You are on page 1of 16

Appl Microbiol Biotechnol (2013) 97:3811–3826

DOI 10.1007/s00253-013-4831-z

MINI-REVIEW

Optimisation of signal peptide for recombinant protein


secretion in bacterial hosts
Kheng Oon Low & Nor Muhammad Mahadi &
Rosli Md. Illias

Received: 5 January 2013 / Revised: 3 March 2013 / Accepted: 4 March 2013 / Published online: 26 March 2013
# Springer-Verlag Berlin Heidelberg 2013

Abstract Escherichia coli—the powerhouse for recombi- recombinant proteins are low yields and inefficient purifi-
nant protein production—is rapidly gaining status as a reli- cation processes. The ability of many organisms to transport
able and efficient host for secretory expression. An native proteins away from the cytoplasm has sparked an
improved understanding of protein translocation processes interest in developing a secretory production process for
and its mechanisms has inspired and accelerated the devel- recombinant proteins. Secretory expression offers many ad-
opment of new tools and applications in this field and, in vantages when compared with cytoplasmic expression: it
particular, a more efficient secretion signal. Several impor- simplifies the detection and purification of the product, re-
tant characteristics and requirements are summarised for the duces the complexity of the bioprocess, minimises the cell-
design of a more efficient signal peptide for the production associated proteolytic degradation and improves the protein
of recombinant proteins in E. coli. General approaches and folding and quality (Choi and Lee 2004).
strategies to optimise the signal peptide, including the se- As one of the most well-known microbes, Escherichia coli
lection and modification of the signal peptide components, is the most commonly used host for the production of recom-
are included. Several challenges in the secretory production binant proteins. Although not suitable for proteins that are rich
of recombinant proteins are discussed, and research ap- in cysteine residues or that require extensive post-translational
proaches designed to meet these challenges are proposed. modifications, these bacteria offer the easiest, most econom-
ical, and fastest expression of heterologous proteins (Terpe
Keywords Signal peptide . Recombinant protein . 2006; Demain and Vaishnav 2009; Chen 2012). The features
Mutation . Secretory production . Sec system . of E. coli as a secretory expression host have been reviewed
Escherichia coli extensively (Choi and Lee 2004; Mergulhao et al. 2005; Wong
et al. 2012), as have strategies to achieve high levels of
recombinant protein production (Yoon et al. 2010; Shokri et
Introduction al. 2003; Choi et al. 2006). This robust host has multiple
complex secretion pathways that are capable of exporting a
Recombinant protein production processes in microbial wide range of recombinant proteins when fused to specific
hosts are a major factor in modern biotechnology and bio- secretion signals. However, its secretion capacity is often
based economies. The main obstacles in the production of limited to a frustratingly low level compared to other expres-
sion hosts (i.e. Bacillus subtilis). Recent findings suggest that
K. O. Low : R. Md. Illias (*) the full potential of this microbe has yet to be fully explored
Department of Bioprocess Engineering, Faculty of Chemical
and realised. Here, we discuss several of the latest develop-
Engineering, Universiti Teknologi Malaysia, 81310, Skudai, Johor
Bahru, Malaysia ments of this field, particularly as related to E. coli.
e-mail: r-rosli@utm.my The hallmark of gram-negative bacteria is that the cell
K. O. Low envelope is composed of two membranes (the inner and
e-mail: khengoon@hotmail.com outer membranes), which are separated by a compartment
(the periplasm) that contains a thin peptidoglycan (or mu-
N. Muhammad Mahadi
rein) layer. In E. coli, the export of protein is generally
Malaysia Genome Institute (MGI), Ministry of Science,
Technology and Innovation, Jalan Bangi Lama, 43000, Kajang, accomplished by the Sec secretion system, also known as
Selangor, Malaysia the general secretory pathway, through one or both of the
3812 Appl Microbiol Biotechnol (2013) 97:3811–3826

membranes. A target protein, fused to an N-terminal secre- a translocation complex with the Sec translocase—
tion signal known as a signal peptide, can be recognised and SecY/SecE/SecG (Georgiou and Segatori 2005). SecE and
translocated by the Sec machinery to the periplasm. The SecY form a protein-conducting channel, with SecG stimulat-
process of passing the protein through the inner membrane ing the translocation activity. Proteins are translocated in an
and into the periplasm is called secretion or secretory pro- unfolded state using energy from ATP hydrolysis and a proton
duction. If the target protein further crosses the outer mem- gradient (Papanikou et al. 2007). The translocation process is
brane to the extracellular medium, the process is termed as completed after the signal peptidase cleaves the signal peptide
excretion or excretory production. On the one hand, gram- (Wickner et al. 1987) and SecD releases the mature protein
positive bacteria, such as Bacillus, contain only a single cell from the cytoplasmic membrane (Matsuyama et al. 1993).
membrane with a thick peptidoglycan layer. In this case, The protein intended for secretion is routed to the
exported protein traverses only one membrane to reach the SecA/SecYEG complex in a SecB-dependent or signal recog-
extracellular medium and the process is called secretion. nition particle (SRP)-dependent manner. SecB, or in some
Despite differences in their final destinations, the secretion cases DnaK (Mujacic et al. 2004), targets less hydrophobic
and excretion processes rely on the same Sec machinery to signal peptides—post-translationally—to the Sec translocase
traverse the inner membrane. The signal peptide plays a (Papanikou et al. 2007). SRP targets highly hydrophobic signal
direct and decisive role in directing the target protein to peptides and membrane proteins—co-translationally—to the
the Sec machinery and across inner membrane. To achieve Sec complex. The SRP pathway includes the E. coli SRP (ffh),
a high level of protein secretion, the choice of signal peptide a cytosolic ribonucleoprotein composed of 4.5S RNA and a
remains the first and most important hurdle to be overcome, single polypeptide, and FtsY (Luirink and Dobberstein 1994).
but a comprehensive paper that reviews the state-of-the-art SRP binds to the newly synthesised signal peptide, which then
in signal peptide optimisation is not available. This article targets the ribosome to the translocase via the FtsY receptor.
aims to complement the existing reviews by focusing on the The secretion machinery differentiates secreted from
optimisation of the signal peptide, particularly for the selec- non-secreted proteins through recognition of the signal pep-
tion and modification of the signal peptide for high levels of tide (Gouridis et al. 2009). Typically 15–30 amino acids
recombinant protein secretion in E. coli. Relevant studies long, these amino-terminal pre-sequences are characterised
that used signal peptide in other bacterial hosts, such as by their three distinct regions, a 5- to 8-residue-long posi-
Bacillus, are also included. Several aspects that were cov- tively charged N terminus (n-region), an 8- to 12-residue-
ered in other recent reviews are not discussed, including the long hydrophobic core (h-region), and a polar 5- to 7-
role of codon bias (Zalucki et al. 2009) and translation residue-long carboxyl-terminal cleavage region (c-region)
(Zalucki et al. 2011) of the signal sequence in protein (Fig. 2). Although all signal peptides contain these regions,
export. Strategies to release recombinant proteins from the variability at the amino acid level is high. This variation
periplasm to the extracellular medium include the use of cell likely determines the capacity of the signal peptide to drive
envelope mutants, co-expression of lysis promoting proteins protein secretion. By understanding the role of the different
and manipulation of the culture conditions and medium (Li amino acids at specific positions, signal peptides with an
et al. 2010; Sommer et al. 2010; Ni and Chen 2009). The improved ability to confer high levels of protein secretion
literature survey was performed using the PubMed database can be designed. Biochemical data (Mori et al. 1997) and
and references therein. high-resolution nuclear magnetic resonance images of the
signal peptide-bound SecA (Gelis et al. 2007; Chou and
Gierasch 2005) have suggested several important consider-
Signal peptide and protein translocation across ations and characteristics to be used in designing a function-
the cell membrane al signal peptide. For example, the signal peptide h-region
adopts an extended α-helical conformation that binds to the
In E. coli, as much as 50 % of all cellular proteins are SecA groove. In addition, the positively charged residues in
exported, with more than 90 % translocated via the Sec the n-region form salt bridges with the acidic residues adja-
pathway. After emerging from the ribosome, the process of cent to the SecA groove, while the c-region remains largely
secreting a protein involves a cascade of interactions with exposed and without detectable interactions.
the Sec machinery before reaching the final destination Another protein secretion system that utilises N-terminal
(Fig. 1) (reviewed in Rusch and Kendall 2007). SecB binds signal peptides is the Sec-independent twin-arginine trans-
to the pre-protein, maintaining a transport-competent state port (Tat) system (Palmer and Berks 2012) (Fig. 1). In E.
(Hardy and Randall 1991). SecA functions as a translocation coli, the Tat translocon consists of three different membrane
ATPase (Economou and Wickner 1994). Upon binding with proteins, TatA, TatB and TatC (Muller 2005). TatBC com-
the pre-protein and adenosine triphosphate (ATP), SecA plex functions in the recognition of targeted proteins, which
inserts into the membrane (Ulbrandt et al. 1992), forming triggers the recruitment of the TatA complex to form a
Appl Microbiol Biotechnol (2013) 97:3811–3826 3813

