Recent advances in the production of proteins in insect and mammalian cells
for structural biology
Joanne E. Nettleship, René Assenberg1, Jonathan M. Diprose, Nahid Rahman-Huq, Raymond J. Owens*
The Oxford Protein Production Facility and Division of Structural Biology, Henry Wellcome Building for Genomic Medicine, University of Oxford, Roosevelt Drive, Oxford OX3 7BN, UK
a r t i c l ei n f o
Received 23 December 2009
Received in revised form 4 February 2010
Accepted 7 February 2010
Available online 11 February 2010
a b s t r a c t
The production of proteins in sufficient quantity and of appropriate quality is an essential pre-requisite
for structural studies. Escherichia coli remains the dominant expression system in structural biology with
nearly 90% of the structures in the Protein Data Bank (PDB) derived from proteins produced in this bac-
terial host. However, many mammalian and eukaryotic viral proteins require post-translation modifica-
tion for proper folding and/or are part of large multimeric complexes. Therefore expression in higher
eukaryotic cell lines from both invertebrate and vertebrate is required to produce these proteins.
Although these systems are generally more time-consuming and expensive to use than bacteria, there
have been improvements in technology that have streamlined the processes involved. For example, the
use of multi-host vectors, i.e., containing promoters for not only E. coli but also mammalian and baculo-
virus expression in insect cells, enables target genes to be evaluated in both bacterial and higher eukary-
otic hosts from a single vector. Culturing cells in micro-plate format allows screening of large numbers of
vectors in parallel and is amenable to automation. The development of large-scale transient expression in
mammalian cells offers a way of rapidly producing proteins with relatively high throughput. Strategies
for selenomethionine-labelling (important for obtaining phase information in crystallography) and con-
trolling glycosylation (important for reducing the chemical heterogeneity of glycoproteins) have also
been reported for higher eukaryotic cell expression systems.
? 2010 Elsevier Inc. All rights reserved.
Over the last 10 years there have been major advances in the
technology for recombinant protein production specifically for
structural biology. Much of this has been led by structural genom-
ics centres which have pioneered high throughput approaches to
sample preparation including the use of laboratory automation to
achieve parallel processing. Escherichia coli remains the dominant
host for producing recombinant proteins as shown by an analysis
of expression systems used for structures deposited in the Protein
Data Bank (PDB) (Table 1). Thus 88% of the protein chains from
structures for which expression system annotation is available
were produced in E. coli compared to 9% for all eukaryotic hosts
combined. Of these, baculovirus/insect cells accounted for 4%,
whilst mammalian cells, including both stable cell lines and more
recently transient expression, represented 2.4%. Although multi-
construct approaches have increased the success rate for obtaining
soluble protein from E. coli (Fogg et al., 2006; Graslund et al., 2008),
there are proteins which are not amenable to bacterial expression,
e.g., many membrane proteins, multi-protein complexes and cell
surface or secreted glycoproteins (Aricescu et al., 2006b). It is esti-
mated that as many as 50% of all human sequences may be glycos-
ylated (Apweiler et al., 1999) representing a major challenge for
structural biology in terms of producing soluble proteins. This
has stimulated the use of higher eukaryotic cells for the production
of proteins for structural studies, reflected in the steady increase in
the number of structures deposited in the PDB of proteins pro-
duced using these systems (Fig. 1). Notable recent examples in-
clude the first reported structures of G-protein coupled receptors
(Cherezov et al., 2007; Rosenbaum et al., 2007) and the ATP-gated
ion channel P2X4 (Kawate et al., 2009), both produced by baculo-
virus infection of insect cells, and viral glycoproteins obtained from
transient expression in mammalian cells (Bowden et al., 2008a,b).
In this article we review the recent advances in insect and mam-
malian cell expression technology which have improved their ease
of use, increasing both throughput and robustness of these
2. Baculovirus expression system
The baculovirus expression system is a well-established meth-
od for the production of recombinant proteins with the major
1047-8477/$ - see front matter ? 2010 Elsevier Inc. All rights reserved.
* Corresponding author.
E-mail address: email@example.com (R.J. Owens).
1Present address: Novartis Pharma AG/NIBR, Fabrikstrasse 16.2.16, CH-4056 Basel,
Journal of Structural Biology 172 (2010) 55–65
Contents lists available at ScienceDirect
Journal of Structural Biology
journal homepage: www.elsevier.com/locate/yjsbi
advantage over E. coli that expressed proteins undergo post-trans-
lational modifications, e.g., phosphorylation, myristoylation and
glycosylation. Unlike mammalian cells (see Section 3), the N-gly-
cans attached to proteins expressed in insect cells are character-
ised by paucimannose-type structures (Harrison and Jarvis,
2006). The principal baculovirus used for recombinant protein
expression is Autographa californica multiple nucleopolyhedrovirus
(AcMNPV) with Spodoptera frugiperda 9 (Sf9) insect cells as the
expression host. Traditionally, the construction of recombinant
baculoviruses required time-consuming rounds of plaque purifica-
tion to isolate the recombinant virus from a background of wild-
type non-expressing viruses. However, the propagation of the bac-
ulovirus genome as a bacmid in E. coli has revolutionised the
manipulation of the virus and opened the way for efficient and rel-
atively high throughput virus generation (Section 2.1). Small-scale
expression screening combined with easy-to-use disposable biore-
actors has also greatly improved the efficiency of baculovirus
expression technology (Sections 2.2 and 2.3).
2.1. Vectors and viruses
2.1.1. One-step virus production by homologous recombination
The construction of baculoviruses by transposition of the gene
to be expressed into the BAC10 bacmid in E. coli (Luckow et al.,
1993), commercialised as the Bac-to-Bac™ system (Invitrogen), is
probably still the most widely used and enables the routine and ra-
pid production of viruses. A wide variety of vectors are available for
the system, including ones which have been engineered to enable
insertion of genes using ligation-independent cloning methods,
e.g., the Gateway?system (Abdulrahman et al., 2009). Alternative
approaches to simplifying the generation of recombinant baculovi-
ruses have focused on disabling the viral genome used for homol-
ogous recombination in insect cells as the means of generating
recombinant viruses (Je et al., 2001; Zhao et al., 2003). In one
example of this approach a version of the BAC10 bacmid was con-
structed in which the essential gene, ORF1629, had been inacti-
vated by the insertion of the chloramphenicol acetyl transferase
gene to produce BAC10:KO1629(Zhao et al., 2003). Co-transfection
of insect cells with linearized BAC10:KO1629and any standard bac-
ulovirus transfer vector results in the generation of 100% recombi-
nant viruses, removing the need for plaque purification. A similar
approach has been described by Possee et al. (2008) with the added
feature that the chitinase gene (chiA) has been deleted from the
virus (see Section 2.1.2). The linearized bacmid and derivatives
are commercially available as flashBAC™ (Oxford Expression Tech-
nologies Ltd.). Without the need to plaque purify viruses, both
these systems can be readily automated for parallel virus construc-
tion (Possee et al., 2008). In addition, fewer steps are required com-
pared to the Bac-to-Bac™ system thereby saving time in the virus
construction process. As a note of caution, there is some evidence
that all bacmid-derived baculovirus expression vectors undergo
spontaneous deletion following repeated passaging of viruses but
only following more than five rounds of amplification (Pijlman
et al., 2003).
2.1.2. Multi-functional and co-expression vectors
A variety of transfer vectors are available for the construction of
recombinant baculoviruses that encode resident fusion proteins
which have been reported to improve protein expression, including
maltose binding protein (Pengelley et al., 2006), glutathione S-
transferase (Abdulrahman et al., 2009; Romier et al., 2006) and
SUMO (Liu et al., 2008b). In addition, by incorporating promoter
elements for expression in E. coli and/or mammalian cells, parallel
screening of constructs in both bacterial, mammalian and insect
cell systems can be carried out (Berrow et al., 2007; Chambers
et al., 2004; Pengelley et al., 2006). This has the obvious benefit
that switching between expression systems is easy and enables
parallel screening to identify the best expression host. Using such
a dual promoter vector, it has been reported that for a set of 62 hu-
man kinases, 29 were expressed as soluble proteins in E. coli,
whereas 61 were obtained as soluble in insect cells (Chambers
et al., 2004).
To address the structure of multi-protein complexes may re-
quire the co-expression of several components in the same host
since combining individual components produced separately is of-
ten not possible due to poor levels of expression and/or solubility
of the component proteins when expressed on their own. At its
simplest level, co-infection of insect cells with recombinant viruses
for each component can be used to identify suitable constructs
and/or sub-complexes. To maximise co-infection and hence overall
expression of the complex requires optimisation of the multiplicity
of infection (MOI) for each virus. Alternatively, a single virus can be
constructed expressing multiple genes by using a transfer vector
consisting of each opening reading frame under the control of a
separate promoter. Such dual expression vectors are available
Fig. 1. Plot of the cumulative total number of chains deposited in the PDB whose
expression system was identified as either baculovirus or mammalian by year of
deposition. Expression data were parsed from PDB files as described for Table 1. The
data for 2009 (and even some of 2008) are incomplete as some 3450 structures
deposited over this period are yet to be released at the time of writing. Despite this,
Mammalian-derived depositions have already risen ?16% from 2008 to 2009 (139–
162 chains). Otherwise, the rate of deposition has been increasing for both systems
in a trend that is similar to the overall rate of PDB depositions, though, as described
by Levitt (2007), the growth is not exponential. For a more detailed analysis of the
growth of the PDB as a whole without consideration of the expression system, see
Expression systems used for producing proteins for structural biology.
