Directed evolution of mammalian anti-apoptosis
proteins by somatic hypermutation
Brian S.Majors1,2, Gisela G.Chiang2, Nels E.Pederson2
and Michael J.Betenbaugh1,3
1Department of Chemical and Biomolecular Engineering, The Johns
Hopkins University, 3400 North Charles Street, 221 Maryland Hall,
Baltimore, MD 21218-2694, USA and2Cellular Engineering, Biogen Idec,
Inc., 14 Cambridge Center, Cambridge, MA 02142, USA
3To whom correspondence should be addressed.
Received June 27, 2011; revised October 12, 2011;
accepted October 13, 2011
Edited by Stephen Bottomley
Recently, researchers have created novel fluorescent pro-
teins by harnessing the somatic hypermutation ability of
B cells. In this study, we examined if this approach could
be used to evolve a non-fluorescent protein, namely the
anti-apoptosis protein Bcl-xL, using the Ramos B-cell line.
After demonstrating that Ramos cells were capable of
mutating a heterologous bcl-xLtransgene, the cells were
exposed to multiple rounds of the chemical apoptosis
inducer staurosporine followed by rounds of recovery in
fresh medium. The engineered B cells expressing Bcl-xL
exhibited progressively lower increases in apoptosis acti-
vation as measured by caspase-3 activity after successive
rounds of selective pressure with staurosporine treatment.
Within the B-cell genome, a number of mutated bcl-xL
transgene variants were identified after three rounds of
evolution, including a mutation of Bcl-xLAsp29 to either
Asn or His, in 8 out of 23 evaluated constructs that repre-
sented at least five distinct Ramos subpopulations.
Subsequently, Chinese hamster ovary (CHO) cells engi-
neered to overexpress the Bcl-xL Asp29Asn variant
showed enhanced apoptosis resistance against an orthog-
onal apoptosis insult, Sindbis virus infection, when com-
pared with cells expressing the wild-type Bcl-xLprotein.
These findings provide, to our knowledge, the first dem-
onstration of evolution of a recombinant mammalian
protein in a mammalian expression system.
Keywords: apoptosis/Bcl-xL/directed evolution/Ramos B
Directed evolution has become a powerful tool for biomole-
cular engineers interested in generating proteins with novel
characteristics or enhanced function. Directed evolution
involves diversification of a gene of interest through muta-
genesis, expression of the mutant protein and selection based
upon an easily identifiable and favorable trait. Such altered
proteins may exhibit increased activity (Kauffmann and
Schmidt-Dannert, 2001), specificity (Kuchner and Arnold,
1997; Liebeton et al., 2000) or properties including improved
stability (Miyazaki et al., 2000; Lehmann and Wyss, 2001),
novel fluorescent characteristics (Zacharias and Tsien, 2006)
or ability to act as molecular switches (Guntas et al., 2005;
Error-prone PCR has been the method of choice for
generation of mutant libraries for directed evolution, and
selection of mutant proteins of interest is typically carried
out in bacteria or yeast. These organisms allow for high
rapid molecular biology manipulation. However, the direc-
ted evolution of mammalian genes is not as well estab-
lished for a number of reasons. Mammalian genes can be
evolved in bacterial systems (Aharoni et al., 2004; Kumar
et al., 2005), although the protein is not in its native
physiological environment and its structure or function
may be affected by the presence or absence of other pro-
teins, ion concentrations and pH gradients. In addition,
many mammalian proteins are transported to specific cellu-
lar compartments, perform specific cellular functions or
undergo post-translational modifications, such as complex
glycosylation, that are specific to mammalian cells. These
differences between mammalian and bacterial proteins may
significantly impact the protein structure and activity. A
number of properties make mammalian cells less than
ideal for expression and selection in traditional directed
evolution experiments including the slow growth of mam-
malian cells, low efficiency of stable integration, tendency
toward multiple gene insertions, highly variable expression
levels and time-consuming molecular biology (Majors
et al., 2009b).
Alternatively, nature’s method for directed evolution in
mammalian cells is most evident in B cells of the immune
system. B cells, formed in the bone marrow, are responsible
recognition of any number of antigens an organism may
encounter. To generate antibodies with increasing specificity
to antigens, B cells undergo random mutagenesis of their
antibody-encoding genes, a process known as somatic
hypermutation. Somatic hypermutation requires the protein
activation-induced (cytidine) deaminase (AID), which prefer-
entially targets transcriptionally active genes in the Ig vari-
able locus (Bachl et al., 2001) and causes the deamination of
cytosines to uracil. The mutagenesis is thought to occur
during an error-prone repair process of the U:G base pairing,
causing single-nucleotide changes and less often insertions
activity (Perez-Duran et al., 2007). In B cells, the resulting
mutations give rise to an antibody diversity that can bind
new antigens or known antigens with increasing specificity
(Cumbers et al., 2002).
# The Author 2011. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: email@example.com
Protein Engineering, Design & Selection vol. 25 no. 1 pp. 27–38, 2012
Published online December 9, 2011 doi:10.1093/protein/gzr052
Researchers have utilized this antibody diversification ma-
chinery to evolve both endogenous IgM from B cells (Cumbers
et al., 2002) and exogenous genes introduced into the B cells
(Bachl et al., 2001; Wang et al., 2004a,b; Kanayama et al.,
2006; Wang and Tsien, 2006).The ability to evolve exogenous
transgenes in B cells was first demonstrated using a novel red
fluorescent protein (RFP). The scientists inserted the gene for
RFP into the constitutively hypermutating B-cell line Ramos
and selected cells containing fluorescent proteins with shifted
emission wavelengths using fluorescence-activated cell sorting.
They found that the hypermutation ability of the B cells created
a collection of RFP protein variants (Wang et al., 2004b) in-
cluding one RFP variant protein exhibiting an emission wave-
length shift of 45 nm into the infrared. This protein is now
being investigated for possible use in in vivo imaging.
Additional studies utilizing the chicken DT40 B-cell line have
shown the capacity to mutate fluorescent proteins by somatic
hypermutation (Arakawa et al., 2008) and another endogenous
B-cell gene diversification mechanism, gene conversion
(Kanayama et al., 2006).
While using B cells for directed evolution has been demon-
strated using fluorometric proteins, to our knowledge, no
studies have been undertaken to demonstrate the potential of
this technology for evolution of non-fluorescent recombinant
proteins. One of the limitations is that a reliable method for
selection of mutants is needed in order to be able to isolate
proteins with improved function or altered activity. The goal
of the current study was to use the induction of apoptosis as a
method for evolving heterologous genes with improved anti-
apoptotic function. In this study the target gene was bcl-xL,
one of the principal anti-apoptotic Bcl-2 family members.