Fig. 1 Schematic diagram of the major signal peptide-based secretion and guides the pre-protein through the SecYEG translocon, at the
system and process. A and B represent the Sec system, and C represents expense of ATP. SecDF aids in the release of the pre-protein into the
the Tat system. A Membrane proteins and highly hydrophobic signal periplasm. SPase I cleaves the signal peptide, releasing the mature
peptides follow the co-translational translocation, or SRP, route. SRP domain to the periplasm. Periplasmic chaperones may aid in protein
binds to an emerging nascent polypeptide chain to form an SRP– folding. C Rapidly folded proteins that harbour the Tat signal peptide
nascent chain–ribosome complex. A receptor, FtsY, then binds to the follow the Tat route. After being synthesised, the pre-protein folds
complex and guides it to the SecYEG translocon in the presence of rapidly into its native conformation, sometimes with the help of cyto-
GTP. Once it has arrived at the SecYEG, the pre-protein enters the plasmic chaperones and cofactors. Tat protein is not recognised by the
membrane through the SecYEG gate/‘tunnel’ as it is being synthesised. Sec components, possibly because of the Tat signal sequence and/or
SecA provides mechanical work by hydrolysing ATP. B Most periplas- because of the rapidly folded nature of the pre-protein. TatBC recog-
mic proteins and less hydrophobic signal peptides follow the post- nises the Tat protein and assembles the TatA translocon, through which
translational translocation, or SecB, route. SecB binds to a largely or the Tat protein is traversed into periplasm. Similar to the Sec system,
fully synthesised secreted protein and maintains it in an unfolded state. SPase I (not shown) cleaves the Tat signal peptide and releases the
Cytoplasmic or SecYEG-bound SecA searches for the signal peptide mature domain into the periplasm

functional translocon (70 Å or more in diameter) within the Passenger protein: secreted or not secreted?
cell membrane (Muller 2005). In addition to the three-
distinct-domain architecture of the Sec signal peptide, Tat One of the pressing concerns in secretory recombinant pro-
signal peptides contain a few unique characteristics (Berks tein production is understanding the characteristics of the
et al. 2000): (1) an (S/T)-R-R-x-F-L-K sequence motif at the target protein that allow the secretion process to occur. This
n/h-region boundary, from which the twin arginine pathway remains a black-box issue and involves trial-and-error rather
gets its name; (2) a relatively longer length and an extended than rational evaluation. It is generally believed that natively
n-region; and (3) a less hydrophobic h-region, with a higher secreted proteins are good candidates for secretory produc-
occurrence of glycine and threonine residues than leucine. tion in heterologous hosts. However, studies have shown
Unlike secretion by the Sec pathway, which requires that the that natively cytoplasmic proteins can also be secreted in
protein remain in an unfolded state for export, the Tat high levels when fused to a signal peptide. Some examples
pathway recognises folded protein for export to the peri- include the secretion of winter flounder antifreeze protein
plasm. In general, proteins that have cofactors or fold rap- (Tong et al. 2000) and Manduca diuresin (Wong et al. 2003)
idly in the cytoplasm can be transported along the Tat route. using the OmpA signal peptide in E. coli. For the Sec
3814 Appl Microbiol Biotechnol (2013) 97:3811–3826

the amount of aircraft that can flow through the system.


Choosing a suitable signal peptide constitutes the first im-
portant step in constructing an extracellular recombinant
protein secretion system. For E. coli, the type of signal
peptide may be chosen based on the commercial availability
(i.e. PelB in a pET vector), a literature survey (i.e. MBP,
OmpA) or personal preferences. Signal peptides that have
previously performed well in E. coli may serve as good
candidates; several are listed in Table 1. With the limited
Fig. 2 General features of a Sec system signal peptide. a A typical
signal peptide contains three functional domains: a positively charged choices available in the literature, more efficient signal
n-region, a hydrophobic h-region and a polar c-region. A pro-region, peptides may need to be developed. Systematic approaches
also known as the export initiation domain, is also found in some in searching for the ideal natural signal peptide have been
secreted proteins. Signal peptidase I recognises and cleaves at the c-
developed in several laboratories. With ambitious genome
region AxA motif (indicated by downward-pointing arrow). b A
summary of a more efficient signal peptide design. A net positive projects and sophisticated in silico prediction programs, the
charge in the n-region and an AAA codon at the second codon (P2). identification of all signal sequences in a host cell can be
The AAA codon permits strong translation initiation and codes for a made and evaluated to enhance recombinant protein secre-
positively charged amino acid, lysine. A highly hydrophobic h-region
tion. Currently, the National Center for Biotechnology
arising from a stretch of leucine amino acids and the presence of a helix
breaker residue, glycine, in the middle. An extended motif, xxAxA, for Information (NCBI) hosts more than 1,000 genome se-
the c-region, where x represents average-to-large amino acids, from quences for a range of organisms, including several popular
position −1 to −5. A net negative charge in the pro-region, with small bacterial expression hosts such as E. coli, Bacillus, and
and negatively charged amino acids in the first two positions
Lactobacillus (ftp://ftp.ncbi.nih.gov/genomes/Bacteria).
Table 2 lists several of the commonly used computational
pathway, one of the most important characteristics of the tools for signal sequence prediction. SignalP (Petersen et al.
secreted protein is that it has a slow folding rate or remains 2011) and Phobius (Kall et al. 2007) rank among the most
in an unfolded state for a sufficient time to allow recognition accurate and versatile tools that permit high-throughput
and translocation by the secretion pathway. If the target processing of protein sequences in a relatively short period
protein folds rapidly or tightly in the cytoplasm, one may of time. A signal peptide library can be constructed by
opt to use a Tat signal peptide or engineer the target protein cloning the signal-encoding nucleotides into an expression
to have slower folding kinetics. The latter approach was vector, and these can be screened for protein production.
shown to be feasible by Bakkes et al. (2010) for protein This approach has been proven feasible, as demonstrated by
exported by the type I haemolysin transport system in E. several studies. For instance, Brockmeier et al. (2006) and
coli. Similar to the Sec system, the haemolysin transport Degering et al. (2010) constructed libraries consisting of all
system recognises and translocates unfolded polypeptides. naturally occurring (non-lipoprotein) Sec-type signal se-
The authors created a ‘folding’ mutation, Y283D, of malt- quences from B. subtilis (173 signal sequences) and
ose binding protein (MBP) and showed that the mutant was Bacillus licheniformis (220 signal sequences) and screened
translocated more efficient than its wild-type counterpart. them for heterologous protein secretion (cutinase from
The approach may also be practical for the Sec pathway. On Fusarium solani pisi and subtilisin from Bacillus
the other hand, recent evidence has shown that proteins that amyloliquefaciens). These studies revealed that signal pep-
have undergone post-translational modification were equal- tides can have large variations in production levels,
ly well recognised and translocated through the Sec pathway depending on the type of recombinant protein fused to them.
(Chen et al. 2012). The optimal signal peptide for one protein was inefficient
for another protein and vice versa. A similar outcome was
obtained for a library of 76 Sec-type signal peptides from
Optimisation of the signal peptide Lactobacillus plantarum, which was screened for the effi-
ciency of protein secretion using staphylococcal nuclease
Signal peptide selection (NucA) and lactobacillal amylase (AmyA) as the reporter
proteins (Mathiesen et al. 2009). No correlation with the n-
The signal peptide functions like a global positioning sys- region net charge, hydrophobicity level or length could be
tem, providing the information needed to direct the target identified for the high secretion performance signal pep-
protein to the final destination. After its synthesis in the tides. In another example, 108 signal peptides derived from
cytoplasm, a polypeptide competes with other transport- a 408-signal-peptide library from Corynebacterium
competent polypeptides for export. As an analogy, one can glutamicum R were evaluated for secretion efficiency using
imagine a busy airport with only one take-off lane, limiting active α-amylase from Geobacillus stearothermophilus
Appl Microbiol Biotechnol (2013) 97:3811–3826 3815

Table 1 Examples of several signal peptides that perform well in E. coli

Signal sequence Origin Recombinant protein Origin Yield Reference

Heat-labile E. coli Cholera toxin (ctxB) Vibrio cholerae 86–259 μg/ml (total of periplasm Jobling et al. (1997)
enterotoxin and supernatant)
LTIIb Alkaline phosphatase E. coli 0.1 U (1.7-fold greater than native
(phoA) phoA signal sequence under similar
expression conditions)
Pneumococcal surface Streptococcus Most abundant protein in periplasm
protein A (pspA) pneumoniae and supernatant
Alkaline E. coli Glucagon Human Up to 3.46 mg/L/OD600 at supernatant Wen et al. (2003)
phosphatase fraction
phoA Human epidermal Human 380 mg/L Zhang et al. (2007)
growth factor (hEGF)
Pectate lyase Erwinia Phospholipase A2 Streptomyces Up to 6 U/ml in the extracellular Takemori et al.
B PelB carotovora (PLA2) violaceoruber fraction; undetected in intracellular (2012)
fraction
Lipase Bacillus 35 % (U/g wet weight cell) in the Cho et al. (2000)
thermoleovorans extracellular fraction.
ompA E. coli Calcitonin peptide Salmon More than 200 mg/L in extracellular Ray et al. (2002)
(sCTgly) fraction when coexpressed with secE
and secY
GM-CSF Human Up to 0.7 g/l in soluble fraction under Sletta et al. (2007)
high cell density cultivation
Bordetella Avidin Chicken Up to 20 mg/L after purification from Hytonen et al.
avium periplasm (2004)
lamB E. coli Insulin-like growth Human Up to 8.5 g/L in the periplasm when Joly et al. (1998)
factor-I (IGF-I) coexpressed with dsbA/dsbC
Endoxylanase Bacillus sp. Alkaline phosphatase E. coli 5.2 g/L in periplasm under high cell Choi et al. (2000)
(AP) density cultivation

(Watanabe et al. 2009). Among those peptides, 11 signal assumes that abundantly secreted proteins should have an
peptides conferred a 50- to 150-fold higher secretion level above-average signal peptide efficiency, making these sig-
than the well-known corynebacterial secretory protein PS2. nal peptides ideal for high levels of recombinant protein
As the authors did not test the best signal peptides using secretion. Two-dimensional gel electrophoresis followed
another heterologous protein, this improvement in perfor- by mass spectrometry is used to determine the identity of
mance may not be applicable for other proteins. secreted proteins and then its associated signal peptide. In
Another approach utilises the proteomic profiles, partic- addition to being a tool to verify genome-based signal
ularly the secretome, of a host cell to identify better- peptide predictions, this approach avoids the tedious and
performing signal peptides. The rationale of this strategy laborious tasks of constructing a signal peptide library and