Expression systemNo. chains % of identified
The number of chains deposited in the PDB by expression system, and as a per-
centage of the total number of chains with an identifiable expression system, as of
December 2009. Information about the expression system was parsed from the set
of PDB files available from: ftp://ftp.wwpdb.org/pub/pdb/data/structures/divided/
pdb. Incomplete or inconsistent information was resolved to the best of the authors’
ability or marked as unknown when unresolvable; the vast majority (33422 out of
33550) of the unknowns had no expression information. The system was assigned
as baculovirus where the expression organism was indicated as S. frugiperda or T. ni
in addition to when specifically indicated. Chains were counted rather than PDB
entries as expression information is recorded by chain in the PDB.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
commercially (e.g., pFastBac Dual, Invitrogen) and typically consist
of two baculovirus late promoters (polH or p10) arranged diver-
gently with multiple cloning sites to enable sequential addition
of two genes. To construct double gene vectors in a single step
we have developed a protocol based on ligation-independent clon-
ing using In-fusion™ PCR cloning as outlined in Fig. 2. The In-
Fusion™ reaction catalysed by vaccinia virus polymerase (Hamil-
ton et al., 2007) enables several sequences to be seamlessly joined
through hybridisation of short compatible 50and 30overhangs
(Baogong Zhu et al., 2007). In this way two genes and two promot-
ers can be assembled into a single vector in one step (Fig. 2).
To express a complex with more than two components using
these dual promoter vectors requires co-infection of two or more
transfer vectors which may affect the stoichiometry of compo-
nent/subunit expression. Therefore to produce larger than binary
complexes, it is worth considering incorporating all the compo-
nents into one transfer vector. This is possible using the MultiBac
system developed by Berger et al. (2004) in which separate vectors
are constructed containing each component of a complex under
the control of a promoter and then combined into a multi-gene
transfer vector. In the most recent version of this method the mul-
ti-gene vector is assembled iteratively using cre-lox recombination
in vitro. Construction of the recombinant baculovirus is based on
integration into a bacmid propagated in E. coli using either Tn7
based transposition (Bac-to-Bac™ system) or cre-lox recombina-
tion providing this has not already been used to build the multi-
gene vector (Fitzgerald et al., 2006, 2007). Using this method the
co-expression of six open reading frames (a three component hu-
man TAF5,6,9 transcription factor complex plus three fluorescent
protein markers) has been described (Fitzgerald et al., 2007). As
an alternative to the MultiBac system, Noad et al. (2009) recently
reported the production of two recombinant virus-like particles
in insect by co-expression of multiple protein subunits, e.g., blue
tongue virus VP2/VP3/VP5/VP7. Recombinant baculoviruses were
constructed in which multiple genes had been sequentially inte-
grated into different loci of the baculovirus genome by cre-lox
recombination (Noad et al., 2009). Irrespective of the co-expression
strategy, multi-gene vectors, multiple viruses or a combination of
the two, the overall yield of a complex is obviously determined
by the component that is expressed at the lowest level which will
depend upon factors such as translational efficiency, mRNA stabil-
ity and protein half-life.
2.1.3. Genetically modified viruses
Infection of insect cells in culture with recombinant baculovi-
ruses leads to cell death and release of proteases. To improve the
stability of expressed proteins, attention has focused on two genes
in the baculovirus genome which are not essential for growth of
the virus in cell culture, namely chiA and cath. These encode the en-
zymes chitinase and cathepsin, respectively which are involved in
Fig. 2. Co-expression strategy based on In-Fusion™ cloning. The cloning of the co-expression vector was achieved by the construction of an intermediate vector, carrying the
polh and p10 promoters in relative inverse orientation, and coupled to the SV40 and b-globin polyA regions, respectively. This vector was linearized with PmeI/MscI
restriction cleavage, and the double promoter region was PCR amplified. The two genes, encoding eGFP and RFP, were PCR amplified with primers containing in-fusion tags
homologous to one end of the vector and one end of the promoter region (e.g., the eGFP gene was amplified with primers compatible with the SV40 polyA region of the vector
and polh promoter of the promoter PCR product, whilst the RFP gene was amplified with primers compatible with the p10 promoter and b-globin polyA region). This allowed
the cloning of the dual expression vector by a single-tube, four-way In-Fusion™ reaction. The figures on the right show two images of the same Sf9 cells infected with the dual
expressing baculovirus, one showing GFP fluorescence and one RFP fluorescence.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
liquefaction of infected larvae. As referred to above, the flashBAC™
system uses a virus with a chiA deletion, which appears to favour
the production of secreted proteins (Possee et al., 2008). The ratio-
nale for this is that chitinase accumulates in the endoplasmic retic-
ulum (Thomas et al., 1998) and may thus interfere with those
recombinant proteins that are processed through the secretory
pathway of the cells (Possee et al., 1999). It has been reported that
chitinase may also act as a molecular chaperone for pro-cathepsin,
the precursor of the viral cathepsin, since processing of the enzyme
into a soluble active form does not occur in cells infected with
chiA?baculoviruses (Hom and Volkman, 2000). Preventing the for-
mation of this protease may also contribute to improved product
stability from chiA?viruses. This has led to the introduction of
viruses in which both chiA and cath genes have been removed (Ber-
ger et al., 2004; Kaba et al., 2004) which appear to give an incre-
mental improvement in expression over the single chitinase
deleted virus (Hitchman et al., 2010). However, most recently it
has been shown that combining these knock-outs with deletions
of three other genes (p10, p26 and p73) which are not required
for virus infectivity in cell culture, further enhances recombinant
protein expression (Fig. 3) (Hitchman et al., 2010).
2.2. Optimising expression
The bacmid-based systems described above eliminate the need
for plaque purification and enable multiple viruses to be generated
in parallel, typically by (co-)transfection of attached Sf9 insect cells
in either 6-well or 24-well plates. The primary virus (P0) is usually
amplified by infection of cells in either attached culture or grown
in suspension in 24-deep well blocks and then expression tested
using P1virus again in 24-deep well blocks. Similar optimal condi-
tions for growing Sf9 cells in 24-deep well blocks have been re-
ported by several groups (Bahia et al., 2005; McCall et al., 2005;
Redaelli et al., 2005), namely 3 ml culture volume incubated at
27 ?C on an orbital shaker (Multitron II, InFors-HT Bottmingen,
Switzerland or equivalent, with an orbit of 30 mm) at 250 rpm. Un-
der these conditions cells can be grown to a density of
?5 ? 106cells/ml of culture, which is adequate for expression
studies. Interestingly, it has been reported that P1viral titres ob-
tained from amplification of viruses in suspension are usually 10-
fold higher (?5 ? 108pfu/ml) than for static cultures (Buchs et al.,
2009), removing the need for further rounds of amplification to ob-
tain a high titre viral stock. Although multiplicity of infection can
be determined empirically by varying the volume ratio of P1virus
to the number of cells infected, determining the precise titre of a
baculovirus may sometimes be necessary, e.g., trouble-shooting
lack of expression. Traditional plaque assays require experience
and are time-consuming and so do not fit into a streamlined work-
flow. This has led to the development of alternative methods,
including end-point dilution using fluorescence from GFP co-ex-
pressed with the protein of interest (Philipps et al., 2005), measur-
ing infection by monitoring cell viability using the sensitive redox
indicator AlamarBlue™ (Pouliquen et al., 2006) and more indirect
approaches based on quantitative polymerase chain reaction (Q-
PCR) (Hitchman et al., 2007; Lo and Chao, 2004). Recently the accu-
racy of all of the available methods for estimating baculovirus ti-
tres has been compared and, with the exception of Q-PCR, shown
to give similar values for viral titre (Roldao et al., 2009). Real-time
Q-PCR methods for the assay of viral DNA generally over-estimate
titres since they do not discriminate between infectious viral par-
ticles and defective interfering particles (Roldao et al., 2009). How-
ever, we have found that this over-estimation can be mitigated to a
large extent by inclusion of a pre-titred (by plaque assay) virus
DNA control in the Q-PCR experiment (see also Hitchman et al.,
2007). Quantification of the virus DNA in the samples relative to
this control improves the accuracy. The important advantage of
Q-PCR over all other methods is that it is fast, does not require
the preparation of multiple sub-samples (e.g., dilutions) for each
virus, and is amenable to high-throughput experiments, allowing
the estimation of titres of large numbers of different virus samples
to be measured within a day.
Once a high titre virus stock has been obtained, achieving the
best expression of a recombinant protein requires the optimisation
of two parameters, namely multiplicity of infection (MOI) and time
of harvest post-infection (TOH). Note that absolute quantitation of
virus titres is often not necessary as a simple dilution experiment
with three different virus dilutions spanning three orders of mag-
nitude is often the best way to establish optimal expression condi-
tions. There is a considerable cost benefit if this carried out at
small-scale since the optimal construct and expression parameters
can be defined for a given protein product before committing re-
sources to large-scale production. Again, infection of cells can be
carried out in 24-deep well plates, though 24-well static cultures
can equally well be used. The method for assessing protein expres-
sion levels depends upon the localisation of the product, e.g., cell
surface membrane proteins by FACS (Hanson et al., 2007). For high
throughput applications expression constructs generally include a
hexahistidine tag (His-tag) enabling purification of proteins from
cell lysates using metal affinity chromatography. We and others
have adapted the procedure used for solubility screening in
E. coli (Berrow et al., 2007, 2006) to insect cells using either Ni–
NTA–Sepharose or NiNTA magnetic beads to capture soluble pro-
teins. The same approach can be used to assess levels of secreted
proteins, though a buffer exchange step is required prior to Ni–
NTA purification because the insect media is known to interfere
with binding. Depending upon expression levels and required uses,
expression can be scaled up appropriately using the parameters
optimised by screening.