Bcl-xLhas been shown to inhibit apoptosis through direct
maintenance of the mitochondrial membrane and through
binding of pro-apoptosis proteins in the presence of apoptotic
stimuli (Majors et al., 2007). In addition, expression of ex-
ogenous Bcl-xLhas been employed in a biotechnology role to
prevent cell death in mammalian cells used to produce recom-
binant proteins. Bcl-xLoverexpression has been applied to
enhance culture viability (Figueroa et al., 2003, 2004), in-
crease viable cell density (Chiang and Sisk, 2005; Kim et al.,
2009), and increase productivity (Meents et al., 2002; Chiang
and Sisk, 2005; Majors et al., 2008) in mammalian cell lines
engineered to express recombinant therapeutic proteins.
In this study, we employed the Ramos B-cell line as a
gain-of-function mutant of the bcl-xLgene. Cells expressing
high levels of Bcl-xLwere selected based on survival in the
presence of apoptotic stimuli and the mutated transgenes
characterized for novel gain of function. This study repre-
sents the first example of applying the B cell’s mutation cap-
abilities to evolve a mammalian protein in a mammalian host
by taking advantage of the cell’s natural programmed cell
death pathway in order to select for altered properties of the
Materials and methods
Maintenance of cell lines
Ramos B cells were a generous gift from the laboratory of
Dr Chi Dang (Johns Hopkins Medical School) and were
maintained in suspension in static culture in RPMI 1640
(Invitrogen, Carlsbad, CA) supplemented as instructed by
American type culture collection for Ramos cells. Flp-In
Chinese hamster ovary K1 (CHO-K1) cells (Invitrogen) were
maintained in high glucose Dulbecco’s modified Eagle
medium (Invitrogen) supplemented with 10% fetal bovine
essential amino acids (Invitrogen). CHO cell cultures were
routinely detached using Trypsin–EDTA (Invitrogen) and
passaged in fresh medium.
Creation of Ramos B cells expressing yellow fluorescent
protein (YFP) and YFP–Bcl-xL
The gene for enhanced YFP was PCR amplified from
pEYFP-C1 (Clontech, Palo Alto, CA) and cloned into
pLXSN retroviral plasmid (Clontech). Similarly, wild-type
(WT) human bcl-xL was PCR amplified and cloned into
pLXSN and the gene for YFP inserted in frame, 50of the
bcl-xLgene. When expressed, the resulting YFP and bcl-xL
genes are separated by a five amino acid linker. Expression
of the YFP or YFP–bcl-xLgene is driven by the 50viral long
terminal repeat which contains promoter/enhancer sequences.
The retroviral vectors were transfected into the Amphopack
293 (Clontech) retroviral packaging cell line and a stable
pool of packaging cells was selected in 500 mg/ml G418
(Invitrogen). Retrovirus-containing supernatants from the
packaging cells were filtered through a 0.45 mM cellulose
acetate filter (Nalgene, Rochester, NY) and used to infect
actively dividing Ramos B cells by spin infection in 24-well
plates (Becton Dickinson, Franklin Lakes, NJ) in the pres-
enceof8 mg/ml polybrene
Transfection efficiency was determined by fixing a portion of
the cells and assaying for positive YFP fluorescence using a
flow cytometer (Becton Dickinson, Mountain View, CA).
Western blot detection of Bcl-xL
Cells were lysed in RIPA buffer and the total protein concen-
tration determined using a BCA Protein Assay Kit (Pierce,
Rockford, IL). The indicated amount of total protein was run
on a 4–20% Tris-glycine gel (Invitrogen) and subjected to
western blot analysis according to Majors et al. (2008).
Measurement of caspase-3 activity
Caspase-3 activity of the B cells was measured using the
EnzChek Caspase-3 Assay Kit (Invitrogen) according to the
Induction of apoptosis in B cells expressing YFP or
Stable pools of B cells expressing YFP or YFP–Bcl-xLwere
seeded at 1 ? 107viable cells/ml in triplicate in either fresh
medium or fresh medium containing 1 mM staurosporine
(Cost et al., 2010) (Fisher Scientific, Fair Lawn, NJ). Three
aliquots of 5 ? 106cells were removed at the indicated times
for each condition. Cells were pelleted by centrifugation,
washed once with PBS and frozen at 2808C until the level
of caspase-3 activity was determined using the EnzChek
Caspase-3 Assay Kit as stated above.
Protein evolution by staurosporine apoptosis induction
For rounds of selection by staurosporine-induced apoptosis,
pools of Ramos B cells stably expressing YFP or YFP–
Bcl-xL were expanded to ?10 ? 107total viable cells.
B.S.Majors et al.
Staurosporine was added at 1 mM and the viable cell number
determined using the trypan blue exclusion method on the
indicated days. Once the total viable cell number decreased
to less than 4 ? 107cells, the cells were pelleted by centrifu-
gation, washed once in PBS and resuspended in fresh culture
medium lacking staurosporine for recovery of the culture.
During recovery, the culture medium was exchanged as ne-
cessary until the total number of viable cells again attained
approximately 10 ? 107, at which time an aliquot of the
culture was frozen in cryovials prior to beginning the next
round of staurosporine-induced apoptosis.
Characterization of mutations in B cells expressing
YFP and YFP–Bcl-xL
Genomic DNA from 5 ? 106viable Ramos B cells expres-
sing either YFP or YFP–Bcl-xL was isolated using the
DNeasy Kit (Qiagen, Valencia, CA) and used as a template
for high-fidelity PCR of the YFP or YFP–bcl-xLgene using
primers specific to the pLXSN retroviral vector and VENT
DNA polymerase (New England Biolabs, Ipswich, MA). The
primers contained flanking sequences that allowed for
cloning of the PCR product directly into the Gateway vector
pDONR221 vector using the Gateway system enzymes
(Invitrogen) and transformed into Subcloning EfficiencyTM
DH5aTMCompetent Cells (Invitrogen). Plasmid DNA from
individual bacterial colonies containing the bcl-xLgene was
sequenced on both strands to confirm mutations.
Creation of stable Bcl-xLand mutant Bcl-xLCHO cell lines
The genes for Bcl-xLWT and the Bcl-xLD29N mutant were
PCR amplified from the appropriate template and inserted
into the pcDNA5/FRT vector (Invitrogen) using the EcoRV
Fig. 1. YFP–Bcl-xLnucleotide and amino acid sequence. The region of interest is shown. The nucleotides within the 50UTR are shown in lower case, the
nucleotides and amino acids corresponding to the YFP gene are denoted by the black arrow, and the nucleotides and amino acids corresponding to the bcl-xL
gene are denoted by the gray arrow.
Directed evolution of mammalian anti-apoptosis proteins
and XhoI sites. Constructs were cotransfected with the
pOG44 vector (Invitrogen) into the CHO Flp-In cell line
(Invitrogen) using Lipofectamine 2000 reagent (Invitrogen)
according to the manufacturer’s instructions. Cells expressing
the plasmid of interest were selected in 600 mg/ml hygromy-
cin (Cellgro, Manassas, VA).