Table 2 Several in silico tools for signal sequence prediction

Prediction tools Prediction methods URL References

SignalP Hidden Markov model and neural network cbs.dtu.dk/services/SignalP/ Petersen et al. (2011)
PrediSi Position weight matrix www.predisi.de/index.html Hiller et al. (2004)
Signal-BLAST Sequence alignment (BLASTP) sigpep.services.came.sbg.ac.at/signalblast.html Frank and Sippl (2008)
Signal-3L Multi-program modules www.csbio.sjtu.edu.cn/bioinf/Signal-3L/ Shen and Chou (2007)
Signal-CF Multi-program modules www.csbio.sjtu.edu.cn/bioinf/Signal-CF/ Chou and Shen (2007)
Phobius Hidden Markov model phobius.sbc.su.se/index.html Kall et al. (2007)
SIG-Pred Matrix bmbpcu36.leeds.ac.uk/prot_analysis/Signal.html –
SPEPLip Neural Network gpcr.biocomp.unibo.it/cgi/predictors/spep/pred_ Fariselli et al. (2003)
spepcgi.cgi
OCTOPUS BLAST homology and neural network octopus.cbr.su.se/index.php Viklund and Elofsson
and hidden Markov model (2008)
3816 Appl Microbiol Biotechnol (2013) 97:3811–3826

evaluating secretion efficiencies. For diderm organisms, MS), which restricts the use of these approaches in labora-
such as E. coli, one could perform proteomic verification tories with limited resources. An additional approach that
using the periplasmic or extracellular fraction. While both may ameliorate these difficulties involves the ‘rational
fractions serve well to identify native signal peptides, pro- screening’ of available signal sequences from the literature
teomic analysis of the extracellular fraction may benefit and from protein databases. The availability of public data-
those who favour excretory production of recombinant pro- bases of signal sequences has increased the interest in this
teins. In general, the N-terminal signal sequence together method as an easy and less resource-heavy solution. Table 3
with its mature protein is needed for excretion of recombi- lists several signal sequence databases that are accessible
nant protein, but the mechanism for passing through the from the Internet. Among those, CoBaltDB (Goudenege et
outer membrane is largely unknown. With this approach, al. 2010) represents the most comprehensive and systematic
Qian et al. (2008) isolated 22 proteins that were abundant in database for predicting protein localisation in bacterial and
the fermentation medium of E. coli BL21(DE3). Among those archaeal microbes, integrating 43 in silico localisation tools
potential candidates, OsmY (a periplasmic, osmotically induc- for the proteome of 784 organisms and displaying relevant
ible protein Y) was found to be an efficient fusion carrier to information from NCBI and Kyoto Encyclopedia of Genes
direct excretion of recombinant proteins (E. coli alkaline and Genomes (KEGG) for a given protein. An easy and
phosphatase, human leptin and B. subtilis α-amylase), with rapid analysis of the proteome of an organism can be made
concentrations ranging from 250 to 700 mg/L under high cell with a high degree of confidence. Signal sequences that
density cultivation conditions. The fusion of recombinant pro- consistently score as a potential signal peptide across a
teins to different segments of OsmY yielded localisation to range of prediction tools can be selected for further analyses,
different cell compartments, namely, the cytoplasm for the effectively narrowing the search space and increasing the
protein fused to the mature osmY, the periplasm for the protein possibility of detecting an efficient signal peptide. As each
fused to the signal sequence of osmY and the extracellular database has its own strengths and weaknesses, the choice
medium for the protein fused to the full-length osmY (signal will depend on the purpose of study.
sequence and mature osmY). The usefulness of OsmY was
also shown in recent work by Kotzsch et al. (2011). The Signal peptide modification
excretory production of 24 human target proteins was higher
when the proteins were fused to OsmY instead of with OmpA Protein secretion largely depends on the efficiency of the
or OmpF. A protease cleavage site (TEV site) and 6xHis tag signal peptide in driving protein translocation. Minor mod-
was located between the fusion carrier and the heterologous ifications to the natural signal peptide can often significantly
protein to facilitate the removal of fusion carrier and rapid improve the secretory expression over that of the parent
purification, respectively. In this case, highly purified protein peptide. In this section, we discuss several important con-
samples (in the range of 100–2,700 μg from a 750-ml culture) siderations for the design of an improved signal peptide for
were obtained from different groups of human target proteins the expression in E. coli.
when fused to OsmY. Importantly, the authors suggested that
OsmY may facilitate the solubility, stability and folding of The n-region: position of the basic residues
target protein. These studies noted the importance of the
proteomic approach to isolating novel signal peptides and The n-region, as its name implies, is located at the extreme
excretion factors for high-level secretion of recombinant pro- N-terminal of a Sec signal peptide. A key characteristic of
teins. Another version of the proteome-based method identi- this region is the net positive charge, which arises from one
fied potential signal peptides using whole-colony matrix- or more basic residues. The positive charges (or basic resi-
assisted laser desorption–ionization time-of-flight mass spec- dues) play several roles in the protein translocation process,
trometry (MALDI-TOF MS). This high-throughput method enabling the signal peptide to insert itself into the inner
identifies extracellular proteins directly from intact bacterial membrane by forming a loop with the negatively
colonies and has been used for the identification of type VII charged lipid head groups via electrostatic interactions
secretory proteins from Mycobacterium marinum and type III (Inouye and Halegoua 1980). The highly basic n-region
secretion systems from Pseudomonas aeruginosa (Champion may also promote interactions with SRP, influencing the
et al. 2012). This method specifically detects extracellular signal peptide’s route to the SecA/SecYEG translocase
proteins that attach to the cell surface, limiting its detection (Peterson et al. 2003). Removing the basic residues (by
to a subset of protein secretion systems. reducing the length or substituting with neutral/acidic
The approaches outlined previously require a heavy in- residues) has various effects on the export process,
vestment in labour (signal peptide library construction and including a reduced rate of export (Nesmeyanova et al.
high-throughput screening) and/or advanced equipment 1997), a decreased rate of protein synthesis (Inouye et
(two-dimensional gel electrophoresis and MALDI-TOF al. 1982), a change in the signal peptide cleavage site
Appl Microbiol Biotechnol (2013) 97:3811–3826 3817

Table 3 List of several publicly available signal sequence databases

Database Target Features URL Reference

Effective Bacterial signal Predicts effector (proteins with a effectors.org Jehl et al. (2011)
sequences signal peptide and eukaryotic
domain) in bacterial genome
(symbiotic and pathogenic)
SPdb Signal sequences Experimentally verified signal proline.bic.nus.edu.sg/spdb/ Choo et al. (2005)
of archaea, sequence from Swiss-Prot
prokaryotes and
eukaryotes
CoBaltDB Bacterial and archaeal Integrates 43 in silico tools to www.umr6026.univ-rennes1.fr/english/ Goudenege et al.
proteome predict protein localisation home/research/basic/software/cobalten (2010)
from genome sequence;
provides links to external
databases (NCBI and KEGG)
Signal Peptide Signal peptides from Collection of signal peptides www.signalpeptide.de/ –
Website eukaryotes, from UniProt Knowledgebase
prokaryotes Release 14.7
and viruses
EXProt Gram-positive and Uses a weight matrix algorithm 142.51.14.12/Laurentian/Home/Departments/ Saleh et al. (2001)
gram-negative and two artificial neural Biology/Faculty_and_Staff/Professors/
signal peptides networks to predict signal saleh/
peptides from genome sequence ExProt.htm?Laurentian_Lang=en-CA
Swiss-Prot Prokaryote (426) and Experimentally verified signal ftp.ebi.ac.uk/pub/contrib/swissprot/testsets/ Menne et al. (2000)
signal test eukaryote (1,142) peptide dataset signal/data/
set signal peptides

(Suominen et al. 1995) or no discernible effects (Vlasuk of the lipid molecule (Nesmeyanova et al. 1997) and the
et al. 1983). acidic residues of SecA (Gelis et al. 2007), thus stabilising
As these substitution experiments suggest that the pres- the hydrogen bonding of the signal peptide α-helix
ence of (one or more) basic residues in the n-region may be complex.
necessary for the development of a good signal peptide, The preference for a basic residue, particularly lysine, at
optimisation of the peptide may require an examination of P2 is challenged by recent findings from Zalucki et al.
the presence and location of these residues. For maximal (2007), which argued with experimental evidence that a
secretion performance, two factors need to be taken into strong initiation of translation—and not the positive
consideration: (1) the positions of the basic residues and charge—at P2 confers a more efficient signal peptide.
(2) the magnitude of the net positive charge. The presence of Coincidentally, the best initiator codon is AAA (which
a basic residue immediately after the start codon (the second codes for lysine), a basic residue that is commonly found
codon, P2) appears to improve the rate of protein secretion. at P2 of E. coli natural signal peptides. The codon AAT
For instance, the signal peptide of MBP and ribose-binding (which codes for asparagine; neutral) is the second best
protein had high export rates when basic residues (i.e. lysine initiator codon after AAA. Substitution of AAA with AAT
and arginine) were located at P2 (Puziss et al. 1992). at P2 of the MBP and alkaline phosphatase signal se-
Similarly, signal peptide variants of alkaline phosphatase quences did not significantly change the exported levels
harbouring lysine, histidine or tyrosine at P2 yielded large of β-lactamase. In contrast, substitution of AAA with a
quantities of exported mature alkaline phosphatase weak initiator codon at P2 of the MBP signal peptide
(Nesmeyanova et al. 1997). Basic residues, particularly resulted in a significant decrease in protein export. In an-
lysine, at P2 induced the accumulation of anionic phospho- other example, a fusion with an N-terminal signal sequence
lipid, most likely through an increased interaction between (PelB, OmpA and CSP)—containing the AAA codon at
the signal peptide and the inner membrane. Anionic phos- P2—led to a strong stimulation of expression of three
pholipids are essential for efficient protein transport both in difficult-to-express proteins (granulocyte–macrophage
vivo (de Vrije et al. 1988) and in vitro (Phoenix et al. 1993) colony-stimulating factor (GM-CSF), interferon alpha 2b
and have been shown to promote the interaction of a pre- and single-chain antibody variable fragment) (Sletta et al.
cursor protein with phospholipids in a model membrane 2007). The increased transcription and translation levels
system (Keller et al. 1996). Signal peptides may establish arising from the more stable mRNA species were suggested
ionic pairs with the negative charge of the phosphate groups to be the primary factor behind the improved expression.
3818 Appl Microbiol Biotechnol (2013) 97:3811–3826

