Biotechnology Advances 30 (2012) 1102–1107

Contents lists available at SciVerse ScienceDirect

Biotechnology Advances
journal homepage: www.elsevier.com/locate/biotechadv

Bacterial expression systems for recombinant protein production: E. coli and beyond
Rachel Chen ⁎
School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0100, United States

a r t i c l e i n f o a b s t r a c t

Available online 24 September 2011 Escherichia coli expression system continues to dominate the bacterial expression systems and remain to be
the preferred system for laboratory investigations and initial development in commercial activities or as a
Keywords: useful benchmark for comparison among various expression platforms. Some new developments in over-
Recombinant protein production coming its shortcomings are reviewed in this paper, including antibiotics-free selection plasmids, extracellu-
Bacterial expression system lar production, and posttranslational modifications. The ability for E. coli to make mg glycosylated proteins
E. coli
promises even broader applications of the E. coli system in the future. Significant progresses have also been
Lactoccocus lactis
Pseudomonas expression system
made over the past few years in alternative bacterial expression systems. Notably, the Lactoccocus lactis sys-
Streptomyces expression system tem has proven to be a viable choice for membrane proteins. Additionally, several Pseudomonas systems were
developed and achieved product titers comparable to E. coli systems. Other bacterial systems such as
Streptomyces, coryneform bacteria, and halophilic bacteria offer advantages in some niche areas, providing
more choices of bacterial expression systems for recalcitrant proteins.
© 2011 Elsevier Inc. All rights reserved.

1. Introduction
Recombinant protein production is an enormous field. There seems
no sign that the expansion of this field will abate anytime soon. Impres-
sive progresses in the recombinant protein technology over the past de-
cades have brought hundreds of therapeutic proteins into clinical
applications. As there are hundreds more therapeutic proteins in clini-
cal trials, researches aimed to better the technology will continue to
speed ahead. The advent of systems biology era is an important driving
force that propels the field. The desire to understand the functions of
hundreds and thousands of proteins, whose sequences were just re-
cently made available, demands new approaches that deliver each and
every single protein (including membrane proteins) in quantities and
quality dictated by the structural and biochemical analysis. Moreover,
the inherent biodiversity in both peptide sequences and
post-translational modifications adds significantly to the complexity
of the task to deliver a bioactive protein, because it requires not only
the synthesis of peptide backbone but also its authentic modification.
Many ingenious strategies were developed over the recent years to
meet these challenges. This review hopes to capture some of the newest
development.
Understandably, providing a review for such a broad field is diffi-
cult, even with the confine of prokaryotic expression systems. To keep
⁎ Tel.: + 1 404 894 1255; fax: + 1 404 894 2866.
E-mail address: rchen@chbe.gatech.edu.
the task manageable, this review will cover references from 2006
with only a few exceptions, where continuity demands inclusion of
these older references. Many excellent reviews have been published
over the recent years (Demain and Vaishnav, 2009; Makino et al.,
2011a; Zerbs et al., 2009). In order to make this review a meaningful
contribution to the field, I intend to complement the existing reviews
by focusing on areas that have received less coverage in other recent re-
views. Three bacterial expression systems are reviewed in some details,
Escherichia coli, Lactoccocus lactis, and Pseudomonas. Bacillus system is
not reviewed here as excellent reviews are published recently (Nijland
and Kuipers, 2008; Pohl and Harwood, 2010). Only selected aspects are
reviewed for each chosen system. For example, in E. coli the focus is on
posttranslational modification and trend of developing antibiotic-free
selection systems. On L. lactis, membrane protein expression is empha-
sized. Where possible, a comparison to E. coli system is made. Other im-
portant developments are only briefly mentioned in the review and
recent reviews are referenced.
2. E. coli systems: new development
E. coli expression system continues to dominate the bacterial ex-
pression systems and remain to be the first choice for laboratory in-
vestigations and initial development in commercial activities or as a
useful benchmark for comparison among various expression plat-
forms. E. coli system is also the basis for efforts in protein engineering
and high-throughput structural analysis (Gordon et al., 2008; Koehn
and Hunt, 2009). Reviews also appeared recently on general aspects
of using the E. coli system and more specific topics such as
single-domain antibody fragments (De Marco, 2011; Koehn and

0734-9750/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.biotechadv.2011.09.013
 R. Chen / Biotechnology Advances 30 (2012) 1102–1107
Hunt, 2009), stress and stress responses associated with recombinant
protein production (Sevastsyanovich et al., 2010).
Extracellular production of proteins is highly desirable as it could
greatly reduce the complexity of a bioprocess and improve product
quality. The inability of E. coli in secretion of protein products to growth
medium has long been considered as a major drawback of the system.
However, the dogma that E. coli secretes no protein has been challenged
recently by the recognition of both its natural ability to secrete protein in
common laboratory strains and increased ability to secrete proteins in
engineered cells. Significant efforts have been made in recent years to
address this issue (Ni and Chen, 2009). A survey of recent literature
shows that the strategies fall into four categories, 1) engineering dedi-
cated secretion systems that naturally exist in pathogen E. coli; 2) use
carrier proteins with no known translocation mechanisms; 3) use cell
envelope mutants; and 4) co-expression a lysis-promoting protein
(Kil) (Ni and Chen, 2009). Efforts in enhancing its ability to secrete pro-
teins to growth medium are significant and some successful technology
has begun to impact the production beyond laboratory investigations.
Additional reviews on the topic include the review on using autotran-
sporters (Jong et al., 2010).
While potential problems of using antibiotic and antibiotic-resistance
markers in large scale recombinant protein production have long been
recognized, only recently, efforts in developing alternative methods for
selections have born fruits in practical sense. Additionally, engineering
glycosylation and other post-translational modification into the system
is an exciting development in recent years. These two areas will be
reviewed in the following sections.