Sindbis virus infection
CHO cells were seeded in six-well plates at 1 ? 106viable
cells/well and infected with Sindbis virus according to
Figueroa et al. (2003) at a multiplicity of infection (MOI) of
10 the day after seeding. Cell viability was determined using
trypan blue exclusion and caspase-3 activity was measured
as described above.
Generation of B cells expressing YFP and YFP–Bcl-xL
Previous studies applying B cells for directed evolution uti-
lized retroviruses for insertion of the exogenous gene (Wang
et al., 2004b). Retroviral delivery offers the capability to
control gene copy number through adjustment of MOI, and
retroviruses, by nature, provide for stable integration of the
transgene into the target cell’s genomic DNA. We chose to
use a bcl-xLgene fused to that of YFP, which allows for easy
analysis of transfection efficiency and expression levels. The
sequence of the YFP–bcl-xLgene is shown in Fig. 1. To
generate Ramos B cells expressing YFP or YFP–Bcl-xL,
constructs containing the gene of interest and packaging
signal domains were transfected into a retrovirus packaging
cell line and viral supernatants were used to transduce
Ramos cells. Ramos cells transduced with either YFP or
YFP–bcl-xLshowed transduction efficiencies of 5% or less
(data not shown), as determined by YFP fluorescence using
flow cytometry. Of the 5% population that was transduced, a
Poisson distribution suggests 95% of these cells contain a
single copy of the gene of interest (Wang et al., 2004a,b).
Cell pools resistant to G418 selection were expanded, yield-
ing populations exhibiting YFP fluorescence by fluorescence
microscopy (data not shown). To determine whether the
exogenous Bcl-xL protein was expressed, Ramos cells
expressing YFP or YFP–Bcl-xLwere subjected to western
blot using an anti-Bcl-xLantibody. As shown in Fig. 2, a
band corresponding to the 54 kDa size of YFP–Bcl-xLwas
detected only in the Ramos cells infected with retrovirus
containing the YFP–Bcl-xL construct. To ensure that the
western blot band intensities reflected the abundance of
Bcl-xLprotein, equal amounts of total cellular protein were
loaded per lane.
Effect of YFP or YFP–Bcl-xLoverexpression on B cells
exposed to an apoptotic stimulus
The next step was to determine whether exogenous expres-
sion of YFP or YFP–Bcl-xLcould delay apoptosis in Ramos
cells exposed to an apoptotic stimulus. As a model apoptosis
insult, we chose to use the kinase inhibitor staurosporine.
Staurosporine has been shown to cause the rapid onset of
mitochondrial-induced apoptosis that is inhibited by the
expression of Bcl-xLin a number of mammalian cell lines
(Poppe et al., 2001; Takehara et al., 2001). To measure the
apoptosis-inducing effects of staurosporine and the ability of
Bcl-xL overexpression to delay apoptosis, we chose to
monitor the activation of caspase-3 in the B cells. Caspase-3
is a member of the caspase family of proteases involved in
apoptosis and is a crucial mediator of apoptotic cell death
(Porter and Janicke, 1999). Therefore, the level of caspase-3
activation provides a measure of apoptosis activation in the
In the absence of staurosporine, G418-resistant Ramos B
cells expressing either YFP or YFP–Bcl-xLdid not exhibit
any increase in apoptosis as measured by caspase-3 activ-
ity over the 6 h experiment (Fig. 3, no STS). In contrast,
after just 2 h of exposure to 1 mM staurosporine, Ramos
cells expressing YFP and YFP–Bcl-xLexhibited increases
in the caspase-3 activity (Fig. 3, 1 mM STS) that progres-
sively increased from 2 to 6 h. However, the increase in
caspase-3 activity was lower in the YFP–Bcl-xL cells at
the later time points (4–6 h). This reduced caspase-3 ac-
tivity is consistent with the anti-apoptotic nature of the
Fig. 2. Western blot analysis of YFP–Bcl-xLexpression in B cells. Equal
amounts of total cellular protein (25 mg) from stable pools of B cells
expressing YFP or YFP–Bcl-xLwere loaded in each lane and subjected to
SDS–PAGE and western blot analysis with an anti-Bcl-xLantibody. The
expected molecular mass of YFP–Bcl-xLis 54 kDa.
Fig. 3. Caspase-3 activity in B cells. Caspase-3 activity levels in stable
pools of B cells expressing YFP or YFP–Bcl-xLin the presence or absence
of 1 mM staurosporine (STS) were assessed over the time of exposure. The
data represent averages from three independent experiments with two
independent measurements per time point.
B.S.Majors et al.
Mutation of exogenous YFP and YFP–Bcl-xLin B cells before
selection in staurosporine
Directed evolution with B cells relies upon the effective
mutation of exogenous genes incorporated into the cellular
chromosomes. A schematic of the steps taken for the gener-
ation of the B-cell population expressing YFP and YFP–
Bcl-xLduring successive rounds of evolution is shown in
Fig. 4. First, we examined whether the B cells could mutate
exogenously inserted YFP or YFP–bcl-xLconstructs prior to
any direct apoptotic stimuli.
Following G418 selection of stable pools expressing YFP
and YFP–Bcl-xL(Fig. 4, round 0), aliquots of the cultures
were harvested and the genomic DNA extracted from the
population. To allow for efficient bacterial cloning, the inte-
grated YFP–bcl-xLtransgene was PCR amplified from the
genomic DNA of the Ramos cells using a high-fidelity DNA
polymerase, cloned into vectors, transformed into bacteria
and purified for double-strand sequencing analysis.
Shown in Table I are the results of sequencing of bacterial
isolates containing the recovered transgenes from stable
0. Seventeen of 23 isolates had at least one nucleotide muta-
tion within the sequenced region (Table I). Mutations in the
coding sequence for the YFP–Bcl-xL construct included
amino acid changes, silent mutations and premature stop
sequences. A mutation of the Bcl-xLTrp169 codon occurred
in three separate constructs, including one Bcl-xLTrp169Arg
and two Bcl-xLTrp169stop mutations. Similarly, mutation of
the Bcl-xLGln207 codon occurred in four constructs, includ-
ing two Gln207Leu and two Gln207stop changes, suggesting
a slight increase in prevalence of cells harboring these muta-
tions in culture. Sequencing of the B-cell culture expressing
YFP alone that was transduced under identical conditions
yielded a similar number of mutations after the initial selec-
tion in G418 (data not shown). These experiments demon-
strated that native somatic hypermutation activity of the B
cells introduced mutations into the exogenous genes even in
the absence of selection for YFP or Bcl-xLactivity.