2.3. Scaling up production and labelling with selenomethionine
Scale-up of the culture and infection of insect cells can be car-
ried out in shake flasks or shaking roller bottles; handling the latter
can be automated using the Cellmate™ robotic system (The
Automaton Partnership, Royston, UK) (McCall et al., 2005). In this
way multiple products can be readily expressed at the 1–2 L scale.
Fig. 3. eGFP expression analysis for different viral genotypes. To analyse intracel-
lular enhanced green fluorescent protein (eGFP), Sf9 cells in 24-well plates were
infected using either chiA?, chiA?/cath?or chiA?/cath?/p10?viruses generated in a
24-well plate format, as described previously (Possee et al., 2008). Briefly, 100 ll of
each co-transfection from a 24-well plate was added to each well of a fresh 24-well
plate and incubated at 27 ?C for 72 h. Cells were then harvested, washed with PBS
and pelleted at 200g for 5 min. Pelleted cells were lysed in 50 mM NaH2PO4, pH 8.0
containing 300 mM NaCl, 10 mM imidazole, 1% v/v Tween 20 and centrifuged at
10,000g for 10 min. Supernatants were distributed into 96-well black/clear poly-
styrene assay plates (BD Biosciences, Oxford, UK) and eGFP fluorescence for each
individual plate well measured at 535 nm following excitation at 485 nm in an
InFinite F200 micro-plate reader (Tecan Group, Theale, UK). Results shown
represent means and +/? standard errors of the mean for 24 replicate samples
from a 24-well plate experiment (adapted from Hitchman et al., 2010).
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
However, oxygen transfer rates limit the culture volume in shake
flasks so for multi-litre scale production, some form of bioreactor
is required. As an alternative to traditional stirred tank bioreactors,
which are complex and expensive devices requiring sterilisation by
autoclaving, a number of disposable cell reactors have been devel-
oped for insect culture at the 5–100 L scale. The Wave™ bag sys-
tem (Wave Biotech) was the first of these disposable bioreactors
to be commercialised and is probably the most widely used. It con-
sists of a single-use plastic bag, inflated with air and then partially
filled with cell suspension which is then is aerated by a continuous
flow of air across the surface of the culture. The bag is rocked on a
heated platform which ensures that the culture is uniformly
warmed to 27 ?C and that cells are continuously mixed (Singh,
1999). Oxygen transfer rates in the Wave™ bag system (or equiv-
alent) are similar to conventional bioreactors (Weber et al., 2002)
and have been incorporated into semi-automated workflows for
protein production in insect cells (Schlaeppi et al., 2006). Neverthe-
less, we and others have found that scale-up from shake flasks to
Wave™ bags usually is accompanied by a protein yield penalty
although the exact reason is unknown. Recently, New Brunswick
(FibraStage™) and Sartorius (Superspinner™ D1000) have devel-
oped disposable bioreactors, offering a potential alternative to
Selenomethionine (SeMet) labelling of proteins enables the
phasing of X-ray diffraction data by multi-wavelength anomalous
diffraction (MAD) (Hendrickson et al., 1990). For recombinant pro-
teins expressed in E. coli incorporation of SeMet is routinely 100%
whereas for other expression systems, including insect cells, yields
are more variable, e.g., 76% for palmitoyl protein thioesterase 1
(Bellizzi et al., 1999), 85% for envelope glycoprotein D from HSV1
(Carfi et al., 2002) and 40% for Traf2 (McWhirter et al., 1999).
Although the number of examples is limited, it appears that higher
levels of incorporation are achieved for secreted compared to intra-
cellular proteins. Recently a systematic study of SeMet labelling of
proteins expressed using the baculovirus system has been reported
(Cronin et al., 2007). The critical step for eliminating protein lack-
ing SeMet from the final product was shown to be the time post-
infection that SeMet is added to the cell culture, since expression
from the late onset polh promoter begins 16–20 h post-infection.
Thus the cell media is replaced with methionine-free media 8 h
post-infection and then after a further 8 h SeMet is added. Increas-
ing the concentration of SeMet in the media was shown to improve
the incorporation of label, with 250 mg/L routinely giving an aver-
age of 77% substitution (for N = 6 proteins) but at the expense of
protein yield, which was reduced to 36% of native expression due
to the toxicity of SeMet. The results were independent of the num-
ber of methionines in the test proteins and whether the product
was intracellular or secreted. Experiments were carried out in pro-
tein-free media with cells grown for a total of 48–96 h post-infec-
tion in both shake flask and Wave™ bioreactors with SeMet
incorporation measured by mass spectrometry (Cronin et al.,
2007). This protocol has recently been used to successfully label
native polyhedra of the AcMNPV baculovirus for structural deter-
mination (Ji et al., 2010).
3. Mammalian expression system
There appears to be a progressive increase in the use of mam-
malian cells for producing proteins for structural biology (Fig. 1).
Traditionally regarded as time-consuming, the introduction of
large-scale transient transfection has made the use of mammalian
cells more attractive in terms of speed and ease of use, particularly
for cell surface and secreted glycoproteins (Sections 3.1 and 3.3). In
contrast to insect cells, glycosylation in mammalian cells results in
attachment of large and complex glycans to the expressed proteins
which generally interfere with crystallization. However, strategies
are available that enable the glycosylation to be modified which in
turn has led to the successful crystallization of glycoproteins pro-
duced in mammalian cells (Section 3.2).
3.1. Cell lines
Various mammalian cell lines have been used for protein
expression with the most common being HEK 293 (human embry-
onic kidney) and CHO (Chinese hamster ovary) which can be trans-
fected using chemicals such as polyethyleneimine (PEI) or calcium
phosphate. HEK 293 cells show the highest level of PEI-mediated
transfection with 50–80% of cells showing GFP expression (Huh
et al., 2007) and are now widely used for production of recombi-
nant proteins both by transient transfection and formation of sta-
ble cell lines. There are two main genetic variants of the HEK 293
cell line, namely 293T which expresses the SV40 large-T antigen
and 293E which expresses the Epstein–Barr virus (EBV) nuclear
antigen 1 (EBNA1). These variants allow episomal amplification
of plasmids containing SV40 or EBV origins of replication, respec-
tively, thus giving more copies of the vector in the transfected cell
enenbroeck et al., 2000). Protein expression in mammalian cells
can also be performed using viral-mediated transduction by such
techniques as the BacMam system (Boyce and Bucher, 1996). This
technology uses recombinant baculoviruses for simple transduc-
tion of mammalian cells, allowing for production of milligram
quantities of protein for structural studies (Dukkipati et al.,
2008). Other cell lines such as COS and Vero (both green African
monkey kidney), HeLa (human cervical cancer), and NS0 (mouse
myeloma) have also been used for structural studies. Some of these
cell lines such as NS0 are more difficult to transfect, usually
achieved using electroporation, and are only used in stable cell line
production (See Section 3.4.3).
A number of vector suites are available for use in mammalian
cells, most of which contain the machinery for high-copy number
replication in E. coli, good for production of DNA, and a strong
mammalian promoter such as the human cytomegalovirus (CMV)
promoter. Such vector suites include the pTT vectors designed by
Durocher et al. which give up to a 10-fold increase in secreted pro-
tein production using HEK 293E cells over non-EBV ori-P contain-
ing vectors (Durocher et al., 2002). Vectors such as the pTriEx
vector suite from Merck have been designed for protein expression
in three hosts: E. coli, insect cells and mammalian cells. In addition,
mammalian expression vectors amenable to ligation-independent
cloning strategies, such as the Gateway™ (Invitrogen) and In-
Fusion™ (Takara Bio) systems, are available for high throughput
vector construction (Berrow et al., 2007; Hartley, 2003).
3.2. Expression screening
In a high-throughput environment, when multiple constructs of
the same protein or multiple proteins are cloned in parallel, assess-
ment of expression level is an essential step. In general, this is done
by small scale transient transfection of mammalian cells which al-
lows screening and analysis to be completed in approximately
4 days (Lee et al., 2009; Nettleship et al., 2009). Expression screen-
ing can be performed with either attached (Lee et al., 2009; Nettle-
ship et al., 2009) or suspension (Chapple et al., 2006; Davies et al.,
2005) cell cultures and for both intracellular and secreted prod-
ucts. For example, Chapple et al. have used transient transfection
of HEK 293E suspension cells in 24-deep well format for the assess-
ment of expression levels of 22 different proteins (Chapple et al.,
2006). Evaluation of expression levels was by dot blot and was
shown to correlate with expression levels subsequently obtained
in 50 ml cultures (Chapple et al., 2006). The small scale transfec-
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
tion protocol is amenable to automation, for example using a Tecan
Evo75™ (Tecan Group Ltd.) liquid handling robot (Fig. 4) (Berrow
et al., 2007). To assess the robustness of automated transfections,
FabOX117 (Nettleship et al., 2008a) was transiently transfected
using the Tecan Evo75™ into every well of three 24-well plates con-
taining HEK 293T attached cells followed by analysis by ELISA. The
average yield was seen to be 3.2 lg/ml (1 ml media per well), with
a 12% deviation across the whole experiment.
