Protein Expression and Purification 32 (2003) 317–322
www.elsevier.com/locate/yprep
Expression of N-formylated proteins in Escherichia coli
Shari Spector,a,b Julia M. Flynn,a Bruce Tidor,b Tania A. Baker,a,c and Robert T. Sauera,*
a
Department of Biology, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA
Biological Engineering Division and Department of Electrical Engineering and Computer Science, 77 Massachusetts Avenue,
Massachusetts Institute of Technology, Cambridge, MA 02139, USA
Howard Hughes Medical Institute, 77 Massachusetts Avenue, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
b
c
Received 26 June 2003, and in revised form 6 August 2003
Abstract
In bacteria, protein expression initiates with a formyl-methionine group. Addition of the antibiotic actinonin, a known peptide
deformylase inhibitor, at the time of induction of protein expression results in the retention of the formyl group by the overexpressed
protein. In addition, because deformylation is a prerequisite for removal of the initiating methionine, this post-translational processing step is also prevented by actinonin, and the N-formyl methionine residue is retained by proteins from which it is normally
removed. We have demonstrated the applicability of this system for obtaining N-modified forms of several different proteins and use
one of these modified molecules to show that the N-terminal amino group is not required for ClpXP degradation of proteins bearing
an N-terminal recognition signal.
Ó 2003 Elsevier Inc. All rights reserved.
Keywords: Protein modification; a-Amino group
A common way to test the role of particular side-chain
groups in protein function is to perform site-directed
mutagenesis at those positions and then to compare the
activity of the mutants with that of the wild-type protein.
When the moiety of interest is not a side-chain, directed
mutations are not possible. For instance, to probe the
function of the N-terminal a-amino group, the length of
the protein chain is usually altered by adding or deleting
N-terminal residues. This leaves a charged amino group
but alters its relative location in the protein sequence.
Alternatively, the a-amino group may be modified
chemically by transamination [1,2] or by acetylation, but
these reactions can fail or result in undesired side reactions with the e-amino group of lysine side chains.
In bacteria, protein synthesis initiates with formylmethionine (fMet).1 As schematized in Fig. 1, the formyl
*
Corresponding author. Fax: 1-617-258-0673.
E-mail address: [email protected] (R.T. Sauer).
1
Abbreviations used: ESI, electrospray ionization mass spectrometry; fMet, formyl-methionine; IPTG, isopropyl-b-D -thiogalactoside;
MALDI-TOF, matrix assisted laser desorption and ionization time of
flight mass spectrometry; MAP, methionine aminopeptidase; PDF,
peptide deformylase; scArc, single-chain Arc repressor.
1046-5928/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.pep.2003.08.004
group is then removed post-translationally by peptide
deformylase (PDF), leaving a free a-amino group that is
positively charged at neutral pH. Depending on the
identity of the second amino acid in the protein chain,
deformylation may be followed by removal of the initiating methionine by the enzyme methionine aminopeptidase (MAP). If the deformylation step in protein
expression could be blocked, then proteins should retain
the modified and uncharged amino-terminus. Actinonin
is a PDF inhibitor [3] but is normally ineffective in
Escherichia coli because it is removed from the cell by
efflux pumps involved in multidrug resistance. However,
the antibiotic kills E. coli strains bearing a deletion
of the acrAB efflux pump genes [3], suggesting that in
this genetic background actinonin might be useful in
obtaining N-modified proteins.
