Protein Engineering, Design & Selection vol. 25 no. 11 pp. 699–703, 2012
Published online October 18, 2012 doi:10.1093/protein/gzs074
A highly stable protein chimera built from fragments
of different folds
Sooruban Shanmugaratnam†, Simone Eisenbeis† and
Birte Höcker1
Max Planck Institute for Developmental Biology, Spemannstr. 35, 72076
Tübingen, Germany
1
To whom correspondence should be addressed.
E-mail: [email protected]
Received June 29, 2012; revised August 24, 2012;
accepted September 10, 2012
Edited by Vladimir Tishkov
Proteins increased in complexity during the course of
evolution. Domains as well as subdomain-sized fragments
were recruited and adapted to form new proteins and
novel folds. This concept can be used in engineering to
construct new proteins. We previously reported the combination of fragments from two ancient protein folds, a
flavodoxin-like and a (ba)8-barrel protein. Here we
report two further attempts at engineering a chimeric
protein from fragments of these folds. While one of the
constructs showed a high tendency to aggregate, the other
turned out to be a highly stable, well-structured protein.
In terms of stability against heat and chemical denaturation this chimera, named NarLHisF, is superior to the
earlier presented CheYHisF. This is the second instance
of a chimera build from two different protein folds,
which demonstrates how easily recombination can lead to
the development and diversification of new proteins – a
mechanism that most likely occurred frequently in the
course of evolution. Based on the results of the failed and
the successful chimera, we discuss important considerations for a general design strategy for fold chimeras.
Keywords: (ba)8-barrel/chimera/flavodoxin-like/hybrid
proteins/protein design
Introduction
Protein chimeras are hybrids formed by substitution of fragments between two parent proteins, ranging from short peptides to entire domains. Hybrid proteins occur in nature due
to dramatic gene reorganization events such as horizontal
gene transfer, gene duplication and fusion (Huynen and
Bork, 1998, Lynch and Conery, 2000). These recombination
events have led to major advances in the natural evolution of
organisms as well as protein architecture and their function
(Grishin, 2001, Söding and Lupas, 2003). Many methods in
protein engineering successfully imitate evolutionary
mechanisms to create proteins with new and useful properties
(Eisenbeis and Höcker, 2010). Similarly, possible evolutionary scenarios that might have led to a stable protein fold can
be explored.
†
Thease authors contributed equally to this work.
We have used the concept of illegitimate recombination
between non-homologous or very divergent sequences to
sample possible diversification of the (ba)8-barrel fold. The
(ba)8-barrel is a ubiquitous fold, commonly found among
enzymes (Sterner and Höcker, 2005). It consists of a closed
eight-stranded parallel b-sheet that is surrounded by eight
a-helices. Based on the remarkable 2-fold symmetry in two
enzymes of the histidine biosynthesis pathway, it was suggested that the (ba)8-barrel fold evolved through duplication
and fusion of an ancestral half-barrel (Lang et al., 2000).
This scenario has been tested in a number of engineering
experiments with the imidazole glycerol phosphate synthase
(HisF) from Thermotoga maritima (Höcker et al., 2001,
2004, 2009; Seitz et al., 2007; Fortenberry et al., 2011).
Diversification of (ba)8-barrel enzymes was further elucidated by the recombination of half-barrel-sized fragments
from closely related (ba)8-barrel proteins (Höcker et al.,
2004; Claren et al., 2009).
Prompted by the question where half-barrels could have
originated from, we searched structural databases and found
a striking similarity to a number of proteins with the
flavodoxin-like fold, a doubly wound (ba)5 protein with a
central five-stranded parallel b-sheet. Among the best hits
the following proteins were found: the nitrite response regulator NarL from Escherichia coli, the chemotaxis response
regulator CheY from T.maritima, and the flavodoxin-like
domain of the methylmalonyl CoA mutase (MMCoA) from
Propionibacterium shermanii (Höcker et al., 2002). To test
whether these structurally similar parts could substitute for
each other, we created a fold hybrid using parts of CheY,
which is a well-characterized single-domain protein, to substitute for structurally similar parts in the (ba)8-barrel
protein HisF from T.maritima (Bharat et al., 2008). The
resulting protein, which is named CheYHisF, was a stable
monomer that unfolds cooperatively. When analyzed by circular dichroism (CD) and fluorescence spectroscopy, it
appeared similar to the wild-type HisF protein. Solution of
the crystal structure, however, revealed unexpected details
for the interactions at the new interface. While the fragments
used for the construction of CheYHisF for the most part
retained their structure, we were surprised to find an additional ninth b-strand in the core of the barrel. This strand
invades between b1 and b2 of the CheY part and consists of
residues from the C-terminus of HisF and from the histidinetag included for purification purposes (Bharat et al., 2008).