For process development using either transient transfection or
stable cells, expression at medium scale is needed which is compa-
rable to the bioreactor used at large scale. For suspension cultures,
the TubeSpin™ system, based on 50 ml centrifugation tubes, al-
lows for a large number of experiments to be run in parallel and
is particularly useful in the evaluation of different, inter-related
cell culture parameters such as media type and sodium butyrate
induction level (De Jesus et al., 2004). Expression levels for His-
tagged proteins can be tested using either IMAC resin in 96-well
or 24-well filter plate format for intracellular and secreted pro-
teins, respectively. The 24-well filter plate allows for the larger vol-
ume of media containing the protein of interest to be applied to the
3.3. Dealing with glycosylation
Unlike proteins produced from insect cells, those secreted from
mammalian cells have a variety of sugar chains attached to the gly-
cosylation sites (Fig. 5). These glycans tend to be large and hetero-
geneous, adversely affecting crystallization of a given glycoprotein
(Butters et al., 1999). O-glycosylation tends to be confined to ex-
tended, unfolded regions of polypeptides which are rich in serine,
threonine and proline, and can therefore be engineered out of the
protein by site-directed mutagenesis or avoiding the region when
designing expression constructs. However, N-glycosylation is often
essential for expression and therefore presents a problem with gly-
can heterogeneity taking two forms: firstly variable occupancy of
glycosylation sites; and secondly variability in the type of glycan
attached. Occupancy of glycosylation sites is usually determined
by mass spectrometry (Dalpathado and Desaire, 2008; Nettleship
et al., 2007) after which any variably occupied sites, which would
lead to heterogeneity in the final product, can be mutated out of
the construct. Because these sites are variably occupied, they are
generally not essential for the stability and function of the glyco-
protein. The glycans attached to a glycoprotein produced in mam-
malian cells can be high-mannose, complex or hybrid in nature
(Fig. 5B). In order to simplify the glycoforms of a protein, formation
of complex glycans is suppressed using two approaches: mutant
cell lines; and glycoprotein processing pathway inhibitors.
3.3.1. Mutant cell lines
Mutant cell lines used for controlling N-glycans are defective in
the production of complex sugars. There are various mutant CHO
cell lines which form modified glycan patterns, such as CHO
Lec22.214.171.124 which produces proteins with sugars of the form
Man5GlcNAc2(Fig. 5C) (Davis et al., 1993). For a recent review of
available lectin-resistant CHO cell lines, see Patnaik and Stanley
(2006). A ricin-selected N-acetylglucosaminyltransferase I (GnTI)
deficient mutant cell line has been established using HEK cells
and is denoted HEK 293S which also gives glycans of the form
Man5GlcNAc2(Fig. 5C) (Reeves et al., 2002b). These CHO and HEK
293 mutant cell lines result in the production of high-mannose gly-
coforms which are sensitive to endoglycosidase (endo) H treat-
ment, thus enabling trimming of the sugars to leave one GlcNAc
attached to the glycosylation site. Mutant cell lines have been used
on a large scale for many crystallization studies both with and
without endo H treatment (Bowden et al., 2009; Langereis et al.,
2009; Zhang et al., 2009b).
In recent years, three main drugs have been used for inhibition
of the glycoprotein processing pathway; N-butyldeoxynojirimycin
(NB-DNJ), swainsonine and kifunensine. NB-DNJ is an a-glucosi-
dase inhibitor which blocks the early stages of N-glycan processing
resulting in endo H sensitive glycoproteins although the best re-
sults are obtained when using the drug with the mutant CHO
Lec126.96.36.199 cell line (Fig. 5D) (Butters et al., 1999). NB-DNJ has been
used for the production of proteins in combination with modified
glycosylation in CHO cells for X-ray crystallographic structure
solution (Evans et al., 2006). Swainsonine inhibits a-mannosidase
II giving predominantly endo H sensitive glycans which are either
high-mannose or hybrid-type, although some sugar chains are not
cleavable by endo H (Fig. 5D) (Chang et al., 2007; Crispin et al.,
2007). Kifunensine is a potent inhibitor of a-mannosidase I (Elbein
et al., 1990) resulting in sugars of the form Man9GlcNAc2(Fig. 5D).
As this glycan-type is high-mannose, the oligosaccharide chains
can be trimmed to leave one GlcNAc using endo H. Kifunensine
has been used in large-scale transient production of glycoproteins
in HEK 293T cells for crystallization studies (Nettleship et al., 2009)
mainly with endo H treatment (Bishop et al., 2009; Bowden et al.,
2008a,b; Carafoli et al., 2009) but also without (Crispin et al., 2009).
3.4. Scaling up production
For scale-up production in mammalian cells, there are many
considerations to take into account such as whether to use tran-
sient transfection or to develop a stable cell line and whether to
use suspension or attached cells. These considerations also have
an impact on the type of vessel or bioreactor necessary for growth
of the cells and in turn what level of automation is possible for the
process. These options and recent advances in technology, in par-
ticular the use of transient transfection of HEK 293E suspension
cells, are discussed in this section.
Fig. 4. Photograph of a Tecan EVO75two-probe liquid handling robot configured for
a small-scale expression screen using transient transfection of HEK 293T cells in
4 ? 24-well plate format. The transfection protocol is controlled by scripts written
in-house using the Tecan EVOWare?software and uses both disposable tips for
handling DNA, transfection agent and transfection cocktail (volume range from 2 to
200 ll using both 10 and 200 ll liquid level sensing conductive tips) and a fixed
steel tip for media, pipetting up to 800 ll per well with eight wells per aspiration.
The liquid handling robot is contained in a Class 100 flow cabinet to maintain an
aseptic environment for all of the robotic manipulations which include preparation
of the DNA/transfection agent cocktail, addition of this to cells and subsequent
media top-up. Following transfection cells are returned by hand to a 37 ?C incubator
gassed with 5% CO2/95% air.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
3.4.1. Transient transfection of attached cells
The advantage of using transient transfection of cells is the
speed of obtaining the final product as milligram quantities of pro-
tein can be obtained within a week (Aricescu et al., 2006a; Lee
et al., 2009; Nettleship et al., 2009). Importantly, the results from
small-scale expression screens performed by transient transfection
are generally predictive of the yields that are obtained at larger
scale. However, the major disadvantage of the transient system is
that milligram quantities of DNA need to be prepared for the trans-
fection. The transfection is usually performed with polyethylene-
imine (Boussif et al., 1995; Huh et al., 2007) due to the cost of
using commercially available transfection reagents on a large scale.
Although the scalability of using attached cells grown in monolay-
ers is limited by the culture surface available (approximately
1 ? 105cells/cm2), the method is useful for the production of se-
creted glycoproteins where maximising cell number is not so crit-
ical (Aricescu et al., 2006a; Nettleship et al., 2009). For example, up
to 40 mg/L of secreted protein can be obtained during a production
run using transient transfection of adherent HEK 293T cells (Arice-
scu et al., 2006a). Such adherent cell systems are also amenable to
automation (see Section 3.4.4).
3.4.2. Transient transfection of suspension cells
In recent years, the use of mammalian cells cultured in suspen-
sion for protein production by transient transfection has increased
dramatically (Derouazi et al., 2004; Durocher et al., 2002; Liu et al.,
2008a; Pham et al., 2006; Rosser et al., 2005). High cell density of
up to 2.4 ? 107cells/ml can be achieved, though transfection usu-
ally takes place at around 1 ? 106cells/ml with production runs
giving over 100 mg/L of secreted proteins, such as the chimeric
heavy chain antibodies produced by Zhang et al. (2009a), and
around 1 mg/L of membrane proteins, such as the GPCR Ste2p
(Shi et al., 2005) from HEK 293E cells. Backliwal et al. have
achieved over 1000 mg/L of IgG production using HEK 293E cells
with PEI-mediated co-transfection of the IgG heavy chain; IgG light
chain, the cell cycle regulators p18 and p21, and the growth factor
acidic Fibroblast Growth Factor (FGFa) (Backliwal et al., 2008).
Transient transfection of HEK 293E cells can be easily scaled up,
for example Girard et al. have performed 100 L production runs
of human anti-RhD-IgG giving a yield of 4 mg/L after 10 days (Gir-
ard et al., 2002). Slightly lower yield can be obtained from transient
transfection of CHO cell lines, with up to 10 mg/L being obtained
for intracellular proteins and 8 mg/L for secreted proteins from
suspension cultures (Derouazi et al., 2004; Haldankar et al., 2006).