Evaluation of Ramos B cells expressing YFP or YFP–Bcl-xL
after multiple rounds of selection in staurosporine
B cells expressing YFP or YFP–Bcl-xLwere then subjected
to successive rounds of apoptosis induction by exposure to
staurosporine followed by recovery in staurosporine-free
media with the intent of generating cells with increased
resistance to apoptosis. B cells were expanded to approxi-
mately 10 ? 107viable cells in culture and treated with 1 mM
staurosporine to begin the first round of selection and recov-
ery. When the number of viable cells declined to 4 ? 107, the
staurosporine was removed from this culture to limit further
cell death. Cells were maintained in fresh medium until the
number of viable cells again returned to 10 ? 107at which
time the next round of selection by staurosporine would
begin. This process of selection and recovery was repeated
multiple times. During the third round of exposure to stauros-
porine the total number of viable cells declined for both YFP
and YFP–Bcl-xLcells as shown in Fig. 5. For Ramos cells
expressing the YFP construct, the total viable cell number
began to decline rapidly after staurosporine treatment was
initiated at day 0. After 4 days of staurosporine exposure, when
the number of viable cells declined to 4 ? 107, the staurospor-
ine was removed. In contrast, Ramos cells expressing YFP–
Bcl-xLshowed a more gradual decline in viable cell number in
response to staurosporine exposure, and the staurosporine was
removed only after 6 days of exposure to allow for recovery of
the pool. Similar death profiles were seen during the initial
rounds (one and two) as well as in the following rounds of se-
lection in staurosporine for these cell lines. The Ramos cells
expressing YFP did not recover following round 3 of selection
and thus only the Ramos cells expressing YFP–Bcl-xLwere
subsequently maintained and subjected to additional rounds of
staurosporine selection followed by recovery.
Next, we wanted to assess the apoptosis resistance of B
cells expressing YFP–Bcl-xL generated from successive
rounds of apoptosis induction by staurosporine followed by
recovery in fresh medium. To do so, we compared the level
of caspase-3 activity in response to staurosporine exposure in
B cells pools generated from 0, 2, 4 and 6 successive rounds
of selection and recovery (Fig. 6). Aliquots of cells from the
end of each round of recovery were frozen and later cultured
and tested in parallel for this experiment. In this experiment,
the level of caspase-3 activity was measured after subtracting
the basal level of caspase-3 signal in cells not exposed to
staurosporine. For B cells expressing YFP from round 0, the
caspase-3 activity level increased after 2 h of exposure to
staurosporine. When the B cells expressing YFP–Bcl-xL
Fig. 4. Workflow for generating B cells expressing YPF and YFP–Bcl-xLfollowed by rounds of selection in staurosporine and recovery.
Directed evolution of mammalian anti-apoptosis proteins
Table I. Mutations in the YFP–Bcl-xLgene region of transduced Ramos B cells at round 0
ConstructComment50UTR (nt. 1–
YFP (nt. 104–820)Bcl-xL(nt. 836–1537)
17 Deletion at nt. C20 T38CG1016A
18 No mutations
19 No mutations
20 No mutations
22 No mutations
23 No mutations
The YFP–Bcl-xLgene from genomic DNA of a pool of Ramos B cells from round 0 was PCR amplified, cloned and sequenced. The sequencing data for 23 different constructs isolated from bacterial colonies are
shown. The location and nature of the nucleotide (nt.) mutation is shown as well as the resulting amino acid changes (amino acid location refers to residue within the YFP or Bcl-xLprotein). Silent (sil) mutations
are also shown.
B.S.Majors et al.
obtained from rounds of selection and recovery were exposed
to staurosporine in this experiment, all cultures exhibited
lower increases in caspase-3 activity compared to the YFP
cells from round 0. Furthermore, Ramos cells expressing
YFP–Bcl-xLfrom round 2 showed lower caspase-3 activity
increases compared to cells from round 0 and cells from
round 4 showed lower caspase-3 activity levels compared
to cells from round 2. In fact, the Ramos B expressing
YFP–Bcl-xLfrom rounds 4 and 6 showed almost no increase
in caspase-3 activity after staurosporine exposure compared
to the same cells in the absence of staurosporine.
One possible explanation for the increased survival during
repeated rounds of staurosporine selection may be that subse-
quent rounds selected Ramos cells with higher levels of
Bcl-xLexpression. To examine the relative level of Bcl-xL
expression in the Ramos B cells over an increasing number
of selection rounds, the expression level of YFP–Bcl-xLin
lysates from Ramos cells from rounds 0, 2, 4 and 6 was
determined by western blot (Fig. 7). There was no detection
of endogenous Bcl-xLin the Ramos cells expressing YFP
only. Ramos cells from various rounds of treatment expres-
sing the YFP–Bcl-xL gene showed detectable levels of
Bcl-xL, and the relative levels of Bcl-xL protein in the
Ramos B-cell population increased from round 0 to round 2
and round 2 to round 4, suggesting that the selection process
enriched for cells exhibiting increased expression of Bcl-xL.
To ensure that the western blot band intensities reflected the
abundance of Bcl-xLprotein, equal amounts of total cellular
protein were loaded per lane.
Mutation of exogenous YFP–Bcl-xLin B cells following
rounds of selection in staurosporine
Another possible reason for the increased survival and
reduced caspase-3 activity was the presence of an altered
Bcl-xLprotein. The sequence of the YFP–bcl-xLconstruct
was analyzed from the lysates of cells following rounds 3
and 6 of the selection and recovery experiment. The results
of the sequence analysis for Ramos cells harvested following
round 3 of selection and recovery are shown in Table II.
These constructs exhibited multiple mutations in both the
YFP and Bcl-xLcoding sequence and in multiple colonies
(Table II). If a gain-of-function mutant of the Bcl-xLprotein
were to evolve within the culture, the construct would be
present in the pool of genomic DNA from the selected B
cells and may exhibit an increased predominance afforded by
the survival phenotype. Interestingly, in the Bcl-xLcoding
sequence, 8 of 23 constructs had a conserved amino acid
change at the Bcl-xLAsp29 residue (Table II). This amino
acid was changed to Asn in seven bacterial isolates (as a
result of a G!A transition) and to His in one instance (as a
result of a G!C transversion). A variety of other mutations
in both the YFP coding sequence and Bcl-xLsequence were
also found in cells harboring the Bcl-xL Asp29 mutation.
Therefore, at least five of the Bcl-xLAsn29 variants were
separate isolates because five out of the eight included a
variety of different mutations at other points in the sequence.
Aside from the Bcl-xLAsp29 change, one other mutation,
Bcl-xL Gly117Glu, appeared in two separate Bcl-xL con-
structs. However, this mutation was found in only 2 of 23
isolates and was thus much less frequently observed than the
Bcl-xL Asp29 change. In addition, the Asp29 amino acid
was the only residue that exhibited a transformation to two
Fig. 5. Total viable cell number (TVCN) profiles from Ramos B cells
expressing YFP or YFP–Bcl-xLduring and after exposure to staurosporine
in round 3. The TVCN was plotted against the duration of the culture.