2.1. Glycosylation and other post-translational modifications
Just a few years ago, it was widely believed that bacteria were inca-
pable of making glycosylated proteins. The discovery of N-linked glyco-
sylation system in a Gram-negative bacterium Campylobacter jejuni and
subsequent transfer of the system to E. coli (Wacker et al., 2002) opened
up the exciting possibility to use E. coli to synthesize glycoproteins.
Since this seminal work, E. coli was shown to glycosylate the native
PglB (the oligosaccharide transferase from C. Jejuni) substrate, AcrA,
with various O-antigen polysaccharides native to E. coli (Feldman
et al., 2005; Wacker et al., 2006). More recently, attempts were made
to produce glycosylated proteins other than the native PglB substrate
AcrA or other types of glycans non-native to either E. coli or C. jejuni
(Fisher et al., 2011; Ihssen et al., 2010; Lizak et al., 2011; Pandhal
et al., 2011). Ihssen et al. used the system to attach Schigella O antigens
to two carrier proteins derived from C jejuni and Pseudomonas aeruginosa,
respectively. The resulting glycoproteins have the potential to be used as
vaccines against shigellosis, a disease causing 1 million deaths per year in
developing world, mostly of young children (Ihssen et al., 2010). In this
work, the host strain CLM24 carries a chromosomal deletion of the
waaL gene encoding an O polysaccharide ligase. The genes encoding
the necessary components: PglB, carrier proteins, and Shiga O antigen
biosynthesis were put on three compatible plasmids. Successful pro-
duction of glycoproteins was detected by using the anti-shigella O1 an-
tibodies. A ladder of glycoprotein bands were observed for each
glycoprotein, due to polymerization of the repeating O-units by the ac-
tion of the enzymes Wzy and Wzz. This study also identified several
challenges of the glycosylation. First challenge is that glycosylation is
often incomplete. In fact, glycosylation efficiency is sometimes as low
as single digit percentage. When more than one glycosylation sites are
present, heterogeneity is also an issue. Second challenge is the growth
impairment by one or more glycosylation components and conditions
for optimal production of carrier proteins may be different from those
leading to optimal glycosylation, necessitating an exhaustive search
for best conditions. A fed-batch process established after significant opti-
mization led to a yield of 18–24 mg L− 1 and productivity of 0.75–
1.0 mg L− 1 h− 1 (Ihssen et al., 2010). In another work by Lizak et al., anti-
body fragments, a potential class of therapeutic proteins, were shown to
1103
be effectively N-glycosylated with the C. jejuni PglB system, improving
the stability and solubility of the protein (Lizak et al., 2011). Specifically,
the murine single-chain fragment of the anti-His tag antibody 3D5 was
engineered to carry two N-glycosylation sites with the optimal sequence
DQNAT within the flexible linker region connecting the variable light
chain (VL) with the variable heavy chain (VH) (Lizak et al., 2011). Produc-
tion of the diglycosylated 3D5 was successful, although the efficiency
was decreased when scaled up to 5 L. Change of antibiotic resistance
gene from chloramphenicol to kanamycin was found to be helpful to re-
tain pglB-containing plasmid. After column purification, a yield of
2 mg L− 1 diglycosylated protein was obtained. This compared to
8 mg L− 1 naked protein under similar conditions. While retaining the
same affinity to the target antigen, the diglycosylated 3D5 was found to
be more resistant to proteolysis, presumably due to steric hinderance. Ad-
ditionally, the solubility of the protein was increased 2.5 fold (Lizak et al.,
2011). Thus, N-glycosylation could not only be used to produce pro-
teins with the biological activity but also be used as an effective
tool to modify physical property of a protein. Therefore, it seems
that E. coli can be used to produce mg of glycosylated proteins with
the desired glycans.
It was reported that in eukaryotic cells, up to 98% proteins are N-ter-
minally acetylated (Johnson et al., 2010). Additionally, Nε-Acetylation of
lysine is also common. Functional importance of acetylation probably ri-
vals that of phosphorylation (Newmann et al., 2008). Acetylation,
among others, influences DNA replication, repair and recombination,
maintenance of genomic stability, cytoskeletal dynamics, metabolism,
signal transduction, protein folding and trafficking (Newmann et al.,
2008). It is thus important to generate authentic acetylation on the pep-
tide backbone in order to produce bioactive recombinant proteins. In
2005, a fusion strategy for acetylation was reported (Acharya et al.,
2005). In this work, Acharya et al. fused a yeast acetyltransferase with
the fragment of the tumor suppressor p53 protein (residues 320–356)
and found that the lysine residue at 320 of p53 fragment was successful-
ly acetylated (Acharya et al., 2005), suggesting that the yeast acetyl-
transferase is functional in E. coli system. It is not known from this
work, however, whether the fusion is absolutely necessary. In a more
recent study, a different approach was used. Modification of the lysine
residue, acetyllysine, was incorporated into several recombinant pro-
teins as a non-natural amino acid. Accordingly, N ε-acetyllysyl-tRNA
synthetase/tRNACUA pair was generated through protein engineering
(Newmann et al., 2008). An endogenous activity deacetylase-CobB, ap-
parently interfered with this acetylation strategy. Thus it was necessary
to use an inhibitor, nicotinamide, to suppress the deacetylation activity.
As reported, recombinant acetylated manganese superoxide dismutase
was successfully produced with the modification at position 44
(Newmann et al., 2008). For amino terminal acetylation, Johnson et
al. showed that co-expression of yeast NatB acetylation complex with
the substrate proteins is very effective. The complex recognizes three
terminal sequences, Met.Asp, Met.Glu, and Met.Asn. Several proteins
from both yeast and human origins bearing one of the three sequences
were acetylated (Johnson et al., 2010). In another work, Fang et al.
discovered that RimJ is an endogenous acetyltransferase in E. coli, re-
sponsible for the partial acetylation of recombinant peptide, thymosin
(Fang et al., 2009). Upon further optimization in a follow up work,
they showed that a complete acetylation of the N-terminal serine
residue was achieved when RimJ was overexpressed, resulting a
high-yield and fully acetylated product (Ren et al., 2011). It
would be interesting to see how effective the enzyme works with
other substrates varying in sizes as this has only been shown for a
28-residue peptide.