We show here that N-formylated proteins can be
expressed in high yield in E. coli DacrAB strains if
actinonin is added at the time of induction of protein
expression. This system has been tested using a singlechain variant of phage P22 Arc repressor (scArc) [4],
phage P22 Mnt repressor, the yeast protein Erv2 and
the fusion protein IscS1–11 –Arc. The latter protein is
targeted by its N-terminal 11 residues for degradation
318
S. Spector et al. / Protein Expression and Purification 32 (2003) 317–322
Fig. 1. Processing of newly synthesized proteins in bacteria. Translation initiates with N-formyl methionine. Once translation is complete, peptide
deformylase (PDF) removes the formyl group from fMet. Actinonin inhibits PDF, blocking this step of processing. Depending on the identity of the
second amino-acid in the protein sequence, the methionine residue may be removed by methionine aminopeptidase (MAP), but this step is contingent
on the removal of the formyl group by PDF.
by the ClpXP protease[5]. For N-terminal ClpXP degradation tags, it is not known whether the free a-amino
group is required for recognition and degradation. We
show that the IscS1–11 –Arc fusion is degraded at essentially the same rate whether it bears a free or formylated
a-amino group, indicating that a free N-terminus is not
required for ClpXP degradation of this substrate.
Materials and methods
Materials
Ampicillin and actinonin were obtained from Sigma–
Aldrich (St. Louis, MO). Isopropyl-b-D -thiogalactoside
(IPTG) was purchased from Gold BioTechnology (St.
Louis, MO).
Plasmids and strains
AG100A (E. coli K-12 DacrAB) was a generous gift
from Nikaido and Levy [6,7]. To enable expression from
pET vectors, this strain was transduced with DE3
(Novagen, Madison, WI). The gene for single-chain Arc
repressor-st11 (scArc) was amplified by PCR from
pLA110 [4], using primers to engineer NdeI and BamHI
restriction sites at the 50 and 30 ends of the gene, respectively, and subcloned into pET11a (Novagen).
Plasmid pJC11, a derivative of pET21b(+) bearing the
yeast gene Erv2, was kindly provided by Sevier [8].
Plasmid pET400-st6, derived from pET-24-d(+), contains the full length Mnt-st6 gene [9]. A plasmid derived
from pET11a encodes a fusion protein IscS1–11 –Arc
consisting of the first 11 residues of IscS (MKLPIYLDYSA) followed by Arc-st11 [5]. All genes encoded a
C-terminal His6 tag for Ni–NTA purification.
picked and grown overnight at 37 °C in LB plus 100 lg/
ml ampicillin and the overnight culture was diluted to
prepare 1 L cultures for growth and induction under the
same conditions. Cells were grown to an OD600 between
0.6 and 0.7. Expression was induced either by addition
of 1 mM IPTG or 1 mM IPTG plus actinonin at a final
concentration of 2 lg/ml. This actinonin concentration
is 8-fold higher than the minimum inhibitory concentration, defined as the minimum concentration required
such that no culture growth is observed after 18–24 h at
35 °C, as measured for AG100A [3]. After 2 h, cultures
were harvested by centrifugation in a Beckman J-6B
centrifuge at 4000 rpm, 4 °C, for 10 min. Cell pellets were
stored at )80 °C prior to lysis and protein purification.
All proteins were C-terminally His-tagged and purified by Ni–NTA affinity chromatography using the
standard protocol for Arc-st11 [10]. To purify the wildtype and formylated proteins, cell pellets were resuspended in a pH 8 buffer containing 0.1 M NaH2 PO4 ,
0.01 M Tris, 6 M guanidine hydrochloride, and 10 mM
imidazole. Cells were lysed by stirring for 1 h at 4 °C and
then sonicated. After centrifugation, the supernatant
Expression and purification
For each protein, AG100A(DE3) cells were transformed with the appropriate plasmid and plated on LB
agar with 100 lg/ml ampicillin. A single colony was
Fig. 2. SDS–PAGE analysis of protein purity. SDS–PAGE shows that
all of the proteins in this study are >95% pure. The only observable
impurity at a molecular weight of approximately 25 kDa corresponds
to SlyD, a histidine-rich E. coli protein which often co-purifies on Ni–
NTA resin with His-tagged protein.