The presence of a basic residue or AAA codon at P2 does assay, the hydrophobicity levels of the signal peptide deter-
not always guarantee an optimum recombinant protein se- mined its affinity towards the protein secretion pathway in
cretion. Through saturation mutagenesis of the positively E. coli. The assay involved the expression of a modified
charged n-region (positions 2–7) of a B. subtilis α-amylase phoA that possessed two signal peptides arranged in tandem
signal peptide, Caspers et al. (2010) isolated four single- (signal peptide 1–signal peptide 2–PhoA). The efficiency of
point mutations that resulted in increased quantities of the signal peptide was determined from the preference of the
cutinase (onefold to twofold increase compared with the signal peptide cleavage point (utilisation of the first signal
wild type). Three of these mutations reduced the net charge peptide (signal peptide 1) was indicated by the signal pep-
from +3 to +2 (F2D, F2E, K4L and F6W). The mutations tide cleavage at the first site and vice versa). Mutant signal
did result in a slow rate of recombinant protein synthesis, peptides with a marginal increase in hydrophobicity and
thus preventing a jamming of the secretion translocase and length (by addition of leucine in the h-region) effectively
the induction of cell-associated proteases and stress-related outcompeted the original signal peptide for secretion effi-
genes, in addition to enabling high levels of protein secre- ciency (Chen et al. 1996). Signal peptide variants with up to
tion and processing. ten leucine residues in the hydrophobic core were efficiently
Current evidence suggests that a functional n-region translocated and cleaved. The authors suggested that an effi-
should possess a net charge of at least +1 for the efficient cient signal peptide should have a high ‘hydrophobic density’
export of the recombinant protein and that different signal in the core region. This hydrophobic core promotes high
peptides may require different magnitudes of positive levels of protein secretion, with a highly hydrophobic h-
charge for maximum efficiency. For example, export of the region compensating for a ‘weaker’ or suboptimal n-region.
MBP was the most favourable for signal peptides with at For instance, the translocation of E. coli lipoprotein was
least a net charge of +1 (Puziss et al. 1992). The maturation insensitive to the n-region positive charge for OmpF signal
of PhoA was also most efficient in the presence of a +1 net peptides containing nine or more leucine residues in the
charge (lysine, histidine or tyrosine) (Nesmeyanova et al. hydrophobic core (Hikita and Mizushima 1992b). At a low
1997). In another example, a mutant L-asparaginase II signal hydrophobicity level (eight or fewer leucines in the h-region),
peptide with a +5 net charge had higher intracellular solu- protein translocation was dependent on the n-region positive
bility and periplasmic export of a recombinant cyclodextrin charge, with the best signal peptide containing four lysine and
glucanotransferase in E. coli compared with the parent sig- eight leucine residues (MMKKKK LLLLLLLL GTANAAS).
nal peptide (net charge of +2) (Ismail et al. 2011). A subsequent in vitro experiment (exerted membrane vesicles
of E. coli) revealed that the total hydrophobicity and the length
The h-region: levelling off of hydrophobicity played a role in determining the secretion efficiency (Hikita
and Mizushima 1992a). OmpF signal peptides with a stretch
The hydrophobic core, generally known as the h-region, is a of eight leucine residues (MMKRN LLLLLLLL GTANFPG)
hallmark feature of signal peptides, functioning like a secu- or ten alanine–leucine residues (MMKRN • ALALALA
rity pass for the secreted protein to exit through the guarded LA• GTANFPG) in the hydrophobic core translocated large
gate—SecA/SecYEG. Two characteristics—helical propen- quantities of lipoprotein. Similarly, a site-directed mutagene-
sity and hydrophobicity level—are important for the recog- sis coupled with a pulse-chase analysis demonstrated that an
nition of the signal peptide by SecA. Upon contact with the h-region with five or more leucine residues rendered the n-
SecA hydrophobic groove (the substrate binding site), the region net charge unimportant for determining protein secre-
highly hydrophobic signal peptide shifts into a helical con- tion levels (Rusch et al. 2002). These findings are in agree-
formation (Chou and Gierasch 2005), triggering the SecA- ment with early cross-linking data that suggested that the
ATPase activity (Kebir and Kendall 2002). Either a deletion interactions between the signal peptides and SecA are propor-
or a substitution of the hydrophobic residues dramatically tional to the length (Mori et al. 1997) and hydrophobicity
impairs protein export (Suominen et al. 1995; Izard and levels of the h-region (Valent et al. 1998).
Kendall 1994), resulting in pre-protein accumulation at the The extent of the effect of h-region hydrophobicity on the
cell membrane with a non-native conformation. A sufficient optimisation of the signal peptide may depend on the pas-
level of hydrophobicity is needed for efficient processing of senger protein. As stated earlier, highly hydrophobic signal
the signal peptide and translocation of the protein. Rusch et p e p ti d e s p r o m o t e S R P - d e p e n d e n t r o u t e s t o t h e
al. (1994) demonstrated that at least five leucine residues SecA/SecYEG translocon. Increasing the hydrophobicity
were required in the h-region for the efficient function of an of natural signal peptides may switch the targeting route
alkaline phosphatase signal peptide. from a SecB-dependent to an SRP-dependent pathway. For
Increasing the hydrophobicity levels of the h-region can example, a LamB signal peptide with four additional hydro-
improve the rate of protein secretion. As Chen et al. (1996) phobic residues avoided recognition by SecB and was
elegantly demonstrated using a signal peptide competition rerouted to an SRP-dependent pathway for export, differing
Appl Microbiol Biotechnol (2013) 97:3811–3826 3819

from its less hydrophobic parent peptide (Bowers et al. competitive signal peptides (with a ten-leucine h-region) sig-
2003). In this case, the extent of the SecB and SRP pathway nificantly inhibited the translocation of native β-lactamase.
efficiencies for recombinant protein secretion was not deter- Reduced expression/processing of the native protein may
mined. A secreted protein that requires some sort of post- have severely influenced important cell functions and retarded
translational modification might benefit from the SecB path- cell growth, creating a major problem in the up-scaling to
way (Bensing et al. 2007), as the SRP-dependent pathway production. In this case, an optimisation of the hydrophobicity
occurs co-translationally, preventing any form of cytoplas- levels by substituting less hydrophobic residues with more
mic modification. By using SRP-dependent signal peptides hydrophobic residues (e.g. leucine) should be performed, one
(i.e. DsbA, TorT, TolB or SfmC), Steiner et al. (2006) residue at a time, until an optimal balance between protein
produced more than a 1,000-fold enrichment of phage- secretion and cell growth is obtained.
displayed protein compared with the Sec-dependent signal
peptides (i.e. LamB, MalE, MglB, OmpA, PelB or PhoA). The c-region: beyond the ‘−3, −1 rule’
The unnecessary processing of the phage-displayed protein
that follows the SecB route stalled in the cytoplasm and After passing through the SecYEG translocon, the signal
retarded the secretion of the protein. Non-problematic pro- peptide is embedded into the inner membrane and has to be
teins may benefit from exportation through the SecB- cleaved off from the mature protein to complete the secre-
dependent pathway, as the SecB pathway approach has a tion process. The cleavage of the signal peptide requires the
large capacity for protein export in E. coli (more than 90 % recognition and proteolytic action of signal peptidase at a
of secreted protein in E. coli follows a SecB-dependent specific site on the signal peptide. This site is commonly
route). As stated earlier in this article, SRP binds to highly known as the c-region, and it is most likely the least ambig-
hydrophobic signal peptides that form long α-helix struc- uous region to optimise. Based on the involvement of the c-
tures. To reduce the affinity of highly hydrophobic signal region with signal peptidase I in a specific enzyme–substrate
peptides towards SRP, a helix breaker (particularly glycine) interaction, special limits on the sequence characteristics,
can be introduced within the hydrophobic core of the signal conformation and polarity may be imposed on the cleavage
peptide. Extending the α-helix structure, through substitu- site in the c-region. The most prominent rule is the ‘−3, −1
tion of glycine with cysteine and leucine (mutations G10C rule’ or AXA motif (von Heijne 1984), with position −1 and
and G10L), of a PhoE signal peptide significantly shifted the −3 preceding the cleavage site favouring residues with small
affinity of this peptide towards the SRP component (P48) neutral side chains (particularly alanine). The c-region
and away from the SecB pathway (Adams et al. 2002). should have a five-residue length to ensure both high export
Similarly, the replacement of glycine within the hydropho- and cleavage efficiency (Suominen et al. 1995). The effi-
bic region of the GspB signal peptide reduced the accessory ciency of the signal peptide cleavage significantly influ-
Sec-dependent transport (i.e. SecB), inducing canonical ences protein secretion levels. Using a novel biosensor
Sec-dependent exportation (Bensing et al. 2007). The pres- technique, Geukens et al. (2004) demonstrated that the
ence of a helix breaker in the hydrophobic core of the signal cleavage of the signal peptide was the rate-limiting step
peptide can dramatically affect recombinant protein export. for protein secretion in Streptomyces lividans. The authors
The introduction of a glycine residue in the middle of the h- also reported that the signal peptidase interacted with the
region doubled the periplasmic and extracellular recombi- signal peptide at the region −8 to +1. In another example,
nant cyclodextrin glucanotransferase activity, compared coexpression of signal peptidase I (sipM) increased
with the no-glycine, wild-type signal peptide (Jonet et al. dextransucrase secretion in Bacillus megaterium by 3.7-fold
2012; Low et al. 2012). The position of the helix breaker (Malten et al. 2005).
(glycine) played a role in protein export levels. A glycine Recent studies have further detailed the requirements for
residue at position −7 improved the pre-protein export com- an efficient c-region. Using a directed evolution approach,
pared with a glycine residue at position −5 or −9 (the Karamyshev et al. (1998) provided new experimental evi-
predicted total length of the h-region was 12 residues). dence to support the ‘−3, −1 rule’, with alanine preferred for
This increase in the export phenotype may have resulted position −1, while position −3 can accommodate several
from a change of the targeting pathway or another residues (i.e. alanine, serine, glycine, leucine and cysteine).
unidentified mechanism. Position −2 favoured large amino acids (i.e. phenylalanine,
Reports in the literature have suggested that the best signal tyrosine, leucine and histidine) for efficient processing. This
peptide should be the most hydrophobic and have a glycine observation was similar to the report by Wrede et al. (1998)
residue in the middle of the hydrophobic core. At first glance, on the design of the c-region using computer-based tech-
this may seem logical to some extent, but there are several niques. Position −4 favoured serine, phenylalanine, tyrosine,
difficulties that prevent this sequence from becoming a work- leucine and cysteine, but not lysine and glycine. Position −5
able solution. As demonstrated by Rusch et al. (1994), highly preferred mid-sized amino acids, such as serine. A
3820 Appl Microbiol Biotechnol (2013) 97:3811–3826