Although not as widely reported, using similar approach as acet-
ylation, phosphorylation of recombinant proteins could also be car-
ried out in E. coli. Murata et al. showed that by co-expressing
human Jun N-terminal kinase and its cognate substrate (human
Jun dimerization protein 2) in E. coli, phosphorylated human Jun di-
merization protein 2 was produced (Murata et al., 2008). Similarly,
1104 R. Chen / Biotechnology Advances 30 (2012) 1102–1107
phosphorylated activation transcription factor 2 was also produced.
In addition, both acetylation and phosphorylation could be achieved
on a single substrate when appropriate enzymes were co-expressed
(Murata et al., 2008).
2.2. Antibiotics-free plasmid systems
Antibiotics are commonly used to exert selective pressure to enhance
stability of plasmid-bearing cultures. However, the high associated cost
and the potential risk of spreading antibiotics resistance genes make
large scale use of antibiotics highly undesirable. This has prompted re-
searchers to develop alternative method to stabilize plasmid-bearing cul-
tures. Isaksson and coworkers (Hagg et al., 2004) developed a
host/plasmid system that could eliminate the use of antibiotics. This
was achieved by deleting from the host chromosome an essential gene,
infA encoding translation initiation factor 1, which was put on the plas-
mid. The maintenance of the plasmid was ensured as plasmid-free cells
would not survive. The authors showed that the vector was stable for at
least 120 generations in the absence of antibiotics and antibiotics resis-
tance genes. Attractive feature of the system is that the IF1 is small, only
71 amino acids. Therefore, it does not significantly increase the vector
size or exert metabolic burdens as a large protein would (Hagg et al.,
2004). A recent paper by Goh and Good describes an alternative strategy
(Goh and Liam, 2008). Here, the essential gene, fabI (encoding enoyl ACP
reductase), was plasmid-borne. Its overexpression relieves the toxic effect
of a biocide, triclosan, thus establishing the dependence of the culture on
the presence of the plasmid. The system was shown to achieve higher
plasmid stability than an antibiotic-dependent system (parental sys-
tem), 99.5% versus 85.3%. However, as authors themselves pointed
out, although triclosan is currently approved for household use and
over-the-counter medicines, resistance did emerge, which could be a
potential problem. Using a different strategy, a plasmid stabilizing sys-
tem based on toxin and anti-toxin was developed (Peubez et al.,
2010). In this case, the toxin gene (ccdB) was integrated to the host
chromosome. The production of the toxin, a stable protein of 100
amino acids, causes cell death as the toxin binds to gyrase and impairs
DNA replication. The action of toxin is antagonized by the plasmid-
borne ccdA encoded by the anti-toxin gene. Thus, segregation of plas-
mid would lead to cell death. The system was used to produce two
recombinant proteins under antibiotics free condition. In the first case
of producing a recombinant Pseudomonas exotoxin, the antibiotic-free
system was found not only to offer better plasmid stability than the
one relying on antibiotics, it also gave higher product titer
(603 mg L − 1 versus 280 mg L − 1) (Peubez et al., 2010). The comparison
between the new stabilizing system and the antibiotic dependent sys-
tem on a second recombinant protein production, the AlpA protein
from Helicobacter pylori, showed that the antibiotic-free system was
slightly better than the other system with about 16.5% of total cellular
protein being the recombinant target product (Peubez et al., 2010).
Using a different approach, Steinbuchel and coworkers developed two
anabolism-based plasmid addiction system and showed superior stabil-
ity over the antibiotic-dependent systems. (Kroll et al., 2009, 2010,
2011). The first system is based on isopentenyl pyrophosphate (IPP) syn-
thesis pathway. The host E. coli strain was modified by deleting the essen-
tial gene, ispH (encoding 4-hydroxy-3-methylbut-2-enyl diphosphate
reductase), of the deoxyxylulose 5-phosphate (DXP) pathway of IPP path-
way. This modification makes the IPP synthesis in the host strain depend
on the episomal mevalonate (MVA) pathway, synthetically constituted
using genes from Lactococcus lactic and Staphylococcus aureus (Kroll
et al., 2009). The genes for target products (in this case, cyanophycin)
were also on the same plasmid. The authors showed 100% plasmid stabil-
ity, which led to increased production of cyanophycin. Cross-feeding be-
tween plasmid-free cells and plasmid-bearing cells was prevented
because IPP, as a phosphorylated metabolite, would not be readily upta-
ken by the cells, including those having lost the plasmid (Kroll et al.,
2010). This is, in fact, a distinct advantage over the method in which
amino acid auxotroph is used. In principle, any essential genes/pathways
could be exploited to create plasmid-addiction systems. Synthesis of ly-
sine precursor, L, L 2, 6-diaminopipelate, was chosen in developing anoth-
er plasmid-addiction system. In this case, the essential gene, encoding the
last step in the formation of L, L 2, 6-diaminopipelate, was knocked out
from the host. An alternative L, L 2, 6-diaminopipelate synthesis pathway
using (Kroll et al., 2011) aminotransferase pathway was encoded on the
plasmid, along with other genes necessarily for the target synthesis (cya-
nophycin). This system was successfully scaled up to 400 L, illustrating
the effectiveness of this approach.
3. Lactoccocus lactis
Lactoccocus lactis has emerged in recent years as one of the most
important Gram-positive bacterial expression systems (Morello
et al., 2008). Compared to E. coli, it has the advantage of being GRAS
and endotoxin free, important for both therapeutic and food applica-
tions (Yeh et al., 2009). As a Gram-positive, the potential for extracel-
lular production is a big plus. More recently, its success in membrane
protein expression establishes this important niche for this bacterial
system.