S. Spector et al. / Protein Expression and Purification 32 (2003) 317–322
319
Table 1
Mass spectrometry results for proteins and proteolytic fragments
Protein
scArc
Sample
Technique
Peptide
Mass (Da)
Observed
Calculated
Intact protein
V8 digest, 33% acetonitrile
ESI
MALDI-TOF
14,854, 14,871
2137, 2151
14,856
2136
fMet–Mnt
V8 digest, 35%
Intact protein
V8 digest, 34%
V8 digest, 36%
Intact protein
V8 digest, 31%
Intact protein
MALDI-TOF
ESI
MALDI-TOF
MALDI-TOF
ESI
MALDI-TOF
ESI
2463
14,882
2164
2491
10,370
2148
10,529
Erv2
fMet–Erv2
IscS1–11 –Arc
fMet–IscS1–11 –Arc
V8 digest, 31% acetonitrile
V8 digest, 34% acetonitrile
Intact protein
Intact protein
Intact protein
Intact protein
2465, 2478
14,884
2165
2491
10,371
2148
10,530, 10,546,
10,371, 10,388
2148
2307
20,546
20,705
9114
9142
fMet–scArc
Mnt
acetonitrile
acetonitrile
acetonitrile
acetonitrile
MALDI-TOF
MALDI-TOF
ESI
ESI
ESI
ESI
was applied to Ni–NTA resin (Qiagen, Valencia, CA).
The column was washed extensively with the above
buffer and protein was eluted in 0.2 M acetic acid, 6 M
guanidine hydrochloride. After elution, scArc and
fMet–scArc were dialyzed into water and lyophilized.
Mnt and fMet–Mnt were dialyzed into 50 mM Tris,
0.5 mM EDTA, 0.1 M KCl, pH 7.5. Erv2 and fMet–
Erv2 were dialyzed into 10 mM Tris, 0.1 mM EDTA,
150 mM KCl, pH 7.5. IscS1–11 –Arc and fMet–IscS1–11 –
Arc were dialyzed into 50 mM Tris, 250 mM KCl,
0.1 mM EDTA, pH 7.5. Each protein was greater than
95% pure as assayed by SDS–PAGE (Fig. 2).
Proteolysis and HPLC
The scArc and Mnt proteins were digested with V8
protease to obtain 10–20 amino-acid N-terminal fragments. Protein stocks were diluted with 8 M urea and
protease dissolved in 50 mM Tris, pH 8, was added to
result in a final urea concentration of 2 M. Samples were
incubated overnight at 37 °C. Following digestion,
samples were subjected to high-performance liquid
chromatography (HPLC) using a Shimadzu LC10
HPLC (Columbia, MD) equipped with a Vydac C18
analytical column (Hesperia, CA). Peptides were separated using aqueous buffers containing 0.06% trifluoroacetic acid and an organic phase consisting of 80%
acetonitrile, 0.052% trifluoroacetic acid.
2148
2307
20,550
20,709
9111
9139
scArc
Met1–Glu17, Met1–Glu17
(Met–SO)
Met1–Asp20
fMet–scArc
fMet1–Glu17
fMet1–Asp20
Mnt
Ala2–Glu18
fMet–Mnt, fMet–Mnt
(Met–SO), Mnt, Mnt (Met–SO)
Ala2–Glu18
fMet1–Glu18
Erv2
fMet–Erv2
IscS1–11 –Arc
fMet–IscS1–11 –Arc
matrix assisted laser desorption and ionization time of
flight (MALDI-TOF) mass spectrometry (Table 1).
Degradation assays
Degradation reactions were performed as follows:
ClpX6 (0.3 lM), ClpP14 (0.8 lM), ATP (4 mM), and an
ATP regeneration system (50 lg/ml creatine kinase and
2.5 mM creatine phosphate) were mixed in PD buffer
(25 mM Hepes–KOH, pH 7.6, 5 mM MgCl2 , 5 mM KCl,
15 mM NaCl, 0.032% (v/v) Nonidet P-40, 10% (v/v)
glycerol) and incubated for 2 min at 30 °C. The protein
substrate (5 lM) was added and samples were removed
at specific times and analyzed by SDS–PAGE.