conformational analysis of the cleavage site (−5 to −1 region) region, also termed the ‘export initiation domain’, which
suggested that an extended β-conformation was optimal for extends up to 30 residues long immediately downstream of
the signal peptidase binding pocket/active site. This finding is the signal peptidase cleavage site (Andersson and von
in good agreement with the available X-ray structure of E. coli Heijne 1991). The nature of the charged residues in this
signal peptidase (Paetzel et al. 1998). By using in silico region influences the discharge of the pre-protein across the
docking, the authors showed that the side chain of alanine at inner membrane of E. coli. This region is largely hydrophil-
position −1 interacted with the signal peptidase pocket S1 ic, and deletions in this region can cause proteins to accu-
(formed by methionine-91, isoleucine-144, leucine-95 and mulate in the cytoplasm and aggregate as inclusion bodies
isoleucine-86). The side chain of alanine at position −3 ac- (Kim et al. 2007). An introduction of basic residues into this
commodated a shallow hydrophobic pocket S3 (formed by region can block (Andersson and von Heijne 1991; Geller et
phenylalanine-84, isoleucine-86, isoleucine-101, valine-132, al. 1993) or reduce protein export (Campion et al. 1997) and
isoleucine-144 and aspartic acid-142). The S3 pocket was change the signal peptidase I cleavage point (Itoh et al.
larger than the S1 pocket, making room for a larger amino 1990), regardless of the targeting pathway (SecB-dependent
acid (i.e. valine, leucine and isoleucine) at the −3 position. The or SRP-dependent) (Tian and Bernstein 2009). Furthermore,
stringent substrate specificity of the signal peptidase contrib- a recombinant protein fused at position +2 (relative to the
uted to the orientation of the signal peptide’s side chain signal peptide cleavage point) of a signal peptide was more
towards the binding site and to the hydrogen bonds between efficiently secreted than the protein fused at position +1
the signal peptidase and the signal peptide at positions −3, −2 (Geukens et al. 2004), suggesting that the pro-region (posi-
and −1 (Choo et al. 2005). tion +1) is involved in protein translocation. Experiments
Another approach to optimise the c-region sequence is de using in silico docking suggested that the signal peptidase
novo design using computer-based techniques. The combina- interacts with the signal peptide not only at the AXA site but
tion of an artificial neural network and an evolutionary dis- also in the pro-region, from position +1 to +6 (Choo et al.
tance matrix is an attractive method for designing a functional 2005). A statistical analysis of 107 experimentally deter-
and efficient c-region. For instance, Paul Wrede’s group mined signal peptidase substrates suggested a preference for
designed an ‘idealised’ cleavage site using the (1, λ)-evolution certain residues in this region (position +1 to +6). The
strategy and artificial neutral networks to generate optimum c- validity of these findings requires further analysis.
region sequences and to estimate the quality of the generated In contrast with the n-region, the pro-region requires a net
sequences, respectively (Schneider and Wrede 1994; Wrede et negative charge for efficient protein export. For example,
al. 1998). They demonstrated that the Miyata matrix (Miyata substitution of the basic residues with acidic residues dras-
et al. 1979), which serves as a calculator for the evolution tically increased recombinant protein export (Campion et al.
distance measurement between amino acids, produced higher- 1997; Yamanaka and Okamoto 1996). The need for a neg-
quality sequences. In addition to confirming the ‘−3, −1 rule’, atively charged pro-region is not unique to E. coli; this
the authors also indicated the amino acid preferences at posi- requirement was also observed in other gram-positive bac-
tions −6, −2 and −7 to −10, with alanine at position −6, a teria (Kajikawa et al. 2010; Kouwen et al. 2010), suggesting
relatively large residue (such as tryptophan) at position −2 and that the pro-region is actively involved in both pre-protein
phenylalanine (hydrophobic residue) at positions −7 to −10. targeting and translocation. The optimal values of both the
The predicted ‘idealised’ c-region (SME2) was reported to be magnitude of the charge and the position of the acidic
FFFFGWYGWA↓AC, where ‘↓’ indicates the cleavage point. residues may depend on the target protein. Recombinant
SME2 fused to a ribonuclease T1 was recognised and cleaved phosphoglycerate kinase required a net charge of −2 or −3
by signal peptidase I as efficiently as OmpA signal peptide on the first 15 residues for efficient export, while the en-
(LAGFATVAQA↓AC) (Wrede et al. 1998), and the exported zymes glyceraldehyde-3-phosphate dehydrogenase and eno-
quantity of ribonuclease protein, measured using SDS-PAGE, lase required a net charge of −6 (Tian and Bernstein 2009).
appeared to be higher relative to that obtained with the OmpA The net negatively charged pro-region might form loop struc-
signal peptide. tures with the positively charged n-region, complementing the
membrane electrochemical potential (the ‘inside positive
The pro-region: the fourth musketeer (factor) in action? rule’). The membrane potential across the E. coli inner mem-
brane is characterised by a positive charge on the periplasmic
The presence of a secretion signal does not always guarantee surface. Kaderbhai et al. (2004) classified the performance of
an efficient secretion of the recombinant protein. protein exportation based on the nature of the two residues
Problematic pre-proteins that fold prematurely or aggregate immediately adjacent to the signal peptide cleavage site (po-
can be translocated by using an overexpression of chaper- sition +1 and +2). Hyperexporters contained acidic residues,
ones to reduce their folding rate (Park et al. 2004; Kwon et hypoexporters contained basic residues or cysteine and inter-
al. 2002). Another important region to consider is the pro- mediate exporters containing predominantly neutral residues.
Appl Microbiol Biotechnol (2013) 97:3811–3826 3821

Hyperexporters bound less frequently to the membrane com- protein at the SecYEG translocon, preventing the secretion
pared with hypoexporters and intermediate exporters. Indeed, of the endogenous proteins that are important to maintain cell
a comparative analysis of the E. coli and B. subtilis cytoplas- survival.
mic and secretory proteins showed that most secretory pro-
teins have neutral or net negative charges for the first 10–15
residues (Tian and Bernstein 2009). Conclusions and future prospects
The steric bulkiness of amino acids at position +1 is
another factor that may influence protein export. Smaller Constructing a system of secretory expression for recombi-
residues, such as cysteine, glutamine, threonine and serine, nant proteins is a simple task, but achieving a substantially
at position +1 ensured high export levels of the mammalian high level of production requires one to balance between
globular cytochrome b5 precursor in E. coli (Kaderbhai et al. gene expression, the protein transport process, cell fitness
2010). The authors suggested that large (i.e. isoleucine) and (viability) and the composition of the medium. For the
rigid (i.e. proline) amino acids at position +1 may restrict the process of protein transport specifically, the main obstacles
activity of signal peptidase I on the signal peptide (steric can be ameliorated with an improved understanding of the
hindrance), causing a lower cleavage and discharge level of mechanisms of protein secretion. The Sec signal peptide
the mature protein. Small and flexible amino acids in the may appear to be a simple and redundant molecule, but this
pro-region allowed a rapid maturation with improved export short polypeptide contains all of the information needed to
levels of subtilase, possibly due to the tendency of the drive protein secretion through a sequential interaction with
secreted protein to maintain a translocation-competent state the Sec machinery and the passenger protein. Recent find-
(Fang et al. 2010). In another example, a deletion of the pro- ings have suggested that the distinct regions of the signal
region (n-terminal 16 residues) rendered transglutaminase peptide (i.e. the n-, h-, c- and pro-regions) do not work
secretion-defective and insoluble (Liu et al. 2011). A model independently but are organised and function in a coopera-
of the protein suggested that the pro-region may be involved tive manner. A summary of a more efficient signal peptide
in protein folding, suggesting that this region is essential for design is proposed as in Fig. 2. We suggest that an ideal
efficient protein secretion. signal peptide should have an optimal and balanced ‘design’
of properties for each of the distinct regions. For this ex-
Signal sequence as an intramolecular chaperone pression system, solving the X-ray structure of the Sec
machinery (i.e. SecA and the signal peptidase) could pro-
The role of the signal peptide in protein secretion is not vide details as to their interactions with the signal peptide.
limited to the interactions with the secretory machinery (i.e. The future of recombinant protein production may rely
SecA, signal peptidase), but also onto the passenger protein. heavily on using modified microbial cell factories with an
Signal peptides can reduce folding rates (Tomkiewicz et al. efficient protein secretion (and/or excretion) system. The
2008) and change both the kinetic and thermodynamic search for a more efficient signal peptide and secretion
destabilisation of the passenger protein (Beena et al. 2004), pathway will remain as an important highlight and challenge
independent of the folding nature of the target protein. to realising that goal. Several exciting challenges that re-
Together with SecB, the signal peptides (particularly the hy- main are described as follows:
drophobic core of the peptide) can interact with the hydropho-
bic regions of the passenger protein to maintain a non-native 1. The hydrophobic core is most likely the most dominant
conformation and a transport-competent state—properties that characteristic in influencing the capacity and efficiency
are crucial for all secreted proteins. Recent findings suggest of a signal peptide during protein secretion. Despite its
that signal peptides can be engineered to not only enhance importance, little systematic work has been performed
protein export but also to improve the physiological state of on engineering an improved hydrophobic core. Current
the cell. By increasing the net charge on the n-region of the L- research has suggested that an enhancement in the h-
asparaginase II signal peptide (from +2 to +5), Ismail et al. region’s efficiency may require an increase in hydro-
(2011) increased both the protein solubility and the periplas- phobicity levels. Less is known about the effect of
mic export of recombinant cyclodextrin glucanotransferase in different hydrophobic residues in the hydrophobic core
E. coli compared with the wild-type signal peptide. Cell on the efficiency of the signal peptide in driving trans-
viability also improved significantly, most likely due to a location. Two questions that remain to be answered
strong cell membrane. More studies should be performed to include the following: Are highly hydrophobic residues
confirm these observations. One possible explanation may be (i.e. isoleucine) favourable over less hydrophobic or
that the mutant signal peptide may relieve a phenomenon non-hydrophobic residues (i.e. alanine, glycine) or vice
called ‘saturated translocation’ (Fu et al. 2005), which is versa? Is there a maximum hydrophobicity threshold for
caused by an accumulation of the transport-deficient pre- better signal peptide design? The recent discovery that
3822 Appl Microbiol Biotechnol (2013) 97:3811–3826