Staurosporine (1 mM) was added to the cell cultures on day 0 and removed
as indicated by the arrows when the cultures declined to 4 ? 107total viable
Fig. 6. Caspase-3 activation in response to staurosporine in recovered
Ramos B cells expressing YFP (round 0) and YFP–Bcl-xLafter rounds of
staurosporine exposure and recovery. Recovered cells from the indicated
round of selection (R) were exposed to 1 mM staurosporine for 2 h and the
caspase-3 activity was measured. The basal caspase-3 activity for cells in the
absence of staurosporine was subtracted from the caspase-3 activity levels
for cells in the presence of staurosporine.
Fig. 7. Western blot expression of Bcl-xLin Ramos B cells expressing the
YFP or YFP–Bcl-xLprotein. Equal amounts of total cellular protein (50 mg)
from Ramos B cells from the indicated round (R) of the selection process
were loaded in each lane to ensure that band intensities reflect actual relative
expression levels in the cells. Samples were subjected to SDS–PAGE and
western blot analysis with an anti-Bcl-xLantibody.
Directed evolution of mammalian anti-apoptosis proteins
Table II. Mutations in the YFP–Bcl-xLgene region of transduced Ramos B cells following round 3
YFP (nt. 104–820)Bcl-xL(nt. 836–1537)
17 Deletions of nts.
G16 and C21
18Deletions of nts.
A9 and C21
19 No mutations
20 No mutations
22 No mutations
23 No mutations
The YFP–Bcl-xLgene from genomic DNA of a pool of Ramos B cells was PCR amplified, cloned and sequenced. The sequencing data for 23 different constructs isolated from bacterial colonies are shown. The
location and nature of the nucleotide (nt.) mutation is shown as well as the resulting amino acid changes (amino acid location refers to residue within the YFP or Bcl-xLprotein). Silent (sil) mutations are also
B.S.Majors et al.
different amino acids as observed in the sequenced isolates.
From round 0 (Table I) to round 3 (Table II), two silent
mutations and one nonsense mutation were maintained in the
YFP–Bcl-xLcoding region. Of the 23 colonies with sequen-
cing data, 5 did not show any detectable mutations in the
regions) and 3 had mutations in the YFP coding region only.
Other genomic mutations outside the sequenced region may
have allowed these cells to survive three rounds of selection
andrecovery. Surprisingly, sequencing
genomic DNA constructs from round 6 of the selection and
recovery experiment yielded no colonies with the Asp29 mu-
tation (data not shown). Additionally, 12 of 19 isolates from
round 6 were identified with no mutations in the bcl-xL
sequence (including UTR
of the pooled
Effect of Bcl-xLmutant on cell survival for a model Sindbis
To assess the extent that the predominant Bcl-xLAsp29Asn
(D29N) mutation altered the properties of the protein, mam-
malian cells lines expressing the Bcl-xL WT protein, the
Bcl-xLD29N mutant protein or the Null vector were created.
Previously, expression of recombinant WT anti-apoptosis
genes including Bcl-xLhas been shown to inhibit cell death
in CHO cells (Figueroa et al., 2003; Chiang and Sisk, 2005).
Thus, CHO cells in combination with the Flp-In integration
system, which provides for integration into a single, identical
chromosomal site, were used in this study. The use of the
Flp-In system helps to eliminate genomic location effects to
ensure that stable selection of the Bcl-xLWT and Bcl-xL
D29N cells resulted in comparable levels of the Bcl-xL
protein. We determined the relative level of Bcl-xLexpres-
sion by western blot (Fig. 8). While a low level of endogen-
ous Bcl-xLwas detected in the CHO cells expressing the
Null vector, Bcl-xLwas detected at much higher levels in
CHO cells transfected with Bcl-xLor Bcl-xLD29N. Within
the limits of quantification possible using western blot, both
Bcl-xL-expressing cell lines showed the presence of similar,
high levels of stable Bcl-xLprotein. Equal amounts of total
cellular protein were loaded per lane during the western blot
so that relative band intensities reflected actual protein
To determine the effects of Bcl-xLor Bcl-xLmutant over-
expression on apoptosis in CHO cells, we next exposed the
CHO cells to a staurosporine apoptosis insult. While stauros-
porine was able to induce apoptosis rapidly in Ramos cells
expressing YFP–Bcl-xL(Fig. 3) and cell death in CHO cells
containing the null vector, CHO cells expressing Bcl-xLWT
and CHO Bcl-xLD29N exhibited a much reduced rate of
cell death in the presence of staurosporine (data not shown).
Therefore Sindbis virus infection was considered as a more
potent apoptosis inducer as it has been shown to cause rapid
apoptosis (Mastrangelo et al., 1999; Figueroa et al., 2001).
Furthermore, Sindbis virus infection is also known to cause
apoptosis through a mitochondrial (Moriishi et al., 2002) and
caspase-dependent (Nava et al., 1998) pathway.
Therefore, CHO cells expressing the Null, Bcl-xLWT and
Bcl-xLD29N constructs were seeded at identical concentra-
tions in six-well plates and exposed to Sindbis virus at a
MOI of 10 to ensure that all cells were infected. After 24 h
of Sindbis virus infection, the viabilities of the cultures were
determined by the trypan blue exclusion method (Fig. 9).
The null cell line showed less than 15% viability 24 h after
infection, whereas CHO cells expressing either Bcl-xLcon-
struct showed much higher viabilities. Specifically, CHO
Bcl-xLWT cells were 40% viable after 24 h of infection.
However, CHO cells expressing the Bcl-xL D29N protein
showed average viabilities that were even higher (at 55%).
These experiments were repeated multiple times, with the
CHO Bcl-xLD29N cell line exhibiting statistically-significant
increases in viability relative to both Null and CHO Bcl-xL
WT according to the Student’s t-test criterion in all cases.
Directed evolution has already proven to be a powerful tool
for the generation of novel polypeptides in a variety of appli-
cations. Such experiments typically employ bacterial or yeast
systems to evolve proteins with easily identifiable traits.
Ideally, it would be useful to have a mammalian system to
evolve mammalian proteins, especially those in which modi-
fications or functional selection depends on the presence of a
mammalian host. Emerging higher eukaryotic directed evolu-
tion systems that employ self-mutating B-cell lines such as
Fig. 8. Western blot detection of Bcl-xLfrom Flp-In CHO cells. Flp-In
CHO cells were transfected with either the Bcl-xLWT gene, the Bcl-xL
D29N gene or the null vector and lysates from stable pools were subjected
to SDS–PAGE and western blot analysis with an anti-Bcl-xL antibody.
Equal amounts of total cellular protein (50 mg) were loaded per lane to
ensure that band intensities reflected the relative expression levels of Bcl-xL.
The expected molecular weight of Bcl-xL(WT and D29N) is 25 kDa.
