J Mol Biol 300 : 1005 - 1016 .
36. Hutchinson EG , Thornton EG ( 1994 ) A revised set of potentials for beta-turn formation in proteins . Protein Sci . 3 : 2207 - 2213 .
37. Kall L , Krogh A , Sonnhammer E ( 2005 ) An HMM posterior decoder for sequence feature prediction that includes homology information . Bioinformatics 21 : 251 - 257 .
38. Yuan Z , Mattick JS , Teasdale RD ( 2004 ) SVMtm: Support vector machines to predict transmembrane segments . J Comput Chem 25 : 632 - 636 .
39. Berger J , Hauber J , Hauber R , Geiger R , Cullen BR ( 1988 ) Secreted placental alkaline phosphatase: a powerful new quantitative indicator of gene expression in eukaryotic cells . Gene 66 : 1 - 10 .
40. Foulquier F , Harduin-Lepers A , Duvet S , Marchal I , Mir AM , et al. ( 2002 ) The unfolded protein response in a dolichyl phosphate mannose deficient Chinese hamster ovary cell line points out the key role of a demannosylation step in the quality-control mechanism of N-glycoproteins . Biochem J 362 : 491 - 498 .
41. Takasugi N , Tomita T , Hayashi I , Tsuruoka M , Niimura M , et al. ( 2003 ) The role of presenilin cofactors in the gamma-secretase complex . Nature 422 : 438 - 441 .
42. Hansson CA , Frykman S , Farmery MR , Tjernberg LO , Nilsberth C , et al. ( 2004 ) Nicastrin, presenilin, APH-1, and PEN-2 form active gamma-secretase complexes in mitochondria . J Biol Chem 279 : 51654 - 51660 .
43. Robin MA , Anandatheerthavarada HK , Biswas G , Sepuri NB , Gordon DM , et al. ( 2002 ) Bimodal targeting of microsomal CYP2E1 to mitochondria through activation of an N-terminal chimeric signal by cAMP-mediated phosphorylation . J Biol Chem 277 : 40583 - 40593 .
44. von Heijne G ( 1986 ) The distribution of positively charged residues in bacterial inner membrane proteins correlates with the trans-membrane topology . EMBO J 5 : 3021 - 3027 .
45. Szabady RL , Peterson HP , Skillman KM , Bernstein HD ( 2005 ) An unusual signal peptide facilitates late steps in the biogenesis of a bacterial autotransporter . Proc Natl Acad Sci USA 102 : 221 - 226 .