glycine (a non-hydrophobic residue) and its location in a living organism for the generation of knowledge and
the h-region significantly influenced the secretion of products, can be used to achieve the best performance in
recombinant proteins in E. coli (Jonet et al. 2012) re- an integrated manner. In this case, E. coli is advantageous,
vealed our limited understanding of this region. An as it is the most well-known organism with the most
exhaustive approach, such as directed evolution, that complete genome-scale metabolic reconstruction model
probes the effects of different residues in the hydropho- (Orth et al. 2011). A recent study demonstrated the feasi-
bic core could provide an improved understanding of bility and practicality of this approach. By using a 13C
the fundamental requirements for the hydrophobic core. metabolic flux analysis, Umakoshi et al. (2011) identified
2. One of the more ambitious goals in signal peptide the rate-limiting step (NADH/NAD + balance) in
research is the development of a cross-host, universal transglutaminase secretion in C. glutamicum,
secretion signal peptide. The ability of signal peptides to counteracting the negative effects and generating a 1.4-
be recognised and translocated in a wide range of hosts fold increase in the production of the target protein.
would ensure flexibility in protein production, simpli- 4. Another area that is worth pursuing is the potential
fying the optimisation processes. Early studies have activity of the signal peptide as a molecular chaperone.
demonstrated that several natural signal peptides exhibit Recent evidence suggests that the ‘influence’ of signal
some degree of functionality in other heterologous ex- peptide is exerted not only onto its passenger protein but
pression hosts (Uhlen and Abrahmsen 1989; Allet et al. may extend towards the well-being of host cell (cell
1997; Golden et al. 1998; Coloma et al. 1992; physiology, cell viability, etc.). A mutant signal peptide
Kirkpatrick et al. 1995; Gray et al. 1985; Asakura et with an increased positive charge in the n-region (+5)
al. 1995). Humphreys et al. (2000) demonstrated that significantly improved both the cell viability and the
eukaryotic signal sequences can be engineered to per- recombinant protein secretion compared to the wild-
form more efficiently in E. coli through codon optimi- type signal peptide (+2) (Ismail et al. 2011). Whether
sation at the 5′ end of the coding sequence. Wang et al. this phenomenon is the consequence of improved pro-
(2001) and Tan et al. (2002) also engineered a novel tein secretion (i.e. reduced inclusion body formation,
secretion signal peptide that was capable of targeting reduced stress, de-saturation of protein translocon,
recombinant protein secretion in E. coli, Saccharomyces etc.) needs further clarification. The question remains
cerevisiae and six different eukaryotic cell lines. The of how the signal peptide reacts with cellular machinery
authors outlined several key characteristics for cross- to perform such functions. One may rely on advanced
host secretion signal peptides, including the ability to computational algorithms (i.e. molecular dynamics
contain the three typical domains observed in all signal analysis) to simulate protein folding and reveal possible
peptides, to obey the ‘−3, −1’ rule and to comprise an interactions between the signal peptide and its passen-
optimal ratio of charge-to-hydrophobicity level (Tan et ger protein. A systems-level ‘omics’ approach, such as
al. 2002). Enhancing the performance of the secretion transcriptomics or proteomics, could provide insight on
signal in multiple expression hosts while maintaining the effect of the design of the signal peptide on the cell.
flexibility (i.e. functionality in multiple host cells) can
be challenging. The use of computational design to Acknowledgements We would like to thank Noor Faizah Ismail for
match and complement the requirements of efficient the critical reading of the manuscript. Support from Abdul Munir
secretion in multiple expression hosts should produce Abdul Murad, Amir Rabu and Farah Diba Abu Bakar of Universiti
Kebangsaan Malaysia and Raha Abdul Rahim of Universiti Putra
desirable outcomes. Genome-scale comparisons of sig-
Malaysia is greatly appreciated. This work was supported by the
nal peptides between common prokaryotic hosts (i.e. E. Genomics and Molecular Biology Initiatives Programme of the Ma-
coli and B. subtilis) and eukaryotic hosts (i.e. CHO, S. laysia Genome Institute, Ministry of Science, Technology and Innova-
cerevisiae) could reveal similarities and differences in tion Malaysia (Project No. 07-05-MGI-GMB011) and the Exploration
Research Grant Scheme (R.J130000.7835.4L046), Universiti
signal peptide architecture and design. Furthermore, the
Teknologi Malaysia. Kheng Oon Low is a researcher of Universiti
recent discovery of the X-ray structure of Sec machin- Teknologi Malaysia under the Post-Doctoral Fellowship Scheme.
ery (i.e. SecA) may also aid in the understanding of
signal peptide interactions with the respective systems.
3. Protein expression and secretion in biological systems are
References
complex processes that involve a plethora of sequential
protein–protein interactions. Improving recombinant pro-
tein secretion should not be confined solely to engineering Adams H, Scotti PA, De Cock H, Luirink J, Tommassen J (2002) The
presence of a helix breaker in the hydrophobic core of signal
more efficient interactions between the signal peptide and sequences of secretory proteins prevents recognition by the
the Sec machinery. Systems biotechnology, a systems- signal-recognition particle in Escherichia coli. Eur J Biochem
level approach that considers all biological processes in 269:5564–5571
Appl Microbiol Biotechnol (2013) 97:3811–3826 3823

Allet B, Bernard AR, Hochmann A, Rohrbach E, Graber P, Magnenat Choo K, Tan T, Ranganathan S (2005) SPdb—a signal peptide data-
E, Mazzei GJ, Bernasconi L (1997) A bacterial signal peptide base. BMC Bioinformatics 6:249
directs efficient secretion of eukaryotic proteins in the baculovirus Chou KC, Shen HB (2007) Signal-CF: a subsite-coupled and window-
expression system. Protein Expr Purif 9:61–68 fusing approach for predicting signal peptides. Biochem Biophys
Andersson H, Von Heijne G (1991) A 30-residue-long “export initia- Res Commun 357:633–640
tion domain” adjacent to the signal sequence is critical for protein Chou Y-T, Gierasch LM (2005) The conformation of a signal peptide
translocation across the inner membrane of Escherichia coli. Proc bound by Escherichia coli preprotein translocase SecA. J Biol
Natl Acad Sci U S A 88:9751–9754 Chem 280:32753–32760
Asakura A, Minami M, Ota Y (1995) Convenient desktop-scale pro- Coloma MJ, Hastings A, Wims LA, Morrison SL (1992) Novel vectors
duction of the extracellular domain of the human growth hormone for the expression of antibody molecules using variable regions
receptor by an insect–baculovirus secretion system using a generated by polymerase chain reaction. J Immunol Methods
protein-free culture. Biosci Biotechnol Biochem 59:1976–1978 152:89–104
Bakkes PJ, Jenewein S, Smits SH, Holland IB, Schmitt L (2010) The De Vrije T, De Swart RL, Dowhan W, Tommassen J, De Kruijff B
rate of folding dictates substrate secretion by the Escherichia coli (1988) Phosphatidylglycerol is involved in protein translocation
hemolysin type 1 secretion system. J Biol Chem 285:40573– across Escherichia coli inner membranes. Nature 334:173–175
40580 Degering C, Eggert T, Puls M, Bongaerts J, Evers S, Maurer KH, Jaeger
Beena K, Udgaonkar JB, Varadarajan R (2004) Effect of signal peptide KE (2010) Optimization of protease secretion in Bacillus subtilis and
on the stability and folding kinetics of maltose binding protein. Bacillus licheniformis by screening of homologous and heterologous
Biochemistry 43:3608–3619 signal peptides. Appl Environ Microbiol 76:6370–6376
Bensing BA, Siboo IR, Sullam PM (2007) Glycine residues in the Demain AL, Vaishnav P (2009) Production of recombinant proteins by
hydrophobic core of the GspB signal sequence route export to- microbes and higher organisms. Biotechnol Adv 27:297–306
ward the accessory Sec pathway. J Bacteriol 189:3846–3854 Economou A, Wickner W (1994) SecA promotes preprotein translo-
Berks BC, Sargent F, Palmer T (2000) The Tat protein export pathway. cation by undergoing ATP-driven cycles of membrane insertion
Mol Microbiol 35:260–274 and deinsertion. Cell 78:835–843
Bowers CW, Lau F, Silhavy TJ (2003) Secretion of LamB-LacZ by the Fang N, Zhong CQ, Liang X, Tang XF, Tang B (2010) Improvement of
signal recognition particle pathway of Escherichia coli. J extracellular production of a thermophilic subtilase expressed in
Bacteriol 185:5697–5705 Escherichia coli by random mutagenesis of its N-terminal
Brockmeier U, Caspers M, Freudl R, Jockwer A, Noll T, Eggert T propeptide. Appl Microbiol Biotechnol 85:1473–1481
(2006) Systematic screening of all signal peptides from Fariselli P, Finocchiaro G, Casadio R (2003) SPEPlip: the detection of
Bacillus subtilis: a powerful strategy in optimizing heterolo- signal peptide and lipoprotein cleavage sites. Bioinformatics
gous protein secretion in gram-positive bacteria. J Mol Biol 19:2498–2499
362:393–402 Frank K, Sippl MJ (2008) High-performance signal peptide prediction
Campion SR, Elsasser E, Chung R (1997) Amino-terminal charge based on sequence alignment techniques. Bioinformatics
affects the periplasmic accumulation of recombinant heregulin/ 24:2172–2176
EGF hybrids exported using the Escherichia coli alkaline phos- Fu ZB, Ng KL, Lam TL, Wong WK (2005) Cell death caused by
phatase signal sequence. Protein Expr Purif 10:331–339 hyper-expression of a secretory exoglucanase in Escherichia coli.
Caspers M, Brockmeier U, Degering C, Eggert T, Freudl R (2010) Protein Expr Purif 42:67–77
Improvement of Sec-dependent secretion of a heterologous model Gelis I, Bonvin AM, Keramisanou D, Koukaki M, Gouridis G,
protein in Bacillus subtilis by saturation mutagenesis of the N- Karamanou S, Economou A, Kalodimos CG (2007) Structural
domain of the AmyE signal peptide. Appl Microbiol Biotechnol basis for signal-sequence recognition by the translocase motor
86:1877–1885 SecA as determined by NMR. Cell 131:756–769
Champion MM, Williams EA, Kennedy GM, Champion PA (2012) Geller B, Zhu HY, Cheng S, Kuhn A, Dalbey RE (1993) Charged
Direct detection of bacterial protein secretion using whole colony residues render pro-OmpA potential dependent for initiation of
proteomics. Mol Cell Proteomics 11:596–604 membrane translocation. J Biol Chem 268:9442–9447
Chen R (2012) Bacterial expression systems for recombinant protein Georgiou G, Segatori L (2005) Preparative expression of secreted
production: E. coli and beyond. Biotechnol Adv 30:1102–1107 proteins in bacteria: status report and future prospects. Curr
Chen H, Kim J, Kendall DA (1996) Competition between functional Opin Biotechnol 16:538–545
signal peptides demonstrates variation in affinity for the secretion Geukens N, Frederix F, Reekmans G, Lammertyn E, Van Mellaert L,
pathway. J Bacteriol 178:6658–6664 Dehaen W, Maes G, Anné J (2004) Analysis of type I signal
Chen N, Hong F-L, Wang H-H, Yuan Q-H, Ma W-Y, Gao X-N, Shi R, peptidase affinity and specificity for preprotein substrates.
Zhang R-J, Sun C-S, Wang S-B (2012) Modified recombinant Biochem Biophys Res Commun 314:459–467
proteins can be exported via the Sec pathway in Escherichia coli. Golden A, Austen DA, Van Schravendijk MR, Sullivan BJ, Kawasaki
PLoS One 7:e42519 ES, Osburne MS (1998) Effect of promoters and signal sequences
Cho AR, Yoo SK, Kim EJ (2000) Cloning, sequencing and expression on the production of secreted HIV-1 gp120 protein in the
in Escherichia coli of a thermophilic lipase from Bacillus baculovirus system. Protein Expr Purif 14:8–12
thermoleovorans ID-1. FEMS Microbiol Lett 186:235–238 Goudenege D, Avner S, Lucchetti-Miganeh C, Barloy-Hubler F (2010)
Choi JH, Lee SY (2004) Secretory and extracellular production of CoBaltDB: complete bacterial and archaeal orfeomes subcellular
recombinant protein using Escherichia coli. Appl Microbiol localization database and associated resources. BMC Microbiol
Biotechnol 64:625–635 10:88
Choi JH, Jeong KJ, Kim SC, Lee SY (2000) Efficient secretory produc- Gouridis G, Karamanou S, Gelis I, Kalodimos CG, Economou A
tion of alkaline phosphatase by high cell density culture of recom- (2009) Signal peptides are allosteric activators of the protein
binant Escherichia coli using the Bacillus sp. endoxylanase signal translocase. Nature 462:363–367
sequence. Appl Microbiol Biotechnol 53:640–645 Gray GL, Baldridge JS, Mckeown KS, Heyneker HL, Chang CN
Choi JH, Keum KC, Lee SY (2006) Production of recombinant pro- (1985) Periplasmic production of correctly processed human
teins by high cell density culture of Escherichia coli. Chem Eng growth hormone in Escherichia coli: natural and bacterial signal
Sci 61:876–885 sequences are interchangeable. Gene 39:247–254
3824 Appl Microbiol Biotechnol (2013) 97:3811–3826