Fig. 9. Viability of CHO cells following exposure to Sindbis virus. Flp-In
CHO cells stably expressing Bcl-xLWT, Bcl-xLD29N or the null vector
were exposed to Sindbis virus at an MOI of 10. The data represent averages
of two independent experiments with two counts per experiment at 24-h
post-infection. The higher viability of the Bcl-xL D29N cells was
determined to be significantly higher than that of the Bcl-xL WT cells
(Student’s t-test P, 0.001).
Directed evolution of mammalian anti-apoptosis proteins
Ramos, 18–81 (Wang et al., 2004a) and DT-40 (Arakawa
et al., 2008; Arakawa and Buerstedde, 2009) have recently
demonstrated their capacity to evolve protein variants
(Blagodatski and Katanaev, 2011). Yet, demonstration of this
technology has been limited primarily to evolution of fluor-
escent proteins for ease of screening. Here, we describe, to
our knowledge, the first successful directed evolution
experiment of an exogenously inserted but naturally occur-
ring mammalian protein in a mammalian protein evolution
system. This work shows that the ability of using hypermu-
tating B-cell lines for mammalian protein evolution is pos-
sible for proteins using screening criteria other than
fluorescence. Such a study should provide the impetus to use
mammalian evolution systems in other biotechnology and
We first generated Ramos B-cell populations expressing
either YFP or YFP–Bcl-xL. Unlike previous mammalian
evolution experiments (Wang et al., 2004b; Arakawa et al.,
2008), our fluorescent proteins were simply used to deter-
mine the transfection efficiency in our system and not used
for selection. The efficiency of retrovirus transduction of
Ramos cells was 5% for the constructs in this study, suggest-
ing the difficult nature of gene delivery to B cells as seen in
previous reports (Wang et al., 2004b). Viral gene delivery is
reported to follow a Poisson distribution, meaning that at 5%
transduction efficiency, one can assume that 95% of the cells
transduced are the result of a single virion introducing a
single copy of the gene of interest (Wang et al., 2004a). This
is important in our study, since a cell harboring multiple
genes of interest may lead to higher expression levels and
multiple, different mutant proteins in a single cell. Such
effects would likely compromise the cell-versus-cell com-
petitive nature necessary for success in these evolution
After selection of Ramos B cells harboring the YFP or
YFP–Bcl-xLgenes, each was sequenced prior to applying
apoptotic selective pressure. In the case of YFP–Bcl-xL,
many mutations were found in the coding sequence, includ-
ing three that were found in multiple constructs, even prior
to staurosporine exposure when there was no selective pres-
sure other than G418. Thus, these mutations are most likely
the result of random mutagenesis by the hypermutation
machinery. Most importantly, it shows that the Ramos B
cells are able to introduce mutations in the exogenous genes
during the culture process.
The transduced B cells stably expressing the YFP–Bcl-xL
gene showed enhanced tolerance to the kinase inhibitor and
apoptosis inducer staurosporine compared to cells expressing
YFP only. This was expected since Bcl-xL expression has
previously been shown to inhibit apoptosis in a Ramos B
cell (Alam et al., 1997). Furthermore, our results confirm that
the YFP fusion did not inhibit Bcl-xLfunction (Chu et al.,
2004). After staurosporine exposure, the number of viable
Ramos cells expressing only the YFP protein declined
precipitously compared to YFP–Bcl-xL-expressing Ramos
cells, which showed a slower decline. The cells continued to
decrease in total viable number even after staurosporine was
removed from the culture which may indicate either incom-
plete removal of the chemical or that the apoptosis cascade
had already been activated in some cells.
After three rounds of staurosporine selection and cell re-
covery, the genomic DNA was harvested from YFP and
YFP–Bcl-xL cells, PCR amplified with high-fidelity poly-
merase and multiple clones were sequenced. Numerous
mutations were observed throughout the sequenced construct.
The mutations in the Bcl-xLcoding sequence are of particu-
lar interest. It is quite interesting that 8 of 23 sequenced con-
structs at round 3 had a mutation at Asp29. This can arise
from either multiple independent mutations of the codon for
Asp29 in multiple cells or from the survival or growth ad-
vantage of a subset of cells containing the mutant gene over-
taking the culture. Interestingly, seven of the eight constructs
showing the Asp29 mutation were a result of a G to A transi-
tion resulting in an Asn residue, whereas one of the eight
constructs had an atypical G to C transversion that resulted in
a His residue. A silent Leu174 mutation in the Bcl-xLcoding
sequence occurred in three Asp29-containing constructs, a
Ser206Phe mutation was present in the YFP coding sequence
for four Asp29-containing constructs, and another variant
included an Ala73Thr mutation in the YFP coding region.
These changes indicate that the Asp29 mutation arose multiple
times in culture leading to perhaps five separate populations.
Subsequently, we tested the anti-apoptotic potential of the
predominant mutation, Bcl-xLAsp29Asn (or D29N), found
after three rounds in our studies. We chose to test the effects
of Bcl-xL and Bcl-xL D29N expression in a CHO cell
system that enables recombinant protein expression from a
single, common genomic locus. CHO cells overexpressing
the Bcl-xLD29N mutation showed higher viabilities in mul-
tiple experiments following Sindbis virus infection. Thus, we
confirmed, in a non-hypermutating and commercially rele-
vant cell line, the improved anti-apoptotic effects of the
The reason that the D29N mutation is able to impart
improved anti-apoptosis protection may be multifaceted.
Most strikingly, aspartic acid residues are the known sub-
strate for caspase-dependent cleavage during apoptosis sig-
naling. Furthermore, the Asp29 residue is located within the
unstructured loop domain of Bcl-xL, a region where two
other well-characterized caspase cleavage residues, Asp61
and Asp76, are located (Clem et al., 1998). Interestingly,
Bcl-2, an anti-apoptotic member of the Bcl-2 family of
apoptosis-related proteins, exhibits significant homology to
Bcl-xLand has a caspase cleavage residue in the vicinity of
the Bcl-xL Asp29 location. Specifically, Bcl-2 has been
shown to be cleaved by caspases at Asp34, and mutagenesis
of this cleavage site abolished caspase cleavage and
increased protection to apoptosis in cells expressing the
mutant protein (Cheng et al., 1997). Surprisingly, the muta-
tions found in the Bcl-xLgene failed to change amino acids
at the other known Bcl-xLcleavage sites Asp61 and Asp76.
When there are multiple known cleavage locations in a
protein, the cleavage site acted upon by caspases will vary
based on the apoptotic insult (Marie Hardwick, personal
communication) and possibly the particular cell lines used.
It is possible that the D29N mutation in Bcl-xLmay result
in other modifications or interactions that alter Bcl-xLfunc-
tion. Bcl-xLinteracts with the mitochondria (Vander Heiden
et al., 1997; Shimizu et al., 2000), endoplasmic reticulum
(White et al., 2005; Jeong et al., 2004) and with a multitude
of other proteins both with direct apoptosis and non-
apoptosis functions in the different cellular compartments.