3.1. Expression systems and recent improvements
NICE (nisin-inducible controlled gene expression) is the most widely
used inducible system in this microorganism(Morello et al., 2008). Be-
sides the nisin-inducible promoter, NICE includes regulatory elements
from the nis operon native to L. lactis strains. This system affords tightly
controlled expression and relatively high protein expression (Liang
et al., 2006; Zhou et al., 2006). Up to 300 mg L− 1 was reported (Morello
et al., 2008). Alternative inducible expression systems were also devel-
oped in recent years. Siren and coworkers developed a phosphate star-
vation inducible system (Sirén et al., 2008). Using beta-galactosidase as
model protein, they were able to show that the level of expression
using the phosphate system was comparable to the NICE system. Another
promoter, P170, was also shown effective. The P170 gene expression sys-
tem has an advantage as it does not require inducer and is turned on
when pH falls below six during transition from exponential to stationary
phase (Morello et al., 2008; Sriraman and Jayaraman, 2006). Product titer
of 300 mg L− 1 was reported for recombinant Staphylococcal nuclease
(Nuc) and a pilot scale process using the P170 system gave similar yield
to the NICE system, suggesting that it is a viable alternative to NICE. Addi-
tionally, a novel expression system based on a Zinc dependent promoter
was developed with product titers similar to those obtained with NICE.
This system is triggered by addition of EDTA, which chelates the Zn ion
thereby reducing its concentration below the threshold (Morello et al.,
2008). Besides using strong promoters, stabilizing mRNA by inclu-
sion of 5′-untranslated leader sequence (UTLS) was shown to in-
crease recombinant amylase production in L. lactis by several folds
(Narita et al., 2006). Overall, however, product titers have not
exceeded 300 mg L − 1, which is approximately one-order of magni-
tude lower than that of E. coli.
Significant efforts were directed to develop food-grade expression
systems by eliminating antibiotic selection markers (Yeh et al., 2009).
Strategies vary but sharing the common feature by creating dependence
of host survival on a plasmid-born function, such as Nisin resistance
(Luo and Wang, 2009), lactose metabolism (Maischberger et al., 2010;
Xu et al., 2007), vitamin (thymidine) or amino acid (alanine, threonine)
provision.
Another area for improvement is secretion. Secretion using Usp45
signal peptide generally gave satisfactory efficiency. For example,
using Usp45 signal peptide, a Bacillus nattokinase was almost
completely secreted to the medium (Liang et al., 2006). However,
poor secretion with some large proteins such as murine cytokine
interleukin-12 (mIL-12, 70 kDa) was observed. To address the prob-
lem, Fernandez and coworkers (Fernandez et al., 2009) developed a
 R. Chen / Biotechnology Advances 30 (2012) 1102–1107
novel signal peptide SLPmod based on Lactobaccillus S-layer proteins.
Compared to the control which utilizes Usp45 signal peptide, the new
signal peptide allowed much higher secretion of mIL-12, 185 pg/ml
versus 40 pg/ml (Fernandez et al., 2009).
3.2. Membrane protein expression
Membrane proteins represent 30% of the proteome of an organism
(Bill et al., 2011), yet much of their functions are unknown due to lack
of structure-functional studies. The first hurdle for structure study of a
membrane protein is its availability in sufficient quantity. This is cur-
rently addressed by recombinant protein expression (Bill et al., 2011).
Yet low level of expression is often encountered. The unique challenges
of achieving high level expression of a membrane protein in a foreign
host is due mainly to its possible toxicity, space-confinement in a mem-
branous environment, and lack of accessory proteins for proper folding.
E. coli and L. lactis are two prokaryotic systems commonly used for
membrane protein production (Gordon et al., 2008). One study com-
paring the two systems for production of 37 prokaryotic membrane
proteins concluded that E. coli was a better system (Surade et al.,
2006). However, Lactoccocal system is advantageous in having only
one membrane, which potentially makes membrane-targeting and pro-
cessing easier and less prone to inclusion body formation. Additionally,
purification of membrane proteins from a single-membrane bacterium
is less laborious. Kunji and coworkers (Kunji et al., 2003) were the first
to use L. lactis expression system for eukaryotic membrane proteins. In
their 2003 paper, they showed that several eukaryotic membrane pro-
teins, including a human KDEL receptor protein, could be overproduced
using the Lactoccocal system without eliciting inclusion bodies (Kunji
et al., 2003). Along with other prokaryotic membrane proteins, these
proteins represented different sizes, topology, quaternary structure
and/or domain organizations, in each case, it was observed the overex-
pressed protein was inserted to the membrane and assembled into a
functional complex (Kunji et al., 2003), thus convincingly showed that
the Lactoccocal system as a promising host for membrane protein pro-
duction. The expression levels vary widely with the target proteins,
ranging from 0.1% to 30% of the total membrane proteins or several mi-
crograms to a few milligrams (Kunji et al., 2003). In most cases, the ex-
pression levels were sufficient for structural analysis (Berntsson et al.,
2009; Niu et al., 2008). In a recent study, the Lactoccocal system was
tested for six plant membrane proteins, the level of expression
was shown to be sufficient for biochemical analyses (Freket-Barrand
and Boutigny, 2010). The success of the expression of these plant
proteins underscores the advantage of the Lactoccocal systems as
previous attempts of expressing them using E. coli, insect, and yeast
expression systems failed to deliver functional proteins at a useful
level (Freket-Barrand and Boutigny, 2010).
Interestingly, green fluorescent protein was found to be a good indica-
tor of protein folding when fused to the N-terminal of a target membrane
protein (Geertsma et al., 2008). This was conveniently used to optimize
expression conditions for membrane proteins (Niu et al., 2008). As a re-
sult, the target protein, an elongase, was expressed to about 15% of the
total membrane protein. Studies aimed to understand the bottlenecks
and stresses imposed on host strains during recombinant membrane pro-
teins expression showed that Lactoccocal responds to the stress differently
than E. coli. While the study itself does not address the problem, the infor-
mation uncovered from the study is extremely valuable (Steen et al.,
2010). In another study, using proteomic approach, Marreddy et al.