Advantages of using suspension cultures when compared to
adherent ones include: the ease of sampling of cultures to allow
for cell counting and viability checks; the lack of trypsinization
for harvesting the cells; and fewer manipulations needed as no
media exchange needs to take place before transfection. There is
also a cost benefit in terms of the reduced use of plasticware com-
pared to large-scale attached cell culture, though this may be off-
set if relatively expensive serum-free media (e.g., 293Freestyle™,
Invitrogen) are used rather than cocktails of standard serum con-
taining base media (e.g., Dulbecco’s modified Eagle’s medium/
Ham’s F12/Iscove’s modified Dulbecco’s medium (Davies et al.,
3.4.3. Stable cell lines
For proteins requiring repeated supply or where transients give
only poor yields (<1 mg/L) then it is advantageous to establish a
stable cell line. Although creation of a high producing stable cell
line takes 2–6 months, after establishment of the cell line expres-
sion of protein is fast, robust and usually in high yield (Wurm,
2004). As a clonal stable cell line continuously expresses the pro-
tein of interest, problems with variability of transfection level are
not encountered. For example, human recombinant IFNa2b can
be steadily produced at levels over 200 mg/L of culture from a
HEK 293 stable cell line (Loignon et al., 2008). In addition, prepara-
tion of large quantities of DNA is not needed as transfection only
takes place during the initial stage of cell line development which
is at small scale. Stable cell lines have been used for many years
and can also be formed in the GnTI negative strains CHO Lec1
Fig. 5. Diagrammatic representation of the major N-linked glycans produced on proteins by (A) insect cells; (B) mammalian cells; (C) the mammalian mutant cell lines CHO
Lec188.8.131.52 and HEK 293S; and (D) CHO and HEK 293 cells in the presence of the N-glycan processing inhibitors N-butyldeoxynojirimycin (NB-DNJ), swainsonine and
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
(Stanley et al., 1990) and HEK 293S (Reeves et al., 2002b). For pro-
teins, which may compromise cell growth if constitutively ex-
pressed, a tetracycline-inducible expression system has been
developed (Reeves et al., 2002a). This system has been used for
the production of GPCRs, e.g., the human olfactory receptor 17-4
(Cook et al., 2008) and13C-labelled rhodopsin (Ahuja et al., 2009).
3.4.4. Scale-up and automation
The technology for the scale up of mammalian cell culture for
protein production is dependent on whether cells are grown at-
tached or in suspension culture. For adherent cells, scale up is usu-
ally performed either in roller bottles, multilayer flasks or with
microcarriers in stirred flasks (Chu and Robinson, 2001). Roller bot-
tles offer the simplest cell culture system, providing a surface area
for cell attachment and growth of 2125 cm2. Scale-up of transient
transfection of HEK 293S/T cells grown in roller bottles is routinely
used for glycoprotein production in Oxford (Aricescu et al., 2006a).
Cell culture in roller bottles can also be automated using either the
MACCS™ (Matrical Bioscience) or the Cellmate™ robotic systems.
As an alternative, multilayer flasks such as the CellSTACK™ (Corn-
ing) and the Nunc Cell Factory™ have become available which pro-
vide a surface area for cell growth of 6320 cm2for each 10 layer
flask. These cell factories have been incorporated into a pipeline
for producing glycoprotein recently reported by Lee et al. (2009),
though culture of cells in these multilayer systems is not amenable
to automation due to their size and dual filling ports. The HYPER-
Flask™ (Corning) is another multilayer flask which is the same size
as a standard tissue culture flasks but has 10 layers for cell attach-
ment providing a total surface area of 1720 cm2. Handling of cells
in HYPERFlasks™ has been fully automated on the SelecT™ robotic
platform (The Automaton Partnership) (Szymanski et al., 2008).
Suspension cells can be grown in vessels from simple shake or
spinner flasks through to Wave™ bags and bioreactors – Bleck-
wenn and Shiloach have written a recent review of options for
large scale cell culture (Bleckwenn and Shiloach, 2004). Using the
Wave™ system, Haldankar et al. achieved between 0.5 and
9.4 mg/L of expressed protein with PEI-mediated transient trans-
fection of CHO-S cells, the yield being protein specific over 12 pro-
teins (Haldankar et al., 2006). Non-disposable bioreactors can be
used for scale-up of mammalian cell cultures grown in suspension
and are available in a variety of sizes. As the environment of a bio-
reactor is controlled, the protein yield is usually greater than that
of a shake flask. For example, Durocher et al. saw a twofold in-
crease in productivity using a bioreactor for the transient expres-
sion of secretedalkaline phosphatase
(Durocher et al., 2002).
inHEK 293E cells
3.5. Labelling of proteins to aid structure solution
3.5.1. Selenomethionine labelling
If structure solution by X-ray crystallography is not amenable
by molecular replacement techniques, SeMet labelling of the pro-
tein of interest is useful for phasing by multi-wavelength anoma-
lous dispersion (MAD). Protocols for SeMet labelling of proteins
expressed in both stable and transient HEK 293 cells have been re-
ported (Aricescu et al., 2006a; Barton et al., 2006). For stable HEK
293 cell lines, Barton et al. recommend pre-incubation of cells in
media lacking methionine for 12 h in order to deplete intracellular
methionine pools before addition of 60 mg/L selenomethionine
(Barton et al., 2006). This method gave 90.5% selenomethionine
incorporation as measured by MALDI–TOF-MS. Although higher
concentrations of selenomethionine result in higher levels of incor-
poration, toxicity also increases, leading to lower protein yields. In
a similar approach we obtained approximately 78% incorporation
of SeMet for a protein produced by HEK 293T transient cultures
supplemented with kifunensine (Fig. 6).
3.5.2. Stable isotope labelling
Labelling of proteins with stable isotopes such as13C and15N is
essential for structure solution by NMR. One novel method for
achieving this is by the use of stable isotope labelled yeast lysates
which can be used for labelling in both mammalian and insect cells
(Egorova-Zachernyuk et al., 2009). As the yeast is grown with only
13C or15N glucose as the carbon source, the lysate can act as a met-
abolic precursor for stable isotope labelling of mammalian cells
and is more cost effective than synthetic stable isotope-labelled
amino acids (Egorova-Zachernyuk et al., 2009). In experiments
using media containing13C and15N-labelled amino acids as well
as a13C-labelled carbon source, 81% incorporation at the carbon
and nitrogen sites was achieved (Osborn and Abramson, 1997).
Advances in both molecular biology and cell culture technology
have greatly facilitated the use of insect and mammalian cells for
producing proteins for structural biology. It is now possible to ap-
ply many of the lessons learnt from structural proteomics projects
based on E. coli to protein production in insect and mammalian
cells, e.g., rapid screening of multiple constructs at small scale to
identify suitable versions for scale-up and purification. With the
potential for over-expression, baculovirus is probably the system
of choice for producing intracellular proteins, whereas large-scale
transient expression in mammalian cells offers a viable alternative
for glycosylated proteins. Baculovirus and mammalian expression
systems are not mutually exclusive. Recently, a combination of
expression screening in transiently transfected HEK cells with pro-
duction in baculovirus infected insect cells has delivered a number
of first-in-class membrane protein structures (Kawate et al., 2009;
Shaffer et al., 2009; Sobolevsky et al., 2009).
The OPPF is funded by the UK Medical Research Council and
Biotechnology and Biological Sciences Research Council.
Abdulrahman, W., Uhring, M., Kolb-Cheynel, I., Garnier, J.M., Moras, D., Rochel, N.,
Busso, D., Poterszman, A., 2009. A set of baculovirus transfer vectors for
screening of affinity tags and parallel expression strategies. Anal. Biochem. 385,
Ahuja, S., Hornak, V., Yan, E.C., Syrett, N., Goncalves, J.A., Hirshfeld, A., Ziliox, M.,
Sakmar, T.P., Sheves, M., Reeves, P.J., Smith, S.O., Eilers, M., 2009. Helix
movement is coupled to displacement of the second extracellular loop in
rhodopsin activation. Nat. Struct. Mol. Biol. 16, 168–175.
Fig. 6. Mass spectrum of the [M+2]2+ion of the peptide QLDLAALIVYWEMEDK
generated by tryptic digest of the human CD274 antigen PDL1, which was
expressed in HEK 293T cells with addition of kifunensine, and analysis by C18
liquid chromatography–ESI-MS (Nettleship et al., 2008b). This peptide contains one
methionine (underlined) and the isotope patterns of sulphur in unlabelled peptide
and selenium in labelled peptide can be seen in the spectrum. The selenomethi-
onine incorporation was calculated to be 78%.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
Apweiler, R., Hermjakob, H., Sharon, N., 1999. On the frequency of protein
glycosylation, as deduced from analysis of the SWISS-PROT database.
Biochim. Biophys. Acta 1473, 4–8.
Aricescu, A.R., Lu, W., Jones, E.Y., 2006a. A time- and cost-efficient system for high-
level protein production in mammalian cells. Acta Crystallogr. D Biol.
Crystallogr. 62, 1243–1250.
Aricescu, A.R., Assenberg, R., Bill, R.M., Busso, D., Chang, V.T., Davis, S.J., Dubrovsky,
A., Gustafsson, L., Hedfalk, K., Heinemann, U., Jones, I.M., Ksiazek, D., Lang, C.,
Maskos, K., Messerschmidt, A., Macieira, S., Peleg, Y., Perrakis, A., Poterszman, A.,
Schneider, G., Sixma, T.K., Sussman, J.L., Sutton, G., Tarboureich, N., Zeev-Ben-
Mordehai, T., Jones, E.Y., 2006b. Eukaryotic expression: developments for
structural proteomics. Acta Crystallogr. D Biol. Crystallogr. 62, 1114–1124.
Backliwal, G., Hildinger, M., Chenuet, S., Wulhfard, S., De Jesus, M., Wurm, F.M.,
2008. Rational vector design and multi-pathway modulation of HEK 293E cells
yield recombinant antibody titers exceeding 1 g/l by transient transfection
under serum-free conditions. Nucleic Acids Res. 36, e96.
Bahia, D., Cheung, R., Buchs, M., Geisse, S., Hunt, I., 2005. Optimisation of insect cell
growth in deep-well blocks: development of a high-throughput insect cell
expression screen. Protein Expr. Purif. 39, 61–70.