It was also observed that the expression of Bcl-xL
increased at later rounds of selection. Indeed, it has been
B.S.Majors et al.
shown that anti-apoptosis proteins exhibit dose-dependent
behavior in which their anti-apoptotic properties increase
with the levels of these proteins within the cells (Majors
et al., 2009a). The increased stability brought about by
removing a caspase cleavage site such as D29 may increase
the levels of the full length protein. In addition to the D29N
mutation, a number of additional mutations were also detected
in the Bcl-xL, YFP and 50and 30UTR domains, including a
Gly177Glu mutation in two separate isolates. These mutations
as well as others outside the sequenced region may serve to
increaseeither the stability
Modifications that increase expression levels or protein stabil-
ity will act in concert with those modifications that increase
the function of the Bcl-xLprotein and represent two compli-
mentary evolving traits in the hypermutating B cells.
Ramos cell populations expressing YFP–Bcl-xLharvested
after six rounds of staurosporine exposure and recovery no
longer showed the widespread mutation of the Asp29
residue. Indeed, a majority of the cells even at round 3 of
selection did not contain the Asp29 mutation. Prolonged
stress using a B-cell type capable of hypermutating its own
genomic DNA (Parsa et al., 2007) may result in a number of
outcomes that can overcome staurosporine exposure, such as
upregulation of pumps capable of removing the toxin or mu-
tation of the kinases affected by staurosporine treatment.
Furthermore, the apoptotic machinery of these cells may be
rendered non-functional by other mutations, thus conferring
apoptosis resistance superior to any mutation in the Bcl-xL
protein. These particular changes may be superior to the
increased protection afforded by the widespread Asp29 muta-
tion in round 3. What the data suggest is that distinct subpo-
pulations of Ramos can evolve even over a few rounds, and
fewer rounds of B-cell evolution may perhaps be more suit-
able for identification of novel, non-fluorescent proteins
while the phenotype imparted by overexpression of the trans-
gene is most prominent. Extended culture or selection in
hypermutating cells may allow for non-identifiable mutations
that circumvent the overexpressed transgene. For these cases,
systems that allow for inducible hypermutation ability (Faili
et al., 2002) may provide a more flexible alternative to the
constitutive hypermutating systems used in this study.
This study has demonstrated the ability of the Ramos
B-cell line to induce mutations in exogenous non-fluorescent
genes. Using rounds of selection and recovery in the pres-
ence of an apoptotic insult, we were able to isolate a mutant
of Bcl-xLwith improved anti-apoptotic ability. This Ramos
B-cell mutagenesis system relies upon the mutagenesis prop-
erties of AID which may have preferential DNA targets and
limited nucleotide changes (Seki et al., 2005). Future studies
could employ the use of other hypermutating systems to gen-
erate greater diversity. For example, the DT40 chicken B-cell
line can introduce genetic mutations by both somatic hyper-
mutation and gene conversion. Furthermore, as more infor-
mation is gained about the mutation capacity of B-cell lines,
vectors targeting a region of increased hypermutation (such
as the IgV locus in Ramos cells) can be utilized for faster
mutagenesis. Regardless, the main factor to consider in all
directed evolution experiments is identifying a selectable
phenotype based on transgene overexpression. Mammalian
proteins that can link functional changes to an appropriate se-
lectable readout will likely provide excellent opportunities for
directed evolution in mammalian cells in the coming years.
The authors would like to thank the contributions of Christopher Tonkin
from the Biogen Idec sequencing group, Marie Hardwick from the Johns
Hopkins School of Public Health and George Oyler. M.J.B.’s contribution
was supported in part from grant 1RO1GM095685 from NIGMS.
Aharoni,A., Gaidukov,L., Yagur,S., Toker,L., Silman,I. and Tawfik,D.S.
(2004) Proc. Natl Acad. Sci. USA, 101, 482–487. First published on
2003/12/30, doi: 10.1073/pnas.2536901100; 2536901100 [pii].
Alam,M.K., Davison,S., Siddiqui,N., Norton,J.D. and Murphy,J.J. (1997)
Eur. J. Immunol., 27, 3485–3491. First published on 1998/02/17, doi:
Arakawa,H. and Buerstedde,J.M. (2009) Philos. Trans. R. Soc. Lond. B Biol.
95780M5276381811 [pii]; 10.1098/rstb.2008.0202.
Arakawa,H., Kudo,H., Batrak,V., Caldwell,R.B., Rieger,M.A., Ellwart,J.W.
and Buerstedde,J.M. (2008) Nucleic Acids Res., 36, e1. First published on
2007/12/13, doi: gkm616 [pii]; 10.1093/nar/gkm616.
Bachl,J., Carlson,C., Gray-Schopfer,V., Dessing,M. and Olsson,C. (2001)
J. Immunol., 166, 5051–5057. First published on 2001/04/06.
1207–1214.First published on
Ueno,K. and Hardwick,J.M. (1997) Science, 278, 1966–1968. First pub-
lished on 1998/01/07.
Chiang,G.G. and Sisk,W.P. (2005) Biotechnol. Bioeng., 91, 779–792. First
published on 2005/06/30, doi: 10.1002/bit.20551.
Chu,F., Borthakur,A.,Sun,X., Barkinge,J.,
Prasad,K.V. (2004) Apoptosis, 9, 83–95. First published on 2004/01/24,
doi: 10.1023/B:APPT.0000012125.01799.4c; 5256690 [pii].
Clem,R.J., Cheng,E.H., Karp,C.L., Kirsch,D.G., Ueno,K., Takahashi,A.,
Kastan,M.B., Griffin,D.E., Earnshaw,W.C., Veliuona,M.A. et al. (1998)
Proc. Natl. Acad. Sci. USA, 95, 554–559. First published on 1998/01/22.
Collingwood,T.N., Snowden,A. and Gregory,P.D. (2010) Biotechnol.
Bioeng., 105, 330–340. First published on 2009/09/25, doi: 10.1002/
Batista,F.D., Sale,J.E. and Neuberger,M.S. (2002) Nat. Biotechnol., 20,
1129–1134. First published on 2002/10/16, doi: 10.1038/nbt752; nbt752
Weill,J.C. (2002) Nature, 419, 944–947. First published on 2002/11/01,
doi: 10.1038/nature01117; nature01117 [pii].
Figueroa,B., Jr, Sauerwald,T.M., Mastrangelo,A.J., Hardwick,J.M. and
Betenbaugh,M.J. (2001) Biotechnol. Bioeng., 73, 211–222. First pub-
lished on 2001/03/21, doi: 10.1002/bit.1053 [pii].
Betenbaugh,M.J. (2003) Metab. Eng., 5, 230–245. First published on
2003/12/04, doi: S1096717603000442 [pii].
Figueroa,B., Jr, Chen,S., Oyler,G.A., Hardwick,J.M. and Betenbaugh,M.J.