Results
C-terminally His6 -tagged variants of scArc, Mnt,
Erv2, and IscS1–11 –Arc were expressed from overproducing plasmids to good yield in the DacrAB E. coli
strain AG100A(DE3) in both the presence and absence
of actinonin. Following a single Ni–NTA affinity
chromatography step, each protein was greater than
95% pure as assayed by SDS–PAGE (Fig. 2). Proteinchemical analyses for each purified test protein are
described below.
Single-chain (sc) Arc
Mass spectrometry
All mass spectrometry was performed at the MIT
Biopolymers Laboratory. Full-length proteins were
submitted for electrospray ionization mass spectrometry
(ESI), and HPLC purified peptides were subjected to
Electrospray ionization mass spectrometry of scArc
expressed in the absence of actinonin gave masses of
14,854 and 14,872 Da (Table 1). The smaller mass corresponds within the error of ESI (approximately 0.01%)
to the wild-type protein (14,856 Da calculated), and the
320
S. Spector et al. / Protein Expression and Purification 32 (2003) 317–322
larger mass, 16 Da higher than expected, is consistent
with oxidation of a single methionine residue (8 total) to
methionine sulfoxide. After expression in the presence of
actinonin, scArc gave a single peak in the ESI spectrum
(14,884 Da observed and 14,882 Da calculated), suggesting that this protein retained its formyl group with a
molecular weight of 28 Da. Interestingly, no oxidation of
methionine was detected in the formylated scArc protein.
To confirm that the increase in mass was caused by the
retention of the formyl group at the N-terminus, scArc
(no actinonin) and fMet–scArc (plus actinonin) were
subjected to overnight digestion by V8 protease. Analysis
of the digested products by HPLC (Fig. 3) revealed two
peaks (1/1 and 2/2 ) that eluted at distinct positions for
each protein. Peak 1 from scArc contained peptides with
masses of 2137 and 2151 Da, as determined by MALDITOF. These masses agree, within the error of MALDITOF (approximately 0.1%), with the mass expected for
Met1–Glu17 (2136 Da calculated) and the same peptide
with a single methionine sulfoxide. The corresponding
peak (1 ) from the protein sample expressed in the
presence of actinonin showed a single peptide with a
mass of 2165 Da, as expected for fMet1–Glu17 (2164 Da
calculated). Peak 2 from scArc contained peptides with
masses of 2465 and 2478 Da, consistent with Met1–
Asp20 (2463 Da calculated) and the same peptide with a
single oxidized methionine. Peak 2 from the actinonin
sample had a peptide with the mass expected for fMet1–
Asp20 (2491 Da observed and 2493 Da calculated).
Mnt
The N-terminal methionine of Mnt is normally
removed post-translationally by MAP. Thus, the
predicted mass difference between the processed protein
with its free N-terminus and the formylated and unprocessed protein is 159 Da. Mnt expressed in the absence of actinonin had a mass of 10,371 Da (10,370 Da
calculated). Mnt expressed in the presence of actinonin
showed major ESI mass spectrometry peaks consistent
with the predicted mass for fMet–Mnt (10530 Da observed, 10,529 Da calculated) and the same sequence
bearing a single oxidized methionine (10,546 Da), as well
as two minor peaks consistent with the unformylated,
processed protein (10,371 Da) and this protein with an
oxidized methionine group (10,388 Da). Following V8
proteolytic digestion and HPLC, the sample expressed
in the absence of actinonin yielded a peptide that eluted
at 31% acetonitrile (Fig. 3B, peak 1) and had a mass
observed by MALDI-TOF corresponding to Ala2–
Glu18 (2148 Da observed and calculated). The same
peptide was also present at roughly one-third the concentration (peak 1 ) in the digest of the protein expressed in the presence of actinonin. The latter
chromatogram also contained a unique peak at 34%
acetonitrile (peak 2 ) that contained a single peptide
corresponding to fMet1–Glu18 (2307 Da observed and
calculated).