Baogong Zhu, G.C., Hall, Emily O., Freeman, Gordon J., 2007. In-Fusion™ assembly:
seamless engineering of multidomain fusion proteins, modular vectors, and
mutations. Biotechniques 43, 354–359.
Barton, W.A., Tzvetkova-Robev, D., Erdjument-Bromage, H., Tempst, P., Nikolov,
D.B., 2006. Highly efficient selenomethionine labeling of recombinant proteins
produced in mammalian cells. Protein Sci. 15, 2008–2013.
Bellizzi, J.J., Widom, J., Kemp, C.W., Clardy, J., 1999. Producing selenomethionine-
labeled proteins with a baculovirus expression vector system. Structure 7,
Berger, I., Fitzgerald, D.J., Richmond, T.J., 2004. Baculovirus expression system for
heterologous multiprotein complexes. Nat. Biotechnol. 22, 1583–1587.
Berrow, N.S., Bussow, K., Coutard, B., Diprose, J., Ekberg, M., Folkers, G.E., Levy, N.,
Lieu, V., Owens, R.J., Peleg, Y., Pinaglia, C., Quevillon-Cheruel, S., Salim, L.,
Scheich, C., Vincentelli, R., Busso, D., 2006. Recombinant protein expression and
solubility screening in Escherichia coli: a comparative study. Acta Crystallogr. D
Biol. Crystallogr. 62, 1218–1226.
Berrow, N.S., Alderton, D., Sainsbury, S., Nettleship, J., Assenberg, R., Rahman, N.,
Stuart, D.I., Owens, R.J., 2007. A versatile ligation-independent cloning method
suitable for high-throughput expression screening applications. Nucleic Acids
Res. 35, e45.
Bishop, B., Aricescu, A.R., Harlos, K., O’Callaghan, C.A., Jones, E.Y., Siebold, C., 2009.
Structural insights into hedgehog ligand sequestration by the human hedgehog-
interacting protein HHIP. Nat. Struct. Mol. Biol. 16, 698–703.
Bleckwenn, N.A., Shiloach, J., 2004. Large-scale cell culture. Curr. Protoc. Immunol.
Appendix 1: Appendix 1U.
Boussif, O., Lezoualc’h, F., Zanta, M.A., Mergny, M.D., Scherman, D., Demeneix, B.,
Behr, J.P., 1995. A versatile vector for gene and oligonucleotide transfer into
cells in culture and in vivo: polyethyleneimine. Proc. Natl. Acad. Sci. USA 92,
Bowden, T.A., Aricescu, A.R., Gilbert, R.J., Grimes, J.M., Jones, E.Y., Stuart, D.I., 2008a.
Structural basis of Nipah and Hendra virus attachment to their cell-surface
receptor ephrin-B2. Nat. Struct. Mol. Biol. 15, 567–572.
Bowden, T.A., Crispin, M., Harvey, D.J., Aricescu, A.R., Grimes, J.M., Jones, E.Y., Stuart,
D.I., 2008b. Crystal structure and carbohydrate analysis of Nipah virus
attachment glycoprotein: a template for antiviral and vaccine design. J. Virol.
Bowden, T.A., Crispin, M., Graham, S.C., Harvey, D.J., Grimes, J.M., Jones, E.Y., Stuart,
D.I., 2009. Unusual molecular architecture of the machupo virus attachment
glycoprotein. J. Virol. 83, 8259–8265.
Boyce, F.M.,Bucher, N.L.,1996. Baculovirus-mediated
mammalian cells. Proc. Natl. Acad. Sci. USA 93, 2348–2352.
Buchs, M., Kim, E., Pouliquen, Y., Sachs, M., Geisse, S., Mahnke, M., Hunt, I., 2009.
High-throughput insect cell protein expression applications. Methods Mol. Biol.
Butters, T.D., Sparks, L.M., Harlos, K., Ikemizu, S., Stuart, D.I., Jones, E.Y., Davis, S.J.,
1999. Effects of N-butyldeoxynojirimycin and the Lec184.108.40.206 mutant phenotype
on N-glycan processing in Chinese hamster ovary cells: application to
glycoprotein crystallization. Protein Sci. 8, 1696–1701.
Carafoli, F., Clout, N.J., Hohenester, E., 2009. Crystal structure of the LG1–3 region of
the laminin alpha2 chain. J. Biol. Chem. 284, 22786–22792.
Carfi, A., Gong, H., Lou, H., Willis, S.H., Cohen, G.H., Eisenberg, R.J., Wiley, D.C., 2002.
Crystallization and preliminary diffraction studies of the ectodomain of the
envelope glycoprotein D from herpes simplex virus 1 alone and in complex with
the ectodomain of the human receptor HveA. Acta Crystallogr. D Biol.
Crystallogr. 58, 836–838.
Chambers, S.P., Austen, D.A., Fulghum, J.R., Kim, W.M., 2004. High-throughput
screening for soluble recombinant expressed kinases in Escherichia coli and
insect cells. Protein Expr. Purif. 36, 40–47.
Chang, V.T., Crispin, M., Aricescu, A.R., Harvey, D.J., Nettleship, J.E., Fennelly, J.A., Yu,
C., Boles, K.S., Evans, E.J., Stuart, D.I., Dwek, R.A., Jones, E.Y., Owens, R.J., Davis,
S.J., 2007. Glycoprotein structural genomics: solving the glycosylation problem.
Structure 15, 267–273.
Chapple, S.D., Crofts, A.M., Shadbolt, S.P., McCafferty, J., Dyson, M.R., 2006.
Multiplexed expression and screening for recombinant protein production in
mammalian cells. BMC Biotechnol. 6, 49.
Cherezov, V., Rosenbaum, D.M., Hanson, M.A., Rasmussen, S.G., Thian, F.S., Kobilka,
T.S., Choi, H.J., Kuhn, P., Weis, W.I., Kobilka, B.K., Stevens, R.C., 2007. High-
resolution crystal structure of an engineered human beta2-adrenergic G
protein-coupled receptor. Science 318, 1258–1265.
Chu, L., Robinson, D.K., 2001. Industrial choices for protein production by large-
scale cell culture. Curr. Opin. Biotechnol. 12, 180–187.
Cook, B.L., Ernberg, K.E., Chung, H., Zhang, S., 2008. Study of a synthetic human
olfactory receptor 17-4: expression and purification from an inducible
mammalian cell line. PLoS One 3, e2920.
Crispin, M., Aricescu, A.R., Chang, V.T., Jones, E.Y., Stuart, D.I., Dwek, R.A., Davis, S.J.,
Harvey, D.J., 2007. Disruption of alpha-mannosidase processing induces non-
canonical hybrid-type glycosylation. FEBS Lett. 581, 1963–1968.
Crispin, M., Bowden, T.A., Coles, C.H., Harlos, K., Aricescu, A.R., Harvey, D.J., Stuart,
D.I., Jones, E.Y., 2009. Carbohydrate and domain architecture of an immature
antibody glycoform exhibiting enhanced effector functions. J. Mol. Biol. 387,
Cronin, C.N., Lim, K.B., Rogers, J., 2007. Production of selenomethionyl-derivatized
proteins in baculovirus-infected insect cells. Protein Sci. 16, 2023–2029.
Dalpathado, D.S., Desaire, H., 2008. Glycopeptide analysis by mass spectrometry.
Analyst 133, 731–738.
Davies, A., Greene, A., Lullau, E., Abbott, W.M., 2005. Optimisation and evaluation of
a high-throughput mammalian protein expression system. Protein Expr. Purif.
Davis, S.J., Puklavec, M.J., Ashford, D.A., Harlos, K., Jones, E.Y., Stuart, D.I., Williams,
A.F., 1993. Expression of soluble recombinant glycoproteins with predefined
glycosylation: application to the crystallization of the T-cell glycoprotein CD2.
Protein Eng. 6, 229–232.
De Jesus, M.J., Girard, P., Bourgeois, M., Baumgartner, G., Jacko, B., Amstutz, H.,
Wurm, F.M., 2004. TubeSpin satellites: a fast track approach for process
development with animal cells using shaking technology. Biochem. Eng. J. 17,
Derouazi, M., Girard, P., Van Tilborgh, F., Iglesias, K., Muller, N., Bertschinger, M.,
Wurm, F.M., 2004. Serum-free large-scale transient transfection of CHO cells.
Biotechnol. Bioeng. 87, 537–545.
Dukkipati, A., Park, H.H., Waghray, D., Fischer, S., Garcia, K.C., 2008. BacMam system
for high-level expression of recombinant soluble and membrane glycoproteins
for structural studies. Protein Expr. Purif. 62, 160–170.
Durocher, Y., Perret, S., Kamen, A., 2002. High-level and high-throughput
recombinant protein production by transient transfection of suspension-
growing human 293-EBNA1 cells. Nucleic Acids Res. 30, E9.
Egorova-Zachernyuk, T.A., Bosman, G.J., Pistorius, A.M., DeGrip, W.J., 2009.
Production of yeastolates for uniform stable isotope labelling in eukaryotic
cell culture. Appl. Microbiol. Biotechnol. 84, 575–581.