The importance of the ninth strand was tested through
removal of the strand-forming residues, leading to the construct CheYHisF-sfr. This construct had a high tendency to
aggregate, thus illustrating the importance of this unexpected
structural element for the integrity of the fold. It was possible
to stabilize the CheYHisF-sfr chimera again through the
introduction of five targeted mutations at the fragments’
interface identified by computational design (Eisenbeis et al.,
2012). The design, named CheYHisF-sfr_RM, was targeted
to form a proper eight-stranded ba-barrel, which was
# The Author 2012. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: [email protected]
699
S.Shanmugaratnam et al.
confirmed experimentally by nuclear magnetic resonance
spectroscopy. CheYHisF-sfr_RM was not only a very stable,
well-folded protein, but also inherited the ability to bind
phosphorylated compounds from its parent HisF.
This suite of experiments demonstrated the potential of
fragment recombination for the rational design of new and
functional proteins. After this proof of concept we wanted to
test the generalizability of our chimeragenesis approach and
constructed two more hybrids by the combination of parts
from the flavodoxin-like proteins NarL and MMCoA with
the (ba)8-barrel HisF. The fragments of CheY, NarL and
MMCoA that were introduced into the HisF scaffold are
quite different, they share only sequence identities between
16 and 28%. While one of the hybrids did not fold properly,
the other one even surpasses our earlier chimera in terms of
stability. This second successfully created fold chimera illustrates how quickly proteins can diversify via non-homologous
recombination. The successful as well as the failed attempt
also give us hints for a general design strategy of fold
chimeras.
Results
The parts of the narL gene from E.coli and mmCoA gene
from P.shermanii were combined with a part of the hisF
gene from T.maritima. The combination was done analogous
to the earlier reported construction of the cheYhisF gene
(Bharat et al., 2008). Sequences encoding b-strand 1 and the
range from a-helix 2 to b-strand 5 in narL (residues 1 – 15
and 39 – 108) and mmCoA (residues 598– 609 and 633 – 706)
were cloned upstream of the sequence of the hisF-gene encoding a-helix 4 to a-helix 8 (residues 103 – 253), yielding
the chimeric genes narLhisF and mmCoAhisF (Fig. 1). The
fragments of NarL and MMCoA were selected to substitute
for the N-terminal part of HisF due to their remarkable structural similarity (Höcker et al., 2002). This similarity had
been identified in a search of the Protein Data Bank for
structures resembling the N- and C-terminal halves of HisF
(HisF-N and HisF-C) using the program DALI (Holm and
Sander, 1993). NarL had been the first non-(ba)8-barrel hit
for HisF-N immediately followed by MMCoA, while a
Materials and methods
Cloning of NarLHisF and MMCoAHisF
The narL gene was amplified from E.coli genomic DNA and
the flavodoxin-domain from MMCoA from P.shermanii was
constructed by polymerase chain reaction (PCR) using oligonucleotides. The resulting genes narLhisF and mmCoAhisF
were cloned into pET21a yielding the constructs pET21anarLhisF and pET21a-mmCoAhisF. The constructs were
sequenced to exclude any inadvertent PCR mutations.
Heterologous expression and purification of proteins
The NarLHisF and MMCoAHisF proteins, which both carry
a His6-tag at their C-termini, were produced in E.coli
BL21(gold). NarLHisF was mainly found in the soluble fraction of the cell extract and purified from there, whereas
MMCoAHisF was found in the insoluble fraction from
which it was solubilized and refolded. The expression, purification and refolding were performed as described for
CheYHisF (Bharat et al., 2008).
Analytical methods
Analytical gel filtration was performed using a calibrated
Superdex 75 column (Amersham Pharmacia). Highly concentrated protein was eluted at a flow rate of 0.5 ml min21 in
50 mM Tris and 300 mM KCl, pH 7.5. CD spectra were
recorded with a JASCO model J-810 spectropolarimeter.
Fluorescence measurements were carried out with a JASCO
FP-6500 spectrofluorometer. Temperature-induced unfolding
was analyzed by following the far-ultraviolet (UV) CD signal
at 222 nm at slowly increasing temperatures (1 K min21).