Hardy SJ, Randall LL (1991) A kinetic partitioning model of selective Kall L, Krogh A, Sonnhammer EL (2007) Advantages of combined
binding of nonnative proteins by the bacterial chaperone SecB. transmembrane topology and signal peptide prediction—the
Science 251:439–443 Phobius web server. Nucleic Acids Res 35:W429–W432
Hikita C, Mizushima S (1992a) Effects of total hydrophobicity and Karamyshev AL, Karamysheva ZN, Kajava AV, Ksenzenko VN,
length of the hydrophobic domain of a signal peptide on in vitro Nesmeyanova MA (1998) Processing of Escherichia coli alkaline
translocation efficiency. J Biol Chem 267:4882–4888 phosphatase: role of the primary structure of the signal peptide
Hikita C, Mizushima S (1992b) The requirement of a positive charge at cleavage region. J Mol Biol 277:859–870
the amino terminus can be compensated for by a longer central Kebir MO, Kendall DA (2002) SecA specificity for different signal
hydrophobic stretch in the functioning of signal peptides. J Biol peptides. Biochemistry 41:5573–5580
Chem 267:12375–12379 Keller RC, Ten Berge D, Nouwen N, Snel MM, Tommassen J, Marsh
Hiller K, Grote A, Scheer M, Munch R, Jahn D (2004) PrediSi: D, De Kruijff B (1996) Mode of insertion of the signal sequence
prediction of signal peptides and their cleavage positions. of a bacterial precursor protein into phospholipid bilayers as
Nucleic Acids Res 32:W375–379 revealed by cysteine-based site-directed spectroscopy.
Humphreys DP, Sehdev M, Chapman AP, Ganesh R, Smith BJ, King Biochemistry 35:3063–3071
LM, Glover DJ, Reeks DG, Stephens PE (2000) High-level peri- Kim CK, Lee SY, Kwon OJ, Lee SM, Nah SY, Jeong SM (2007)
plasmic expression in Escherichia coli using a eukaryotic signal Secretory expression of active clostripain in Escherichia coli. J
peptide: importance of codon usage at the 5′ end of the coding Bacteriol 131:346–352
sequence. Protein Expr Purif 20:252–264 Kirkpatrick RB, Ganguly S, Angelichio M, Griego S, Shatzman A,
Hytonen VP, Laitinen OH, Airenne TT, Kidron H, Meltola NJ, Porkka Silverman C, Rosenberg M (1995) Heavy chain dimers as well as
EJ, Horha J, Paldanius T, Maatta JA, Nordlund HR, Johnson MS, complete antibodies are efficiently formed and secreted from
Salminen TA, Airenne KJ, Yla-Herttuala S, Kulomaa MS (2004) Drosophila via a BiP-mediated pathway. J Biol Chem
Efficient production of active chicken avidin using a bacterial 270:19800–19805
signal peptide in Escherichia coli. Biochem J 384:385–390 Kotzsch A, Vernet E, Hammarström M, Berthelsen J, Weigelt J,
Inouye M, Halegoua S (1980) Secretion and membrane localization of Gräslund S, Sundström M (2011) A secretory system for
proteins in Escherichia coli. CRC Crit Rev Biochem 7:339–371 bacterial production of high-profile protein targets. Protein
Inouye S, Soberon X, Franceschini T, Nakamura K, Itakura K, Inouye Sci 20:597–609
M (1982) Role of positive charge on the amino-terminal region of Kouwen TRHM, Nielsen AK, Denham EL, Dubois J-YF, Dorenbos R,
the signal peptide in protein secretion across the membrane. Proc Rasmussen MD, Quax WJ, Freudl R, Van Dijl JM (2010)
Natl Acad Sci U S A 79:3438–3441 Contributions of the pre- and pro-regions of a Staphylococcus
Ismail NF, Hamdan S, Mahadi NM, Murad AM, Rabu A, Bakar FD, hyicus lipase to secretion of a heterologous protein by Bacillus
Klappa P, Illias RM (2011) A mutant L-asparaginase II signal subtilis. Appl Environ Microbiol 76:659–669
peptide improves the secretion of recombinant cyclodextrin Kwon MJ, Park SL, Kim SK, Nam SW (2002) Overproduction of
glucanotransferase and the viability of Escherichia coli. Bacillus macerans cyclodextrin glucanotransferase in E. coli by
Biotechnol Lett 33:999–1005 coexpression of GroEL/ES chaperone. J Microbiol Biotechnol
Itoh Y, Kanoh KI, Nakamura K, Takase K, Yamane K (1990) Artificial 12:1002–1005
insertion of peptides between signal peptide and mature protein: Li Z, Gu Z, Wang M, Du G, Wu J, Chen J (2010) Delayed supple-
effect on secretion and processing of hybrid thermostable α- mentation of glycine enhances extracellular secretion of the re-
amylases in Bacillus subtilis and Escherichia coli cells. J Gen combinant alpha-cyclodextrin glycosyltransferase in Escherichia
Microbiol 136:1551–1558 coli. Appl Microbiol Biotechnol 85:553–561
Izard JW, Kendall DA (1994) Signal peptides: exquisitely designed Liu S, Zhang D, Wang M, Cui W, Chen K, Liu Y, Du G, Chen J, Zhou
transport promoters. Mol Microbiol 13:765–773 Z (2011) The pro-region of Streptomyces hygroscopicus
Jehl M-A, Arnold R, Rattei T (2011) Effective—a database of predict- transglutaminase affects its secretion by Escherichia coli. FEMS
ed secreted bacterial proteins. Nucleic Acids Res 39:D591–D595 Microbiol Lett 324:98–105
Jobling MG, Palmer LM, Erbe JL, Holmes RK (1997) Construction Low KO, Jonet MA, Ismail NF, Illias RM (2012) Optimization of a
and characterization of versatile cloning vectors for efficient de- Bacillus sp signal peptide for improved recombinant protein se-
livery of native foreign proteins to the periplasm of Escherichia cretion and cell viability in Escherichia coli: is there an optimal
coli. Plasmid 38:158–173 signal peptide design? Bioengineered 3:334–338
Joly JC, Leung WC, Swartz JR (1998) Overexpression of Escherichia Luirink J, Dobberstein B (1994) Mammalian and Escherichia coli
coli oxidoreductases increases recombinant insulin-like growth signal recognition particles. Mol Microbiol 11:9–13
factor-I accumulation. Proc Natl Acad Sci U S A 95:2773–2777 Malten M, Nahrstedt H, Meinhardt F, Jahn D (2005) Coexpression of
Jonet MA, Mahadi NM, Murad AM, Rabu A, Bakar FD, Rahim RA, the type I signal peptidase gene sipM increases recombinant
Low KO, Illias RM (2012) Optimization of a heterologous signal protein production and export in Bacillus megaterium MS941.
peptide by site-directed mutagenesis for improved secretion of Biotechnol Bioeng 91:616–621
recombinant proteins in Escherichia coli. J Mol Microbiol Mathiesen G, Sveen A, Brurberg M, Fredriksen L, Axelsson L, Eijsink
Biotechnol 22:48–58 V (2009) Genome-wide analysis of signal peptide functionality in
Kaderbhai MA, Davey HM, Kaderbhai NN (2004) A directed evolution Lactobacillus plantarum WCFS1. BMC Genomics 10:425
strategy for optimized export of recombinant proteins reveals critical Matsuyama S, Fujita Y, Mizushima S (1993) SecD is involved in the
determinants for preprotein discharge. Protein Sci 13:2458–2469 release of translocated secretory proteins from the cytoplasmic
Kaderbhai NN, Ahmed K, Kaderbhai MA (2010) Export of a membrane of Escherichia coli. EMBO J 12:265–270
hyperexpressed mammalian globular cytochrome b5 precursor in Menne KM, Hermjakob H, Apweiler R (2000) A comparison of signal
Escherichia coli is dramatically affected by the nature of the sequence prediction methods using a test set of signal peptides.
amino acid flanking the secretory signal sequence cleavage bond. Bioinformatics 16:741–742
Protein Sci 19:1344–1353 Mergulhao FJM, Summers DK, Monteiro GA (2005) Recombinant
Kajikawa A, Ichikawa E, Igimi S (2010) Development of a highly protein secretion in Escherichia coli. Biotechnol Adv 23:177–202
efficient protein-secreting system in recombinant Lactobacillus Miyata T, Miyazawa S, Yasunaga T (1979) Two types of amino acid
casei. J Microbiol Biotechnol 20:375–382 substitutions in protein evolution. J Mol Evol 12:219–236
Appl Microbiol Biotechnol (2013) 97:3811–3826 3825