(2010) showed that the availability of branched chain amino acids is a
critical factor of the overexpression of proteins in L. lactis, suggesting
that co-expression of selected amino acid transporters may be beneficial
for membrane proteins and other proteins. Another approach targeting
on improving host strain via mutation-selection was also quite successful
(Linares et al., 2010) in obtaining a strain with enhanced production of
membrane proteins (up to 8 fold).
1105
Taken together, we can expect that these new understanding will be
translated to improved production of membrane proteins in L. lactis in
the future.
4. Pseudomonas expression systems
Pseudomonas species are known for their rapid growth and ability
to secrete proteins (Krzeslak et al., 2009). Several Pseudomonas spe-
cies, including P. fluorescens, P. aeruginosa, and P. putida, were
shown to be good alternatives to E. coli in recombinant protein pro-
duction, achieving yield comparable or higher than E. coli systems.
4.1. Pseudomonas fluorescens
Dow has developed a proprietary P. fluorescens expression system
for recombinant protein productions (Huang et al., 2007). There are
several notable advantages of the system, as compared to E. coli system.
While it shares with E. coli for the ability of hyper production of recom-
binant proteins (N 50% of total protein) and the ability to grown to high
cell density (N100 g/L), the production of proteins from P. fluorescens is
less dependent on oxygen concentrations and no acetate accumulation
during production (Huang et al., 2007). Therefore, the fermentation pro-
cess for P. fluorescens may not require strict control of aeration and/or glu-
cose concentration. Product yield as high as 25 g L− 1 was reported for
recombinant nitrilase from the P. fluorescens system. For at least some
proteins tested, the advantage of the system over E. coli was significant.
For example, the P. fluorescens system gave a titer of 3–4 g L− 1 for the
two insecticidal proteins originated from Bacillus thuringiensis, this com-
pares to about 100 mg L− 1 when expressed in E. coli (Huang et al.,
2007). For protein targets that contain disulfide bonds, periplasmic ex-
pression is necessary as its oxidative environment is more conducive for
the proper formation of the disulfide bond. Identifying signal peptides
that direct recombinant protein to periplasm via known translocation
mechanisms is very important to ensure high periplasmic expression
and complete processing (cleavage) of the signal peptides. Using E. coli
PhoA as reporter and a transposon mutant library, several signal peptides
were identified in P. fluorescens, allowing periplasmic expression of a
range of recombinant proteins. Remarkably, production yield was very
high with highest of 10 g L− 1 reported. But the yield varied widely with
different combination of target proteins and signal peptides used. The
cleavage of peptide was not uniform, either (Retallack et al., 2007).
Since underlying mechanism for the variation is not completely under-
stood, optimization through trials and errors is inevitable. However, the
availability of these signal peptides makes it possible to search an optimal
signal peptide for a given target. NeupogenR (fligrastim), marketed by
Amgen, is a recombinant human G-CSF protein expressed as inclusion
bodies in E. coli. Interest of developing biosimilar to NeupogenR is growing
as the patent in Europe has expired (Jin et al., 2011). With the goal of sol-
uble expression in periplamic space, signal peptides identified from previ-
ous study (Retallack et al., 2007) were tested. After considerable
optimization using the 96-well platform, the best yield of 350 mg/L was
achieved at the 1 L fermentation. The cost saving from soluble expressing
is P. fluorescens was speculated to be 50%, compared to extraction of E. coli
inclusion body (Jin et al., 2011), but no rigorous analysis was provided to
support the claim.
4.2. Pseudomonas aeruginosa
P. aeruginosa is of interest as host for recombinant protein produc-
tion due to its native and well-studied secretion systems such as type
II and type III (Derouazi et al., 2008; Krzeslak et al., 2009). In a study
aimed for extracellular production from P. aeruginosa via its type III
system, a novel expression system was developed, in which a
54-amino acid signal peptide from a type III substrate exotoxin S
(ExoS) was fused to the target proteins and the expression of a com-
mon TTSS regulator of P. aeruginosa was under the control of the
1106 R. Chen / Biotechnology Advances 30 (2012) 1102–1107
inducible promoter Ptac. As a result, the plasmid allows, upon IPTG in-
duction, activation of the entire TTSS. Target proteins were shown to
be secreted under calcium deprivation condition generated by addi-
tion of EGTA to the culture medium (Derouazi et al., 2008). Directing
E. coli penicillin G acylase through the type II system of P. aeruginosa
was also attempted but the results were less impressive, due partially
to the requirement of multi-subunits of the target protein and the re-
quirement of maturation. As a result, although some activity was
found extracellularly, cell lysis also occurred, which makes interpre-
tation less conclusive (Krzeslak et al., 2009). Overexpression and se-
cretion of a homologous elastase seemed to be more successful. All
the recombinant elastolytic activities were detected in the culture
medium. Presumably secretion was via type II system (Wong et al.,
2010).
4.3. Pseudomonas putida
P. putida was also used as host for recombinant protein produc-
tion. P. putida KT2440 was certified as a biosafety strain and recent
development of genetic manipulation tools make it possible for engi-
neering recombinant protein production (Dammeyer et al., 2011). Of
particular interest is single chain Fv fragments (scFv). The expression
vector was based on RK2 broad host range plasmid, either Ptac or Pm
promoter from the xyl operon, both inducible, apparently worked
very well in this strain. The vector has the G10L ribosome binding
site with epsilon enhancer and the Erwinia carotovora pelB leader se-
quence to direct the recombinant protein to its periplasm (Dammeyer
et al., 2011). The three model scFvs (murine anti-hen egg-white lyso-
zyme scFv D1.3, and two human scFvs against C-reactive protein,
TOB5-D4, and HT186-D11 against mucin1 as a biomarker for cancer)
were expressed using the vector with multiple affinity tags to facili-
tate their purifications. Single step purification using Strep-Tactin
Superflow resin in gravity flow columns gave a purity of 95%.The
yield of the three scFvs averaged at 3.5 mg L − 1 (Dammeyer et al.,
2011). This product titer is similar to that of the E. coli expression sys-
tem (1–4 mg L − 1), when both are subjected to significant optimiza-
tion(Makino et al., 2011b).