Elbein, A.D., Tropea, J.E., Mitchell, M., Kaushal, G.P., 1990. Kifunensine, a potent
inhibitor of the glycoprotein processing mannosidase I. J. Biol. Chem. 265,
Evans, E.J., Castro, M.A., O’Brien, R., Kearney, A., Walsh, H., Sparks, L.M., Tucknott,
M.G., Davies, E.A., Carmo, A.M., van der Merwe, P.A., Stuart, D.I., Jones, E.Y.,
Ladbury, J.E., Ikemizu, S., Davis, S.J., 2006. Crystal structure and binding
properties of the CD2 and CD244 (2B4)-binding protein, CD48. J. Biol. Chem.
Fitzgerald, D.J., Berger, P., Schaffitzel, C., Yamada, K., Richmond, T.J., Berger, I., 2006.
Protein complex expression by using multigene baculoviral vectors. Nat.
Methods 3, 1021–1032.
Fitzgerald, D.J., Schaffitzel, C., Berger, P., Wellinger, R., Bieniossek, C., Richmond, T.J.,
Berger, I., 2007. Multiprotein expression strategy for structural biology of
eukaryotic complexes. Structure 15, 275–279.
Fogg, M.J., Alzari, P., Bahar, M., Bertini, I., Betton, J.M., Burmeister, W.P.,
Cambillau, C., Canard, B., Corrondo, M.A., Coll, M., Daenke, S., Dym, O.,
Egloff, M.P., Enguita, F.J., Geerlof, A., Haouz, A., Jones, T.A., Ma, Q., Manicka,
S.N., Migliardi, M., Nordlund, P., Owens, R.J., Peleg, Y., Schneider, G., Schnell,
R., Stuart, D.I., Tarbouriech, N., Unge, T., Wilkinson, A.J., Wilmanns, M.,
Wilson, K.S., Zimhony, O., Grimes, J.M., 2006. Application of the use of high-
throughput technologies to the determination of protein structures of
bacterial and viral pathogens. Acta Crystallogr. D Biol. Crystallogr. 62,
Girard, P., Derouazi, M., Baumgartner, G., Bourgeois, M., Jordan, M., Jacko, B., Wurm,
F.M., 2002. 100-liter transient transfection. Cytotechnology 38, 15–21.
Graslund, S., Nordlund, P., Weigelt, J., Hallberg, B.M., Bray, J., Gileadi, O., Knapp, S.,
Oppermann, U., Arrowsmith, C., Hui, R., Ming, J., dhe-Paganon, S., Park, H.W.,
Savchenko, A., Yee, A., Edwards, A., Vincentelli, R., Cambillau, C., Kim, R., Kim,
S.H., Rao, Z., Shi, Y., Terwilliger, T.C., Kim, C.Y., Hung, L.W., Waldo, G.S., Peleg, Y.,
Albeck, S., Unger, T., Dym, O., Prilusky, J., Sussman, J.L., Stevens, R.C., Lesley, S.A.,
Wilson, I.A., Joachimiak, A., Collart, F., Dementieva, I., Donnelly, M.I.,
Eschenfeldt, W.H., Kim, Y., Stols, L., Wu, R., Zhou, M., Burley, S.K., Emtage, J.S.,
Sauder, J.M., Thompson, D., Bain, K., Luz, J., Gheyi, T., Zhang, F., Atwell, S., Almo,
S.C., Bonanno, J.B., Fiser, A., Swaminathan, S., Studier, F.W., Chance, M.R., Sali, A.,
Acton, T.B., Xiao, R., Zhao, L., Ma, L.C., Hunt, J.F., Tong, L., Cunningham, K., Inouye,
M., Anderson, S., Janjua, H., Shastry, R., Ho, C.K., Wang, D., Wang, H., Jiang, M.,
Montelione, G.T., Stuart, D.I., Owens, R.J., Daenke, S., Schutz, A., Heinemann, U.,
Yokoyama, S., Bussow, K., Gunsalus, K.C., 2008. Protein production and
purification. Nat. Methods 5, 135–146.
Haldankar, R., Li, D., Saremi, Z., Baikalov, C., Deshpande, R., 2006. Serum-free
suspension large-scale transient transfection of CHO cells in WAVE bioreactors.
Mol. Biotechnol. 34, 191–199.
Hamilton, M.D., Nuara, A.A., Gammon, D.B., Buller, R.M., Evans, D.H., 2007. Duplex
strand joining reactions catalyzed by vaccinia virus DNA polymerase. Nucleic
Acids Res. 35, 143–151.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
Hanson, M.A., Brooun, A., Baker, K.A., Jaakola, V.P., Roth, C., Chien, E.Y., Alexandrov,
A., Velasquez, J., Davis, L., Griffith, M., Moy, K., Ganser-Pornillos, B.K., Hua, Y.,
Kuhn, P., Ellis, S., Yeager, M., Stevens, R.C., 2007. Profiling of membrane protein
variants in a baculovirus system by coupling cell-surface detection with small-
scale parallel expression. Protein Expr. Purif. 56, 85–92.
Harrison, R.L., Jarvis, D.L., 2006. Protein N-glycosylation in the baculovirus-insect
‘‘mammalianized” recombinant glycoproteins. Adv. Virus Res. 68, 159–191.
Hartley, J.L., 2003. Use of the gateway system for protein expression in multiple
hosts. Curr. Protoc. Protein Sci.. Chapter 5: Unit 5 17.
Hendrickson, W.A., Horton, J.R., LeMaster, D.M., 1990. Selenomethionyl proteins
produced for analysis by multiwavelength anomalous diffraction (MAD): a
vehicle for direct determination of three-dimensional structure. EMBO J. 9,
Hitchman, R.B., Siaterli, E.A., Nixon, C.P., King, L.A., 2007. Quantitative real-time PCR
for rapid and accurate titration of recombinant baculovirus particles.
Biotechnol. Bioeng. 96, 810–814.
Hitchman, R.B., Possee, R.D., Crombie, A.T., Chambers, A., Ho, K., Siaterli, E., Lissina,
O., Sternard, H., Novy, R., Loomis, K., Bird, L.E., Owens, R.J., King, L.A., 2010.
Genetic modification of a baculovirus vector for increased expression in insect
cells. Cell Biol. Toxicol. 26, 57–68.
Hom, L.G., Volkman, L.E., 2000. Autographa californica M nucleopolyhedrovirus chiA
is required for processing of V-CATH. Virology 277, 178–183.
Huh, S.H., Do, H.J., Lim, H.Y., Kim, D.K., Choi, S.J., Song, H., Kim, N.H., Park, J.K., Chang,
W.K., Chung, H.M.,Kim, J.H.,2007.
polyethyleneimine for efficient gene delivery. Biologicals 35, 165–171.
Je, Y., Chang, J.H., choi, J.Y., Roh, J.Y., Jin, B.R., O’Reilly, D.R., Kang, S.K., 2001. A
defective viral genome maintained in Escherichia coli for the generation of
baculovirus expression vectors. Biotechnol. Lett. 23, 575–582.
Ji, X., Sutton, G., Evans, G., Axford, D., Owen, R., Stuart, D.I., 2010. How baculovirus
polyhedra fit square pegs into round holes to robustly package viruses. EMBO J.
Kaba, S.A., Salcedo, A.M., Wafula, P.O., Vlak, J.M., van Oers, M.M., 2004. Development
of a chitinase and v-cathepsin negative bacmid for improved integrity of
secreted recombinant proteins. J. Virol. Methods 122, 113–118.
Kawate, T., Michel, J.C., Birdsong, W.T., Gouaux, E., 2009. Crystal structure of the
ATP-gated P2X(4) ion channel in the closed state. Nature 460, 592–598.
Langereis, M.A., Zeng, Q., Gerwig, G.J., Frey, B., von Itzstein, M., Kamerling, J.P., de
Groot, R.J., Huizinga, E.G., 2009. Structural basis for ligand and substrate
recognition by torovirus hemagglutinin esterases. Proc. Natl. Acad. Sci. USA 106,
Lee, J.E., Fusco, M.L., Ollmann Saphire, E., 2009. An efficient platform for screening
expression and crystallization of glycoproteins produced in human cells. Nat.
Protoc. 4, 592–604.
Levitt, M., 2007. Growth of novel protein structural data. Proc. Natl. Acad. Sci. USA
Liu, C., Dalby, B., Chen, W., Kilzer, J.M., Chiou, H.C., 2008a. Transient transfection
factors for high-level recombinant protein production in suspension cultured
mammalian cells. Mol. Biotechnol. 39, 141–153.
Liu, L., Spurrier, J., Butt, T.R., Strickler, J.E., 2008b. Enhanced protein expression in
the baculovirus/insect cell system using engineered SUMO fusions. Protein
Expr. Purif. 62, 21–28.
Lo, H.R., Chao, Y.C., 2004. Rapid titer determination of baculovirus by quantitative
real-time polymerase chain reaction. Biotechnol. Prog. 20, 354–360.
Loignon, M., Perret, S., Kelly, J., Boulais, D., Cass, B., Bisson, L., Afkhamizarreh, F.,
Durocher, Y., 2008. Stable high volumetric production of glycosylated human
recombinant IFNalpha2b in HEK293 cells. BMC Biotechnol. 8, 65.
Luckow, V.A., Lee, S.C., Barry, G.F., Olins, P.O., 1993. Efficient generation of infectious
recombinant baculoviruses by site-specific transposon-mediated insertion of
foreign genes into a baculovirus genome propagated in Escherichia coli. J. Virol.