Protein unfolding induced by guanidinium chloride was followed by the decrease in fluorescence signal at 320 nm after
excitation at 280 nm. The proteins were incubated with different concentrations of denaturant, and the signals were
recorded after different time intervals until no further change
was observed.
700
Fig. 1. Construction of the fold chimeras. (a) Comparison of the topologies
of HisF (in blue), and the flavodoxin-like domains of MMCoA and NarL (in
green), as well as the chimeric constructs. (b) MMCoAHisF and NarLHisF
are constructed from fragments originating from MMCoA or NarL (residue
598–609 or 1 –15 including b1 and 633– 706 or 39– 108 containing a2-b5,
respectively) and from HisF (residue 103– 253 including a4-a8). (c) Model
of the NarLHisF structure generated by homology modeling with the
program Modeller (Sali and Blundell, 1993) using the crystal structure of
CheYHisF (PDB-id 3cwo) as a template. The parts originating from NarL
and HisF are shown in green and blue, respectively, and the residues after
Val230 that form the ninth strand are shown in orange.
Protein chimera design
search based on HisF-C returned MMCoA as the first hit
with a flavodoxin-like domain. The high similarity of the
structures covers more than three of the ba-modules while
the positioning of the other secondary structural elements
deviates drastically. We chose to swap the largest fragment
possible and thus set the fragment boundaries to where the
structures start to deviate from one another. After heterologous expression of the His6-tagged chimeras in E.coli, most
protein of NarLHisF was found in the soluble fraction of the
cell extract while all MMCoAHisF was found in inclusion
bodies. Thus, NarLHisF was purified from the soluble fraction
via metal affinity and subsequent gel filtration chromatography, while MMCoAHisF was refolded after solubilization
of the inclusion bodies in guanidinium chloride.
To determine the oligomerization state of the chimeras in
solution, analytical gel filtration experiments were conducted.
NarLHisF eluted as a sharp peak with an apparent molecular
mass of 36.7 kDa, which is slightly higher than the calculated molecular mass for the monomeric protein (26.7 kDa)
(Fig. 2a). We observed a similar discrepancy between the apparent and calculated molecular mass of the nine-stranded
barrel CheYHisF. This migration behavior most likely originates from the lesser compactness of the protein and the
wider radius of the barrel. The protein was applied to the
column highly concentrated. Next to the main peak a small
shoulder is visible, which is either a small contamination or
a degradation product. MMCoAHisF on the other hand
eluted in the void volume indicating that the protein forms
soluble aggregates. Consequently, MMCoAHisF was abandoned and only NarLHisF analyzed further.
The secondary structure content of NarLHisF was measured by far-UV CD. The spectrum is indicative of significant a-helical content and compares well to the spectra of
HisF and CheYHisF (Fig. 2b). Tertiary structure was investigated by using fluorescence and near-UV CD spectroscopy.
NarLHisF, CheYHisF and HisF contain the same single tryptophan residue in a-helix 5 providing a convenient way to
compare their tertiary structures. The fluorescence spectra are
Fig. 2. Characterization of the fold chimera NarLHisF. (a) Association state measured by analytical gel filtration. The applied protein was 10 mg ml21. (b)
Secondary structure evaluation by far-UV CD spectroscopy. Protein concentrations were 0.2 mg ml21 (d ¼ 0.1 cm). (c) Tertiary structure evaluation by
fluorescence spectroscopy. Protein concentrations were 0.2 mg ml21. (d) Tertiary structure evaluation by near-UV CD spectroscopy. Protein concentrations
were 0.65 mg ml21 for NarLHisF, 3.4 mg ml21 for HisF and 3.6 mg ml21 CheYHisF (d ¼ 1 cm). (e) Stability measured by thermal denaturation. Proteins
(0.2 mg ml21) were incubated at increasing temperatures while monitored at 222 nm in a 0.1 cm cuvette. All measurements were recorded in 10 mM Tris ( pH
7.5) at 258C.
701
S.Shanmugaratnam et al.
very similar, the emission maxima of NarLHisF, CheYHisF
and HisF occur at 321, 324 and 323 nm, respectively
(Fig. 2c). The near-UV CD spectra of all the proteins are
very distinct and extremely similar (Fig. 2d). These results
indicate that the indole chromophore of the tryptophan is in
a comparable asymmetric environment, and is similarly
shielded in all three proteins.