5. Other bacterial expression systems
Several other bacterial expression systems have emerged in recent
years as useful alternatives. Among them, Streptomyces systems are
particularly noteworthy. As a Gram-positive bacterium, Streptomyces
are known to secrete large amounts of soluble proteins to growth me-
dium, suggesting high capacity of their secretion systems. This is in-
deed one of most important attributes that attract many researchers
over the past decade. A few species are preferred hosts, especially
Streptomyces lividans, due to lack of restriction systems and low en-
dogenous proteolytic activities. This host was used to produce
human granulocyte macrophage-colony stimulating factor to support
clinical trials and many other recombinant proteins (Vrancken and
Anne, 2009). A recent paper by Noda et al. (2010) reported successful
use of new vector for expression of three recombinant proteins to
yield of 64, 114, and 230 mg/ml, respectively. The products could be
detected in growth medium using SDS-PAGE, each as a prominent
band. Further, they showed that the products could be isolated easily
and purified to high purity (Noda et al., 2010). The range of product
titers vary several orders of magnitude with highest of 300 mg L − 1
reported. While variation of product titers with target proteins is
not a unique issue for this system, there are a few inherent bottle-
necks that may be addressed. An excellent review by Vrancken and
Anne (2009) provides an insightful analysis of potential areas for im-
provement. These include Identifying strong inducible promoters to
enhance production yield and separate growth from production
phases, and improving on the translation rate by addressing rare
codon issues. Additionally, the choice of signal peptide, which directs
the target protein to either Sec or Tat secretion pathway and deter-
mines which peptidase for cleavage, is crucial. Finally, morphological
engineering represents a unique opportunity to further enhance re-
combinant protein production in Streptomyces (Vrancken and Anne,
2009).
Coryneform bacteria such as Corynebacterium glutamicum and
Corynebacterium ammoniagenes are also potential hosts for recombi-
nant protein production. Itaya and coworkers recently demonstrat-
ed that many coryneform bacteria strains were able to secrete a
recombinant pro-tranglutaminase to a high level, presumably
through the Tat pathway (Itaya and Kitochi, 2008). One strain,
C. ammoniagenes ATCC6872, gave a titer of 2.5 g L − 1, significantly
higher than E. coli and Streptomyces expression systems (Zhang
et al., 2009), demonstrating the potential of coryneform bacteria in
recombinant enzyme production and secretion.
Expression technologies based on halophilic bacteria and halophilic
proteins were also developed to obtain soluble recombinant proteins in-
cluding mammalian proteins (Tokunaga et al., 2010). Moderate halo-
philic bacteria contain osmolytes in their cytosol, which could be a
better environment for mammalian proteins. One success example is
human brain serine racemase (Nagayoshi et al., 2009). A yield of
250 μg/L was obtained. Additionally, proteins from moderate halophiles
are rich in acidic amino acids, which contribute to their high solubility in
both the native and unfolded states. These proteins could be used as sol-
ubilizing fusion partners (Tokunaga et al., 2010).
6. Future prospective
The wide variety of applications of recombinant proteins promises
their need will only increase in the future. The challenges to meet the
growing demand are multifaceted, in terms of quantities, qualities,
and cost-effectiveness. In some cases, delivering a large number of
proteins in parallel is of interest, thus a need for high-throughput ex-
pression technology. Additionally, systems level understanding de-
mands a comprehensive coverage of all proteins in a proteome.
Therefore, expression technology will have to tackle those “difficult
to express” or recalcitrant proteins. This challenge may be com-
pounded by novel non-natural sequences generated by protein engi-
neering or other in vitro technologies. These multifaceted challenges
will not be met with a single technology. We can therefore expect a
future in which more unconventional bacterial expression systems
will find their use along with the improved veteran systems such as
E. coli.
Acknowledgement
Researches related to recombinant protein expression and secretion
in the author's lab were supported by NSF and Georgia Research
Alliance.
References
Acharya A, Xu XJ, Husain-Ponnampalam RD, Hoffmann-Benning S, Kuo MH. Production
of constitutively acetylated recombinant p53 from yeast and Escherchia coli by
tethered catalysis. Protein Expr Purif 2005;41:417–25.
Berntsson RPA, Alia Oktaviani N, Fusetti F, Thunnissen AMWH, Poolman B, Slotboom
DJ. Selenomethionine incorporation in proteins expressed in Lactococcus lactis.
Protein Science 2009;18:1121–7.
Bill RM, Henderson PJF, Iwata S, Kunji ERS, Michel H, Neutze R, et al. Overcoming barriers
to membrane protein sturcture determination. Nat Biotechnol 2011;29:335–40.
Dammeyer T, Steinwand M, Krüger SC, Dübel S, Hust M, Timmis K. Efficient production
of soluble recombiant single chain Fv fragments by a Pseudomanas putida KT2440
cell factory. Microb Cell Fact 2011;10.
De Marco A. Biotechnological applications of recombinant single-domain antibody
fragments. Microb Cell Fact 2011;10.
Demain AL, Vaishnav P. Production of recombinant proteins by microbes and higher
organisms. Biotechnol Adv 2009;27:297–306.
Derouazi M, Toussaint B, Quenee L, Epaulard O, Guillaume M, Marlu R, et al. High-yield
production of secreted active proteins by the Pseudomonas aeruginosa type III se-
cretion system. Appl Environ Microbiol 2008;74:3601–4.
 R. Chen / Biotechnology Advances 30 (2012) 1102–1107
Fang H, Zhang X, Shen L, Si X, Ren Y, Dai H, et al. RimJ is responsible for N(alpha)-acetylation
of thymosin alpha1 in Escherichia coli. Appl Microbiol Biotechnol 2009;84:99-104.
Feldman MF, Wacker M, Hernandez M, Hitchen PG, Marolda CL, Kowarik M, et al. Engineer-
ing N-linked protein glycosylation with diverse O antigen lipopolysaccharide structures
in Escherichia coli. Proc Natl Acad Sci U S A 2005;102:3016–21.