McCall, E.J., Danielsson, A., Hardern, I.M., Dartsch, C., Hicks, R., Wahlberg, J.M.,
Abbott, W.M., 2005. Improvements to the throughput of recombinant protein
expression in the baculovirus/insect cell system. Protein Expr. Purif. 42, 29–36.
McWhirter, S.M., Pullen, S.S., Holton, J.M., Crute, J.J., Kehry, M.R., Alber, T., 1999.
Crystallographic analysis of CD40 recognition and signaling by human TRAF2.
Proc. Natl. Acad. Sci. USA 96, 8408–8413.
Nettleship, J.E., Aplin, R., Aricescu, A.R., Evans, E.J., Davis, S.J., Crispin, M., Owens, R.J.,
2007. Analysis of variable N-glycosylation site occupancy in glycoproteins by
liquid chromatography electrospray ionization mass spectrometry. Anal.
Biochem. 361, 149–151.
Nettleship, J.E., Ren, J., Rahman, N., Berrow, N.S., Hatherley, D., Barclay, A.N., Owens,
R.J., 2008a. A pipeline for the production of antibody fragments for structural
studies using transient expression in HEK 293T cells. Protein Expr. Purif. 62, 83–
Nettleship, J.E., Brown, J., Groves, M.R., Geerlof, A., 2008b. Methods for protein
characterization by mass spectrometry, thermal shift (ThermoFluor) assay, and
multiangle or static light scattering. Methods Mol. Biol. 426, 299–318.
Nettleship, J.E., Rahman-Huq, N., Owens, R.J., 2009. The production of glycoproteins
by transient expression in mammalian cells. Methods Mol. Biol. 498, 245–263.
Noad, R.J., Stewart, M., Boyce, M., Celma, C.C., Willison, K.R., Roy, P., 2009. Multigene
expression of protein complexes by iterative modification of genomic Bacmid
DNA. BMC Mol. Biol. 10, 87.
Osborn, B.L., Abramson, F.P., 1997. Production of uniformly stable isotope labelled
proteins from mammalian cells. J. Labelled Compd. Radiopharm. 39, 935–953.
of insectcellsto produce
Optimizationof 25 kDa linear
Patnaik, S.K., Stanley, P., 2006. Lectin-resistant CHO glycosylation mutants. Methods
Enzymol. 416, 159–182.
Pengelley, S.C., Chapman, D.C., Mark Abbott, W., Lin, H.H., Huang, W., Dalton, K.,
Jones, I.M., 2006. A suite of parallel vectors for baculovirus expression. Protein
Expr. Purif. 48, 173–181.
Pham, P.L., Kamen, A., Durocher, Y., 2006. Large-scale transfection of mammalian
cells for the fast production of recombinant protein. Mol. Biotechnol. 34, 225–
Philipps, B., Forstner, M., Mayr, L.M., 2005. A baculovirus expression vector system
for simultaneous protein expression in insect and mammalian cells. Biotechnol.
Prog. 21, 708–711.
Pijlman, G.P., van Schijndel, J.E., Vlak, J.M., 2003. Spontaneous excision of BAC vector
sequences from bacmid-derived baculovirus expression vectors upon passage
in insect cells. J. Gen. Virol. 84, 2669–2678.
Possee, R.D., Thomas, C.J., King, L.A., 1999. The use of baculovirus vectors for the
production of membrane proteins in insect cells. Biochem. Soc. Trans. 27, 928–
Possee, R.D., Hitchman, R.B., Richards, K.S., Mann, S.G., Siaterli, E., Nixon, C.P., Irving,
H., Assenberg, R., Alderton, D., Owens, R.J., King, L.A., 2008. Generation of
baculovirus vectors for the high-throughput production of proteins in insect
cells. Biotechnol. Bioeng. 101, 1115–1122.
Pouliquen, Y., Kolbinger, F., Geisse, S., Mahnke, M., 2006. Automated baculovirus
titration assay based on viable cell growth monitoring using a colorimetric
indicator. BioTechniques 40, 1–4.
Redaelli, L., Zolezzi, F., Nardese, V., Bellanti, B., Wanke, V., Carettoni, D., 2005. A
platform for high-throughput expression of recombinant human enzymes
secreted by insect cells. J. Biotechnol. 120, 59–71.
rhodopsin: a tetracycline-inducible system in stable mammalian cell lines
for high-level expression of opsin mutants. Proc. Natl. Acad. Sci. USA 99,
Reeves, P.J., Callewaert, N., Contreras, R., Khorana, H.G., 2002b. Structure and
function in rhodopsin: high-level expression of rhodopsin with restricted and
acetylglucosaminyltransferase I-negative HEK293S stable mammalian cell
line. Proc. Natl. Acad. Sci. USA 99, 13419–13424.
Roldao, A., Oliveira, R., Carrondo, M.J., Alves, P.M., 2009. Error assessment in
recombinant baculovirus titration: evaluation of different methods. J. Virol.
Methods 159, 69–80.
Romier, C., Ben Jelloul, M., Albeck, S., Buchwald, G., Busso, D., Celie, P.H.,
Christodoulou, E., De Marco, V., van Gerwen, S., Knipscheer, P., Lebbink, J.H.,
Notenboom, V., Poterszman, A., Rochel, N., Cohen, S.X., Unger, T., Sussman,
J.L., Moras, D., Sixma, T.K., Perrakis, A., 2006. Co-expression of protein
complexes in prokaryotic and eukaryotic hosts: experimental procedures,
database tracking and case studies. Acta Crystallogr. D Biol. Crystallogr. 62,
Rosenbaum, D.M., Cherezov, V., Hanson, M.A., Rasmussen, S.G., Thian, F.S., Kobilka,
T.S., Choi, H.J., Yao, X.J., Weis, W.I., Stevens, R.C., Kobilka, B.K., 2007. GPCR
engineering yields high-resolution structural insights into beta2-adrenergic
receptor function. Science 318, 1266–1273.
Rosser, M.P., Xia, W., Hartsell, S., McCaman, M., Zhu, Y., Wang, S., Harvey, S.,
Bringmann, P., Cobb, R.R., 2005. Transient transfection of CHO-K1-S using
serum-free medium in suspension: a rapid mammalian protein expression
system. Protein Expr. Purif. 40, 237–243.
Schlaeppi, J.M., Henke, M., Mahnke, M., Hartmann, S., Schmitz, R., Pouliquen, Y.,
Kerins, B., Weber, E., Kolbinger, F., Kocher, H.P., 2006. A semi-automated large-
scale process for the production of recombinant tagged proteins in the
baculovirus expression system. Protein Expr. Purif. 50, 185–195.
Shaffer, P.L., Goehring, A., Shankaranarayanan, A., Gouaux, E., 2009. Structure and
mechanism of a Na+-independent amino acid transporter. Science 325, 1010–
Shi, C., Shin, Y.O., Hanson, J., Cass, B., Loewen, M.C., Durocher, Y., 2005. Purification
and characterizationofa recombinant
Saccharomyces cerevisiae Ste2p, transiently expressed in HEK293 EBNA1 cells.
Biochemistry 44, 15705–15714.
Singh, V., 1999. Disposable bioreactor for cell culture using wave-induced agitation.
Cytotechnology 30, 149–158.
Sobolevsky, A.I., Rosconi, M.P., Gouaux, E., 2009. X-ray structure, symmetry and
Stanley, P., Sallustio, S., Krag, S.S., Dunn, B., 1990. Lectin-resistant CHO cells:
selection of seven new mutants resistant to ricin. Somat. Cell Mol. Genet. 16,
Szymanski, S.L., Huff, K.W., Patel, A.D., Murray, J.R., Feasby, J., Sharma, B.V., Young,
D.K., Strulovici, B., Peltier, R.R., Johnson, E.N., Rush, A., 2008. Automated
application of a novel high yield, high performance tissue culture flask. JALA 13,
Thomas, C.J., Brown, H.L., Hawes, C.R., Lee, B.Y., Min, M.K., King, L.A., Possee, R.D.,
1998. Localization of a baculovirus-induced chitinase in the insect cell
endoplasmic reticulum. J. Virol. 72, 10207–10212.
Van Craenenbroeck, K., Vanhoenacker, P., Haegeman, G., 2000. Episomal vectors for
gene expression in mammalian cells. Eur. J. Biochem. 267, 5665–5678.
Weber, W., Weber, E., Geisse, S., Memmert, K., 2002. Optimisation of protein
expression and establishment of the Wave Bioreactor for Baculovirus/insect cell
culture. Cytotechnology 38, 77–85.
2002a.Structureand function in
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65
Wurm, F.M., 2004. Production of recombinant protein therapeutics in cultivated Download full-text
mammalian cells. Nat. Biotechnol. 22, 1393–1398.
Zhang, J., Liu, X., Bell, A., To, R., Baral, T.N., Azizi, A., Li, J., Cass, B., Durocher, Y., 2009a.
Transient expression and purification of chimeric heavy chain antibodies.
Protein Expr. Purif. 65, 77–82.
Zhang, Q., Zhou, Y.F., Zhang, C.Z., Zhang, X., Lu, C., Springer, T.A., 2009b. Structural
specializations of A2, a force-sensing domain in the ultralarge vascular protein
von Willebrand factor. Proc. Natl. Acad. Sci. USA 106, 9226–9231.
Zhao, Y., Chapman, D.A., Jones, I.M., 2003. Improving baculovirus recombination.
Nucleic Acids Res. 31. E6–6.
J.E. Nettleship et al./Journal of Structural Biology 172 (2010) 55–65