Mori H, Araki M, Hikita C, Tagaya M, Mizushima S (1997) The Schneider G, Wrede P (1994) The rational design of amino acid
hydrophobic region of signal peptides is involved in the interac- sequences by artificial neural networks and simulated molecular
tion with membrane-bound SecA. Biochim Biophys Acta evolution: de novo design of an idealized leader peptidase cleav-
1326:23–36 age site. Biophys J 66:335–344
Mujacic M, Bader MW, Baneyx F (2004) Escherichia coli Hsp31 func- Shen HB, Chou KC (2007) Signal-3L: a 3-layer approach for
tions as a holding chaperone that cooperates with the DnaK-DnaJ- predicting signal peptides. Biochem Biophys Res Commun
GrpE system in the management of protein misfolding under severe 363:297–303
stress conditions. Mol Microbiol 51:849–859 Shokri A, Sanden AM, Larsson G (2003) Cell and process design for
Muller M (2005) Twin-arginine-specific protein export in Escherichia targeting of recombinant protein into the culture medium of
coli. Res Microbiol 156:131–136 Escherichia coli. Appl Microbiol Biotechnol 60:654–664
Nesmeyanova MA, Karamyshev AL, Karamysheva ZN, Kalinin AE, Sletta H, Tøndervik A, Hakvåg S, Aune TEV, Nedal A, Aune R,
Ksenzenko VN, Kajava AV (1997) Positively charged lysine at Evensen G, Valla S, Ellingsen TE, Brautaset T (2007) The
the N-terminus of the signal peptide of the Escherichia coli presence of N-terminal secretion signal sequences leads to
alkaline phosphatase provides the secretion efficiency and is strong stimulation of the total expression levels of three
involved in the interaction with anionic phospholipids. FEBS tested medically important proteins during high-cell-density
Lett 403:203–207 cultivations of Escherichia coli. Appl Environ Microbiol
Ni Y, Chen R (2009) Extracellular recombinant protein production 73:906–912
from Escherichia coli. Biotechnol Lett 31:1661–1670 Sommer B, Friehs K, Flaschel E (2010) Efficient production of extra-
Orth JD, Conrad TM, Na J, Lerman JA, Nam H, Feist AM, Palsson BO cellular proteins with Escherichia coli by means of optimized
(2011) A comprehensive genome-scale reconstruction of coexpression of bacteriocin release proteins. J Biotechnol
Escherichia coli metabolism—2011. Mol Syst Biol 7:535 145:350–358
Paetzel M, Dalbey RE, Strynadka NC (1998) Crystal structure of a Steiner D, Forrer P, Stumpp MT, Pluckthun A (2006) Signal sequences
bacterial signal peptidase in complex with a beta-lactam inhibitor. directing cotranslational translocation expand the range of pro-
Nature 396:186–190 teins amenable to phage display. Nat Biotech 24:823–831
Palmer T, Berks BC (2012) The twin-arginine translocation (Tat) Suominen I, Meyer P, Tilgmann C, Glumoff T, Glumoff V, Käpylä J,
protein export pathway. Nat Rev Microbiol 10:483–496 Mäntsälä P (1995) Effects of signal peptide mutations on process-
Papanikou E, Karamanou S, Economou A (2007) Bacterial protein ing of Bacillus stearothermophilus α-amylase in Escherichia coli.
secretion through the translocase nanomachine. Nat Rev Microbiology 141:649–654
Microbiol 5:839–851 Takemori D, Yoshino K, Eba C, Nakano H, Iwasaki Y (2012)
Park SL, Kwon MJ, Kim SK, Nam SW (2004) GroEL/ES chaperone Extracellular production of phospholipase A2 from Streptomyces
and low culture temperature synergistically enhanced the soluble violaceoruber by recombinant Escherichia coli. Protein Expr
expression of CGTase in E. coli. J Mol Biol 14:216–219 Purif 81:145–150
Petersen TN, Brunak S, Von Heijne G, Nielsen H (2011) SignalP 4.0: Tan NS, Ho B, Ding JL (2002) Engineering a novel secretion signal for
discriminating signal peptides from transmembrane regions. Nat cross-host recombinant protein expression. Protein Eng 15:337–
Methods 8:785–786 345
Peterson JH, Woolhead CA, Bernstein HD (2003) Basic amino acids in Terpe K (2006) Overview of bacterial expression systems for heterol-
a distinct subset of signal peptides promote interaction with the ogous protein production: from molecular and biochemical fun-
signal recognition particle. J Biol Chem 278:46155–46162 damentals to commercial systems. Appl Microbiol Biotechnol
Phoenix DA, Kusters R, Hikita C, Mizushima S, De Kruijff B (1993) 72:211–222
OmpF-Lpp signal sequence mutants with varying charge hydropho- Tian P, Bernstein HD (2009) Identification of a post-targeting step
bicity ratios provide evidence for a phosphatidylglycerol-signal required for efficient cotranslational translocation of proteins
sequence interaction during protein translocation across the across the Escherichia coli inner membrane. J Biol Chem
Escherichia coli inner membrane. J Biol Chem 268:17069–17073 284:11396–11404
Puziss JW, Harvey RJ, Bassford PJ (1992) Alterations in the hydro- Tomkiewicz D, Nouwen N, Driessen AJM (2008) Kinetics and ener-
philic segment of the maltose-binding protein (MBP) signal pep- getics of the translocation of maltose binding protein folding
tide that affect either export or translation of MBP. J Bacteriol mutants. J Mol Biol 377:83–90
174:6488–6497 Tong L, Lin Q, Wong WKR, Ali A, Lim D, Sung WL, Hew CL, Yang
Qian ZG, Xia XX, Choi JH, Lee SY (2008) Proteome-based identifi- DSC (2000) Extracellular expression, purification, and character-
cation of fusion partner for high-level extracellular production of ization of a winter flounder antifreeze polypeptide from
recombinant proteins in Escherichia coli. Biotechnol Bioeng Escherichia coli. Protein Expr Purif 18:175–181
101:587–601 Uhlen M, Abrahmsen L (1989) Secretion of recombinant proteins into
Ray MVL, Meenan CP, Consalvo AP, Smith CA, Parton DP, Sturmer the culture medium by Escherichia coli and Staphylococcus au-
AM, Shields PP, Mehta NM (2002) Production of salmon calci- reus. Biochem Soc Trans 17:340–341
tonin by direct expression of a glycine-extended precursor in Ulbrandt ND, London E, Oliver DB (1992) Deep penetration of a
Escherichia coli. Protein Expr Purif 26:249–259 portion of Escherichia coli SecA protein into model membranes
Rusch SL, Kendall DA (2007) Interactions that drive Sec-dependent is promoted by anionic phospholipids and by partial unfolding. J
bacterial protein transport. Biochemistry 46:9665–9673 Biol Chem 267:15184–15192
Rusch SL, Chen H, Izard JW, Kendall DA (1994) Signal peptide Umakoshi M, Hirasawa T, Furusawa C, Takenaka Y, Kikuchi
hydrophobicity is finely tailored for function. J Cell Biochem Y, Shimizu H (2011) Improving protein secretion of a
55:209–217 transglutaminase-secreting Corynebacterium glutamicum recom-
Rusch SL, Mascolo CL, Kebir MO, Kendall DA (2002) Juxtaposition binant strain on the basis of 13C metabolic flux analysis. J Biosci
of signal-peptide charge and core region hydrophobicity is critical Bioeng 112:595–601
for functional signal peptides. Arch Microbiol 178:306–310 Valent QA, Scotti PA, High S, De Gier JW, Von Heijne G, Lentzen G,
Saleh MT, Fillon M, Brennan PJ, Belisle JT (2001) Identification of Wintermeyer W, Oudega B, Luirink J (1998) The Escherichia coli
putative exported/secreted proteins in prokaryotic proteomes. SRP and SecB targeting pathways converge at the translocon.
Gene 269:195–204 EMBO J 17:2504–2512
3826 Appl Microbiol Biotechnol (2013) 97:3811–3826

Viklund H, Elofsson A (2008) OCTOPUS: improving topology pre- of heterologous proteins for commercial applications. Recent
diction by two-track ANN-based preference scores and an extend- Patents on Chemical Engineering 5:45–55
ed topological grammar. Bioinformatics 24:1662–1668 Wrede P, Landt O, Klages S, Fatemi A, Hahn U, Schneider G (1998)
Vlasuk GP, Inouye S, Ito H, Itakura K, Inouye M (1983) Effects of the Peptide design aided by neural networks: biological activity of
complete removal of basic amino acid residues from the signal artificial signal peptidase I cleavage sites. Biochemistry 37:3588–
peptide on secretion of lipoprotein in Escherichia coli. J Biol 3593
Chem 258:7141–7148 Yamanaka H, Okamoto K (1996) Amino acid residues in the pro region
Von Heijne G (1984) How signal sequences maintain cleavage speci- of Escherichia coli heat-stable enterotoxin I that affect efficiency
ficity. J Mol Biol 173:243–251 of translocation across the inner membrane. Infect Immun
Wang J, Ho B, Ding J (2001) Functional expression of full length 64:2700–2708
Limulus Factor C in stably transformed Sf9 cells. Biotechnol Lett Yoon SH, Kim SK, Kim JF (2010) Secretory production of
23:71–76 recombinant proteins in Escherichia coli. Recent Pat
Watanabe K, Tsuchida Y, Okibe N, Teramoto H, Suzuki N, Inui M, Biotechnol 4:23–29
Yukawa H (2009) Scanning the Corynebacterium glutamicum R Zalucki YM, Power PM, Jennings MP (2007) Selection for effi-
genome for high-efficiency secretion signal sequences. cient translation initiation biases codon usage at second ami-
Microbiology 155:741–750 no acid position in secretory proteins. Nucleic Acids Res
Wen C, Gan R, Zhu S (2003) Construction of secretory expression 35:5748–5754
system suitable to express glucagon under the control of PL Zalucki YM, Beacham IR, Jennings MP (2009) Biased codon usage in
promoter. Curr Microbiol 47:0180–0185 signal peptides: a role in protein export. Trends Microbiol
Wickner W, Moore K, Dibb N, Geissert D, Rice M (1987) Inhibition of 17:146–150
purified Escherichia coli leader peptidase by the leader (signal) Zalucki YM, Beacham IR, Jennings MP (2011) Coupling between
peptide of bacteriophage M13 procoat. J Bacteriol 169:3821–3822 codon usage, translation and protein export in Escherichia coli.
Wong W-KR, Ali AB, Ma MC (2003) Cloning, expression, and Biotechnol J 6:660–667
characterization of diuretic hormone Manduca diuresin from Zhang H, Li Z, Qian Y, Zhang Q, Du P, Gan R, Ye Q (2007)
Manduca sexta in Escherichia coli. Protein Expr Purif 29:51–57 Cultivation of recombinant Escherichia coli for secretory produc-
Wong WKR, Fu Z, Wang YY, Ng KL, Chan AKN (2012) Engineering tion of human epidermal growth factor under control of PL
of efficient Escherichia coli excretion systems for the production promoter. Enzyme Microb Technol 40:708–715

You might also like