(2004) Biotechnol. Bioeng., 85, 589–600. First published on.
Guntas,G., Mansell,T.J., Kim,J.R. and Ostermeier,M. (2005) Proc. Natl
Acad. Sci. USA, 102, 11224–11229. First published on 2005/08/03, doi:
0502673102 [pii]; 10.1073/pnas.0502673102.
Jeong,S.Y., Gaume,B., Lee,Y.J., Hsu,Y.T., Ryu,S.W., Yoon,S.H. and
Youle,R.J. (2004) EMBO J., 23, 2146–2155. First published on 2004/05/
08, doi: 10.1038/sj.emboj.7600225; 7600225 [pii].
Kanayama,N., Todo,K., Takahashi,S., Magari,M. and Ohmori,H. (2006)
Nucleic Acids Res., 34, e10. First published on 2006/01/20, doi: 34/2/e10
Kim,Y.G., Kim,J.Y., Mohan,C. and Lee,G.M. (2009) Biotechnol. Bioeng.,
103, 757–766. First published on 2009/03/07, doi: 10.1002/bit.22298.
Kauffmann,I. and Schmidt-Dannert,C. (2001) Pro. Eng., 14, 919–928.
Kuchner,O. and Arnold,F.H. (1997) Trends Biotechnol., 15, 523–530. First
published on 1998/01/07, doi: S0167-7799(97)01138-4 [pii]; 10.1016/
Kumar,S., Chen,C.S., Waxman,D.J. and Halpert,J.R. (2005) J. Biol. Chem.,
280, 19569–19575. First published on 2005/03/19, doi: M500158200
Lehmann,M. and Wyss,M. (2001) Curr. Opin. Biotechnol., 12, 371–375.
First published on 2001/09/12, doi: S0958-1669(00)00229-9 [pii].
publishedon 2008/11/15, doi:
Directed evolution of mammalian anti-apoptosis proteins
Liebeton,K., Zonta,A., Schimossek,K., Nardini,M., Lang,D., Dijkstra,B.W.,
Reetz,M.T. and Jaeger,K.E. (2000) Chem. Biol., 7, 709–718. First pub-
lished on 2000/09/12, doi: S1074-5521(00)00015-6 [pii].
Majors,B.S., Betenbaugh,M.J. and Chiang,G.G. (2007) Metab. Eng., 9,
317–326. First published on 2007/07/06, doi: S1096-7176(07)00030-4
Majors,B.S., Betenbaugh,M.J., Pederson,N.E. and Chiang,G.G. (2008)
Biotechnol. Bioeng., 101, 567–578. First published on 2008/08/30, doi:
Majors,B.S., Betenbaugh,M.J., Pederson,N.E. and Chiang,G.G. (2009a)
Biotechnol. Prog., 25, 1161–1168. First published on 2009/06/25, doi:
Majors,B.S., Chiang,G.G. and Betenbaugh,M.J. (2009b) Mol. Biotechnol.,
42, 216–223.First published
Mastrangelo,A.J., Zou,S., Hardwick,J.M. and Betenbaugh,M.J. (1999)
Biotechnol. Bioeng., 65, 298–305. First published on 1999/09/15, doi:
Meents,H., Enenkel,B., Eppenberger,H.M., Werner,R.G. and Fussenegger,M.
(2002) Biotechnol. Bioeng., 80, 706–716. First published on 2002/10/16,
Miyazaki,K., Wintrode,P.L., Grayling,R.A., Rubingh,D.N. and Arnold,F.H.
(2000) J. Mol. Biol., 297, 1015–1026. First published on 2000/03/29, doi:
10.1006/jmbi.2000.3612; S0022-2836(00)93612-X [pii].
Moriishi,K., Koura,M. and Matsuura,Y. (2002) Virology, 292, 258–271.
First published on 2002/03/07, doi: 10.1006/viro.2001.1206; S0042682
Hardwick,J.M. (1998) J. Virol., 72, 452–459. First published on 1998/01/
Ostermeier,M. (2005) Protein Eng. Des. Sel., 18, 359–364. First published
on 2005/07/27, doi: gzi048 [pii]; 10.1093/protein/gzi048.
Martin,A. (2007) Mol. Immunol., 44, 567–575. First published on 2006/
03/18, doi: S0161-5890(06)00031-9 [pii]; 10.1016/j.molimm.2006.02.003.
Perez-Duran,P., de Yebenes,V.G. and Ramiro,A.R. (2007) Carcinogenesis,
28, 2427–2433. First published on 2007/09/07, doi: bgm201 [pii];
Poppe,M., Reimertz,C., Dussmann,H.,
J. Neurosci., 21, 4551–4563. First published on 2001/06/28.
Porter,A.G. and Janicke,R.U. (1999) Cell Death Differ., 6, 99–104. First
published on 1999/04/14, doi: 10.1038/sj.cdd.4400476.
Seki,M., Gearhart,P.J. and Wood,R.D. (2005) EMBO Rep., 6, 1143–1148.
Firstpublished on 2005/12/02,
Shimizu,S., Shinohara,Y. andTsujimoto,Y.
4309–4318. First published on 2000/09/12, doi: 10.1038/sj.onc.1203788.
Takehara,T., Liu,X., Fujimoto,J., Friedman,S.L. and Takahashi,H. (2001)
S0270-9139(01)57355-3 [pii]; 10.1053/jhep.2001.25387.
Vander Heiden,M.G., Chandel,N.S., Williamson,E.K., Schumacker,P.T. and
Thompson,C.B. (1997) Cell, 91, 627–637. First published on 1997/12/11,
doi: S0092-8674(00)80450-X [pii].
Wang,C.L., Yang,D.C. and Wabl,M. (2004a) Protein Eng. Des. Sel., 17,
659–664. First published on 2004/11/02, doi: gzh080 [pii]; 10.1093/
Wang,L., Jackson,W.C., Steinbach,P.A. and Tsien,R.Y. (2004b) Proc. Natl
Acad. Sci. USA, 101, 16745–16749. First published on 2004/11/24, doi:
0407752101 [pii]; 10.1073/pnas.0407752101.
Wang,L. and Tsien,R.Y. (2006) Nat. Protoc., 1, 1346–1350. First published
on 2007/04/05, doi: nprot.2006.243 [pii]; 10.1038/nprot.2006.243.
White,C., Li,C., Yang,J., Petrenko,N.B., Madesh,M., Thompson,C.B. and
Foskett,J.K. (2005) Nat. Cell Biol., 7, 1021–1028. First published on
2005/09/24, doi: ncb1302 [pii]; 10.1038/ncb1302.
Zacharias,D.A. and Tsien,R.Y. (2006) Methods Biochem. Anal., 47, 83–120.
First published on 2005/12/13.
doi: 7400582 [pii];10.1038/
publishedon 2001/06/30, doi:
B.S.Majors et al.