Fernandez A, Horn N, Wegmann U, Nicoletti C, Gasson MJ, Narbad A. Enhanced secre-
tion of biologically active murine interleukin-12 by Lactococcus lactis. Appl Environ
Microbiol 2009;75:869–71.
Fisher AC, Haitjema CH, Guarino C, Celik E, Endicott CE, Reading CA, et al. Production of
secretory and extracellular N-Linked glycoproteins in Escherichia coli. Appl Environ
Microbiol 2011;77:871–81.
Freket-Barrand A, Boutigny S. Lactococcus lactis, an alternative system for functional
expression of peripheral and intrinsic arabidopsis membrane proteins. PLoS One
2010;5.
Geertsma ER, Groeneveld M, Slotboom DJ, Poolman B. Quality control of overexpressed
membrane proteins. PNAS 2008;105:5722–7.
Goh S, Liam G. Plasmid selection in Escherichia coli using an endogenous essential gene
marker. BMC Biotechnol 2008;8.
Gordon E, Horsefield R, Swarts HGP, De Pont JJHHM, Neutze R, Snijder A. Effective
high-throughput overproduction of membrane proteins in Escherichia coli. Protein
Expr Purif 2008;62:1–8.
Hagg P, de Pohl J, Abdulkarim F, Isaksson LA. A host/plasmid system that is not depen-
dent on antibiotics and antibiotic resistance genes for stable plasmid maintenance
in Escherichia coli. J Biotechnol 2004;111:17–30.
Huang K, Badger M, Haney K, Evans S. Large scale production of Bacillus thuringiensis
PS149B1 insecticidal proteins Cry34Ab1 and Cry35Ab1 from Pseudomonas fluorescens.
Protein Expr Purif 2007;53:325–30.
Ihssen J, Kowarik M, Dilettoso S, Tanner C, Wacker M, Thoxy-Meyer L. Production of
glycoprotein vaccines in Escherichia coli. Microb Cell Fact 2010;9.
Itaya H, Kitochi Y. Secretion of Streptomyces mobaraensis pro-transglutaminase by
coryneform bacteria. Appl Microbiol Biotechnol 2008;78:621–5.
Jin H, Cantin GT, Maki S, Chew LC, Resnick SM, Ngai J, et al. Soluble periplasmic produc-
tion of human granulocyte cology-stimulating factor (G-CSF) in Pseudomonas
fluorescens. Protein Expr Purif 2011;78:69–77.
Johnson M, Coulton AT, Geeves MA, Mulvihill DP. Targeted amino-terminal acetylation
of recombinant proteins in E. coli. PLoS One 2010;5:1–5.
Jong WSP, Saun A, Luirink J. Extracellular production of recombinant proteins using
bacterial autotransporters. Curr Opin Biotechnol 2010;21:642–52.
Koehn J, Hunt I. High-throughput protein production (HTPP): a review of enabling
technologies to expedite protein production. Methods Mol Biol 2009;498:1-18. [L
High Throughput Protein Expression and Purification].
Kroll J, Steinle A, Reichelt R, Ewering C, Steinbuchel A. Establishment of a novel
anabolism-based addition system with an artificially introduced mevalonate path-
way: complete stabilization of plasmids as universal application in white biotechnol-
ogy. Metab Eng 2009;11:168–77.
Kroll J, Klinter S, Schneider C, Vob I, Steinbuchel A. Plasmid addition systems: perspec-
tives and applications in biotechnology. Microb Biotechnol 2010;3:634–57.
Kroll J, Klinter S, Steinbuchel A. A novel plasmid addiction system for large scale pro-
duction of cyanophycin in Escherichia coli using mineral salts medium. Appl Micro-
biol Biotechnol 2011;89.
Krzeslak J, Braun P, Voulhoux R, Cool R, Quax W. Heterologous production of Eschericia
coli penicillin G acylase in Pseudomanas aeruginosa. J Biotechnol 2009;142:250–8.
Kunji ERS, Slotboom D-J, Poolman B. Lactococcus lactis as host for overproduction of
functional membrane proteins. Biochim Biophys Acta 2003;1610:97-108.
Liang X, Zhang L, Zhong J, Huan L. Secretory expression of a heterologous nattokinase in
Lactococcus lactis. Appl Microbiol Biotechnol 2006;24:285–95.
Linares DM, Geertsma ER, Poolman B. Evoled Lactococcus lactix strains for enhanced
expression of recombinant membrane proteins. J Mol Biol 2010;401:45–55.
Lizak C, Fan Y, Web TC, Aebi M. N-Linked glycosylation of antibody fragments in Escherichia
coli. Bioconjug Chem 2011;22:488–96.
Luo L, Wang C. Cloning of Nisin resistent gene nisI from Lactococcus lactis and its appli-
cation as a food grade selection marker. Wei Sheng Wu Xue Bao 2009;49:1229–33.
Maischberger T, Mierau I, Peterbauer CK, Hugenholtz J, Haltrich D. High-level expres-
sion of Lactobacillus beta-galactosidases in Lactococcus lactis using the
food-grade, nisin-controlled expression system NICE. J Agric Food Chem 2010;58:
2279–87.
Makino T, Skretas G, Georgiou G. Strain engineering for improved expression of recom-
binant proteins in bacteria. Microb Cell Fact 2011a;10.
Makino T, Skretas G, Kang T, Georgiou G. Comprehensive engineering of Escherichia coli
for enhanced expression of IgG antibodies. Metab Eng 2011b;13:241–51.
Marreddy RK, Geertsma ER, Permentier HP, Pinto JP, Kok J, Poolman B. Amino acid accumu-
lation limits the overexpression of proteins in Lactococcus lactis. PLoS One 2010;26.
Morello E, Bermudez-Humaran LG, Llull D, Sole V, Miraglio N, P. L, et al. Lactococcus lactis,
an efficient cell factory for recombinant protein production and secretion. J Mol
Microbiol 2008;14:48–58.