Erv2
The N-terminal methionine of Erv2 is normally removed post-translationally. As a result, protein expressed in the presence of actinonin should be higher in
mass than Erv2 by 159 Da. Indeed, the observed ESI
masses were 20,546 Da (20,550 Da calculated) for Erv2
and 20,705 Da (20,709 Da calculated) for fMet–Erv2.
Both experimental masses were 4 Da lower than
Fig. 3. HPLC traces of V8 digested proteins. (A) scArc (solid line) and fMet–scArc (dashed line). (B) Mnt (solid line) and fMet–Mnt (dashed line).
Peaks bearing the amino-terminal peptides are marked.
S. Spector et al. / Protein Expression and Purification 32 (2003) 317–322
Fig. 4. ClpXP protease assays. IscS1–11 –Arc and fMet–IscS1–11 –Arc
were degraded by ClpXP protease. The fraction of substrate remaining
was determined by SDS–PAGE (inset) and is plotted as a function of
time.
predicted, consistent with the oxidation of 4 of Erv2Õs 6
cysteine side chains to form two disulfide bridges.
IscS1–11 –Arc
Arc repressor is not normally a substrate for the
ClpXP protease; however, fusion of the N-terminal 11
residues of IscS, a cysteine desulfurase, to Arc repressor
(IscS1–11 –Arc) targets this fusion protein for degradation
by ClpXP[5]. IscS is just one of a group of ClpXP
substrates with the consensus NH2 –Met–Lys–U–U–X5 –
U (U ¼ hydrophobic), suggesting that this sequence is
recognized by ClpX and therefore targets proteins for
degradation by the ClpXP protease. To determine
whether a free N-terminus is required for degradation,
IscS1–11 –Arc was expressed in the presence or/absence of
actinonin. The wild-type protein had an ESI mass of
9114 Da (9111 Da calculated) and the protein expressed
in the presence of actinonin had a mass (9142 Da observed and 9139 Da calculated) that is consistent with
retention of the fMet to produce fMet–IscS1–11 –Arc.
IscS1–11 –Arc and fMet–IscS1–11 –Arc were tested for
the ability to be degraded by ClpXP. In each case,
ClpXP was briefly incubated with ATP and an ATP
regeneration system, protein substrates were added, and
aliquots were removed after various times for analysis
by SDS–PAGE. As shown in Fig. 4, IscS1–11 –Arc and
fMet–IscS1–11 –Arc are degraded at very similar rates,
indicating that the N-terminal amino group of the fusion protein is not required for degradation by ClpXP.
Discussion
The results presented in this paper demonstrate that
proteins retaining the N-terminal fMet can be expressed
321
in an E. coli strain lacking the acrAB efflux pump if
actinonin is added at the time of induction of protein
over-expression. Such strains can only be grown for
short periods of time in the presence of this antibiotic
because of toxicity. Nevertheless, good yields of four test
proteins (scArc, Mnt, Erv2, and IscS1–11 –Arc) were obtained. For each of these proteins except Mnt, the purified proteins appeared to be uniformly formylated
based on mass spectrometry. For Mnt, roughly twothirds of the purified protein was formylated and the
remaining one-third had the initiating fMet residue removed. We do not know if the properly processed Mnt
was synthesized prior to induction and addition of actinonin (perhaps from leak-through expression of the
pET vector employed) or if actinonin incompletely inhibited the processing of Mnt by peptide deformylase. In
the latter case, it is possible that a higher actinonin
concentration might improve the ratio of formylated to
processed Mnt as only a single actinonin concentration,
8-fold higher than the minimum inhibitory concentration (see Materials and methods) [3], was tested in these
experiments. In the former case, use of a different expression vector with reduced expression in the absence
of inducer might improve the ratio of fMet protein to
processed protein.