Stability of NarLHisF was investigated by reversible
unfolding with heat as well as guanidinium chloride. A cooperative unfolding behavior was observed during thermal
denaturation followed by CD at 222 nm (Fig. 2e). NarLHisF
exhibits a remarkable resistance to increasing temperatures
with a thermal melting point of 748C. Denaturation of
NarLHisF with guanidinium chloride also showed highly cooperative unfolding (Fig. 3). The data was analyzed on the
basis of the two-state model in comparison to the unfolding
of HisF (Carstensen et al., 2012) and the CheYHisF chimeras
(Bharat et al., 2008; Eisenbeis et al., 2012). NarLHisF
unfolds with a high m-value (m ¼ 17.0 kJ mol21 M21)
leading to a remarkable DGH2 O of 40.0 kJ mol21 (Table I).
Taken together the biophysical data show that the
NarLHisF chimera has adopted a native fold with a remarkable stability. The spectroscopic features of NarLHisF strongly resemble the ones of CheYHisF, the chimera that we
Fig. 3. Stability of the fold chimera NarLHisF measured by GdmCl-induced
denaturation. The measurement is shown in comparison to CheYHisF
(Bharat et al., 2008) and CheYHisF-sfr_RM (Eisenbeis et al., 2012).
Proteins at concentration of 0.1 mg ml21 were incubated with the given
concentrations of GdmCl in 10 mM Tris ( pH 7.5) at room temperature. The
loss of tertiary structure was followed by recording the decrease of the
fluorescence emission at 320 nm after excitation at 280 nm. The lines were
drawn as a visual aid.
Table I. Thermodynamic parameters for GdmCl-induced unfolding of
NarLHisF compared to HisF and computationally optimized chimera
CheYHisF-sfr_RM. The data were analyzed on the basis of the two-state
model
protein
DG (kJ mol21)
m (kJ mol21 M21)
[D]1/2 (M)
NarLHisF
HisFa
CheYHisF-sfr_RMb
40.0 + 1.9
55.6 + 2.3
18.0 + 2.5
17.0 + 0.9
19.9 + 0.9
7.5 + 0.8
2.4
2.8
2.4
The parameters were calculated from the data shown in Fig. 3.
a
Carstensen et al. (2012).
b
Eisenbeis et al. (2012).
702
reported earlier. Solution of its crystal structure revealed that
this chimera had assembled the original fragments quite
well, but had formed an extra b-strand with the C-terminal
residues including the tag used for purification. This led to
the formation of a nine-stranded barrel, which had not been
observed before. The same residues that form the unexpected
ninth b-strand are also present in NarLHisF. Crystallization
trials of NarLHisF unfortunately did not yield any diffracting
crystals. To analyze the contribution of the C-terminal residues to the integrity of the fold further, we prepared a shortened construct named NarLHisF-sfr, which ended after
valine 230, thereby eliminating possibly strand forming residues. This variant was found in the insoluble fraction of the
cell extract, but could be purified via refolding. Analytical
gel filtration showed that NarLHisF-sfr has a strong tendency
to aggregate indicating that the removed residues are
involved in stabilizing the fold and might form similar interactions in NarLHisF as they form in CheYHisF.
Discussion
We created two new protein chimeras, MMCoAHisF and
NarLHisF, in which parts of the (ba)8-barrel protein HisF
are replaced by fragments from the flavodoxin-like domains
of MMCoA and NarL, respectively (Fig. 1). MMCoA is an
enzyme that induces the formation of an adenosyl radical
from its cofacter, the coenzyme B12, which initiates a free
radical rearrangement of the substrate succinyl-CoA to
methylmalonyl-CoA (Mancia et al., 1996). NarL is a transcription factor that regulates genes involved in nitrite and
nitrate metabolism (Stock et al., 2000). Both proteins consist
of two domains. MMCoA comprises an N-terminal (ba)8barrel and a C-terminal flavodoxin-like domain, that share a
large interface. The two domains are connected by a long
linker of around 30 residues, which wraps around the
flavodoxin-like domain. The transplantation of a fragment
from this embedded domain probably led to the exposure of
hydrophobic areas that in the wild type enzyme are not solvent
accessible. Most likely this problem contributed significantly
to the misfolding and aggregation of the MMCoAHisF
chimera.