Murata T, Shinozuka Y, Obata Y, Yokoyama KK. Phosphorylation of two eukaryotic
transcription fcators Jun dimerization protein 2 and activation transcription factor
2 in Escherichia coli by Jun N-terminal kinasse 1. Anal Biochem 2008;376.
1107
Nagayoshi C, Ishibaxhi M, Tokunaga M. Purification and characterization of human
brain serine racemase expressin moderately halophilic bacteria. Protein Pept Lett
2009;16:201–6.
Narita J, Ishida S, Okano K, Kimura S, Fukuda H, Kondo A. Improvement of protein pro-
duction in lactic acid bacteria using 5′ untraslated leader sequence of slpa from
Lactobacillus acidophilus. Appl Microbiol Biotechnol 2006;73:366–73.
Newmann H, Peak-Chew SY, J.W C. Genetically encoding Ne-acetylllysine in recombinant
proteins. Nat Chem Biol 2008;4:232–4.
Ni Y, Chen R. Extracellular recombinant protein production from Escherichia coli. Biotechnol
Lett 2009;31:1661–70.
Nijland R, Kuipers OP. Optimization of protein secretion by Bacillus subtilis. Recent Pat
Biotechnol 2008;2:79–87.
Niu Y, Kong J, Xu Y. A novel GFP-fusion eukaryotic membrane protein expression sys-
tem in Lactococcus lactis and its application to overexpression of an elongase. Curr
Microbiol 2008;57:423–8.
Noda S, Ito Y, Shimizu N, Tanaka T, Ogino C, Kondo A. Over-production of various
secretory-form proteins in Streptomyces lividans. Protein Expr Purif 2010;73:
198–202.
Pandhal H, Ow SY, Noirel J, Wright PC. Improving N-glycosylation efficiency in Escherichia
coli using shotgun proteomics metabolic network analysis, and selective reaction
monitoring. Biotechnol Bioeng 2011;108:902–12.
Peubez I, Chaudet N, Mignon C, Hild G, Husson S, Courtois V, et al. Antibiotic-free selec-
tion in E. coli: new considerations for optimal design and improved production.
Microb Cell Fact 2010;9.
Pohl S, Harwood CR. Heterologous protein secretion by Bacillus species from the cradel
to the grave. Adv Appl Microbiol 2010;73:1-25.
Ren Y, Yao X, Dai H, Li S, Fang H, Chen H, et al. Production of N-acetylated thymosin
alpha 1 in Escherichia coli. Microb Cell Fact 2011;10.
Retallack DM, Schneider JC, Mitchell J, Chew L, Liu H. Transport of heterologous pro-
teins to the periplamic space of Pseudomonas fluorescens using a variety of native
signal sequences. Biotechnol Lett 2007;29:1483–91.
Sevastsyanovich YR, Alfasi SN, Cole JA. Sense and nonsense from a systems biology ap-
proach to microbial recombinant protein production. Biotechnol Appl Biochem
2010;55:9-28.
Sirén N, Salonen K, Leisola M, Nyyssölä A. A new and efficient phosphate starvation in-
ducible expression system for Lactococcus lactis. Appl Microbiol Biotechnol
2008;79:803–10.
Sriraman K, Jayaraman G. Enhancement of recombinant streptokinase production in
Lactococcus lactis by suppression of acid tolerance response. Appl Microbiol Bio-
technol 2006;72:1202–9.
Steen A, Wiederhold E, Gandhi T, Breitling R, Slotboom DJ. Physiological adaptation of
the bacterium Lactococcus lactis in response to the production human CFTR. Cell
Mol Proteomics 2010;10. [page number not found].
Surade S, Klein M, Stolt-Bergner PC, Muenke C, Roy A, Michel H. Comparative analysis
and “expression space” coverage of the production of prokaryotic membrane pro-
teins for structural genomics. Protein Sci 2006;15:2178–89.
Tokunaga H, Arakawa T, Tokunaga M. Novel soluble expression technologies derived
from unique properties of halophilic proteins. Appl Microbiol Biotechnol
2010;88:1223–31.
Vrancken K, Anne J. Secretory production of recombinant proteins by Streptomyces.
Future Microbiol 2009;4:181–8.
Wacker M, Linton D, Hitchen PG, Nita-Lazar M, Haslam SM, North SJ, et al. N-linked glycosyl-
ation in Campylobacter jejuni and its functional transfer into E. coli. Science 2002;298:
1790–3.
Wacker M, Feldman MF, Callewaert N, Kowarik M, Clarke BR, Pohl NL, et al. Substrate
specificity of bacterial oligosaccharyltransferase suggests a common transfer
mechanism for the bacterial and eukaryotic systems. Proc Natl Acad Sci U S A
2006;103:7088–93.
Wong CF, Salleh AB, Basri M, Rahman R. Organic-solvent stability of elastase strain K
overexpressed in an Escherichia–Pseudomonas expression system. Biotechnol
Appl Biochem 2010;57:1–7.
Xu B, Cao YS, Chen Y, Guo XH. Construction of food-grade inducible expression system
in Lactococcus lactis and expression of fusion OprF/H from Pseudomonas aeruginosa.
Wei Sheng Wu Xue Bao 2007;47:604–9.
Yeh C-M, Kao B-Y, Peng H-J. Production of a recombinant type 1 antifreeze protein
anlogue by L. lactis and its applications on frozen meat and frozen dough. J Agric
Food Chem 2009;57:6216–23.
Zerbs S, Frank AM, Collart FR. Bacterial systems for production of heterologous pro-
teins. Methods Enzymol 2009;463:149–67.
Zhang D, Zhu Y, Chen J. Microbial tranglutaminase production: understanding the
mechanism. Biotechnol Genet Eng Rev 2009;26:205–22.
Zhou XX, Li WF, Ma GX, Pan YJ. The nisin-controlled gene expression system: construc-
tion, application and improvements. Biotechnol Adv 2006;24:285–95.