The response regulator NarL is composed of an
N-terminal flavodoxin-like domain (receiver) and a
C-terminal DNA-binding domain (effector). The two
domains are connected by a short linker and share a relatively small interface. The receiver domain belongs to the same
superfamily as the single-domain protein CheY, which was
used in our previously designed chimera CheYHisF (Bharat
et al., 2008). Biophysical characterization of the newly constructed chimera NarLHisF shows that this protein has
adopted a native fold strongly resembling CheYHisF. Even
the C-terminal residues that form the ninth strand in
CheYHisF are equally important for the integrity of the fold
suggesting that NarLHisF might also be a nine-stranded
ba-barrel (Fig. 1c). Surprisingly this protein even surpasses
CheYHisF in terms of stability. NarLHisF has a remarkable
resistance to increasing temperature even though only the HisF
part originates from a thermophile while the NarL part comes
from the mesophilic E.coli. Its thermal melting point of 748C
is 98C higher than the one of CheYHisF. Furthermore, assessment of the stability by chemical denaturation shows that
NarLHisF unfolds with higher cooperativity and possesses a
Protein chimera design
higher stability than CheYHisF and even than the computationally optimized chimera CheYHisF-sfr_RM (Eisenbeis
et al., 2012). Its thermodynamic parameters are in a range with
other natural proteins (Table I), and even compare well with
the extremely thermostable (ba)8-barrel HisF (Carstensen
et al., 2012).
Altogether, the results demonstrate how quickly new proteins can develop and diversify through recombination. With
this second fold chimera build from fragments of a (ba)8barrel and a flavodoxin-like protein we show that our approach is generalizable. By recombining structurally similar
fragments we can build new hybrid proteins with welldefined tertiary structure that are as stable as natural proteins
even from thermophilic organisms. These chimeras provide
suitable scaffolds for the establishment of catalytic activity.
Funding
This work was supported by Deutsche Forschungsgemeinschaft
(Grant HO 3022/1-2).
References
Bharat,T.A., Eisenbeis,S., Zeth,K. and Höcker,B. (2008) Proc. Natl Acad.
Sci. USA, 105, 9942– 9947.
Carstensen,L., Zoldak,G., Schmid,F.X. and Sterner,R. (2012) Biochemistry
Claren,J., Malisi,C., Höcker,B. and Sterner,R. (2009) Proc. Natl Acad. Sci.
USA, 106, 3704–3709.
Eisenbeis,S. and Höcker,B. (2010) J. Pept. Sci., 16, 538– 544.
Eisenbeis,S., Proffitt,W., Coles,M., Truffault,V., Shanmugaratnam,S.,
Meiler,J. and Höcker,B. (2012) J. Am. Chem. Soc., 134, 4019– 4022.
Fortenberry,C., Bowman,E.A., Proffitt,W., Dorr,B., Combs,S., Harp,J.,
Mizoue,L. and Meiler,J. (2011) J. Am. Chem. Soc., 133, 18026– 18029.
Grishin,N.V. (2001) J. Struct. Biol., 134, 167–185.
Höcker,B., Beismann-Driemeyer,S., Hettwer,S., Lustig,A. and Sterner,R.
(2001) Nat. Struct. Biol., 8, 32–36.
Höcker,B., Schmidt,S. and Sterner,R. (2002) FEBS Lett., 510, 133–135.
Höcker,B., Claren,J. and Sterner,R. (2004) Proc. Natl Acad. Sci. USA, 101,
16448– 16453.
Höcker,B., Lochner,A., Seitz,T., Claren,J. and Sterner,R. (2009)
Biochemistry, 48, 1145– 1147.
Holm,L. and Sander,C. (1993) J. Mol. Biol., 233, 123–138.
Huynen,M.A. and Bork,P. (1998) Proc. Natl Acad. Sci. USA, 95,
5849–5856.
Lang,D., Thoma,R., Henn-Sax,M., Sterner,R. and Wilmanns,M. (2000)
Science, 289, 1546– 1550.
Lynch,M. and Conery,J.S. (2000) Science, 290, 1151–1155.
Mancia,F., Keep,N.H., Nakagawa,A., Leadlay,P.F., McSweeney,S.,
Rasmussen,B., Bosecke,P., Diat,O. and Evans,P.R. (1996) Structure, 4,
339–350.
Sali,A. and Blundell,T.L. (1993) J. Mol. Biol., 234, 779 –815.
Seitz,T., Bocola,M., Claren,J. and Sterner,R. (2007) J. Mol. Biol., 372,
114–129.
Söding,J. and Lupas,A.N. (2003) Bioessays, 25, 837–846.
Sterner,R. and Höcker,B. (2005) Chem. Rev., 105, 4038–4055.
Stock,A.M., Robinson,V.L. and Goudreau,P.N. (2000) Annu. Rev. Biochem.,
69, 183–215.
703