Expression in the presence of actinonin in an acrAB
deletion strain of E. coli provides a simple method for
obtaining proteins with modified N-termini. This, in turn,
provides a straightforward way to test the role of the free
a-amino group in systems in which it appears important
for protein function. In the studies described here, we
purified fMet–IscS1–11 –Arc and tested whether blockage
of the a-amino group affected degradation by ClpXP.
The N-terminal residues of E. coli IscS target both this
protein and an Arc fusion protein for ClpXP degradation. Moreover, a number of other ClpXP substrates, like
IscS, share the consensus NH2 –Met–Lys–U–U–X5 –U,
where U represents a hydrophobic side-chain. It seemed
possible therefore that the free a-amino group in these
proteins represented a recognition determinant for
ClpXP. This does not, however, appear to be the case. We
found that fMet–IscS1–11 –Arc was degraded at the same
rate as Met–IscS1–11 –Arc (Fig. 4). Because the N-terminal
amino group is not required for ClpXP binding or degradation, it will be interesting to determine whether the
sequence motif shared by this group of ClpXP substrates
could target proteins for degradation at an exposed internal or even a C-terminal position in a protein sequence.
In the case of scArc, the wild-type protein is oxidized
at a single position, which our studies show is Met1,
Met4, or Met7. Interestingly, when scArc was expressed
in the presence of actinonin, the resulting formylated
protein was not oxidized. This suggests that Met1 is the
site of oxidation and, furthermore, that formylation
protects the protein from oxidation. For proteins with
N-terminal methionines in which a free a-amino group
322
S. Spector et al. / Protein Expression and Purification 32 (2003) 317–322
is not required for function, formylation may offer a
general means of protecting Met1 from oxidation.
Acknowledgments
This work was supported by NIH Grants AI-16892,
AI-15706, and GM-55758. S.S. was supported by NIH
GM-20702. T.A.B. is an employee of HHMI.
References
[1] H.B.F. Dixon, R. Fields, Specific modification of NH2 -terminal
residues by transamination, Methods Enzymol. 25 (1972) 409–419.
[2] C.S. Hayes, E. Alarcon-Hernandez, P. Setlow, N-terminal amino
acid residues mediate protein–protein interactions between DNAbound alpha/beta-type small, acid-soluble spore proteins from
Bacillus species, J. Biol. Chem. 276 (2001) 2267–2275.
[3] D.Z. Chen, D.V. Patel, C.J. Hackbarth, W. Wang, G. Dreyer,
D.C. Young, P.S. Margolis, C. Wu, Z.J. Ni, J. Trias, R.J. White,
Z. Yuan, Actinonin, a naturally occurring antibacterial agent, is a
potent deformylase inhibitor, Biochemistry 39 (2000) 1256–1262.
[4] C.R. Robinson, R.T. Sauer, Covalent attachment of Arc repressor
subunits by a peptide linker enhances affinity for operator DNA,
Biochemistry 35 (1996) 109–116.
[5] J.M. Flynn, S.B. Neher, Y.-I. Kim, R.T. Sauer, T.A. Baker,
Proteomic discovery of cellular substrates of the ClpXP protease
reveals five classes of ClpX-recognition signals, Mol. Cell 11
(2003) 671–683.
[6] H. Okusu, D. Ma, H. Nikaido, AcrAB efflux pump plays a major
role in the antibiotic resistance phenotype of Escherichia coli
multiple-antibiotic-resistance (Mar) mutants, J. Bacteriol. 178
(1996) 306–308.
[7] D.G. White, J.D. Goldman, B. Demple, S.B. Levy, Role of the
acrAB locus in organic solvent tolerance mediated by expression
of marA, soxS, or robA in Escherichia coli, J. Bacteriol. 179 (1997)
6122–6126.
[8] C.S. Sevier, J.W. Cuozzo, A. Vala, F. Aslund, C.A. Kaiser, A
flavoprotein oxidase defines a new endoplasmic reticulum pathway for biosynthetic disulphide bond formation, Nat. Cell. Biol. 3
(2001) 874–882.
[9] A. Berggrun, R.T. Sauer, Contributions of distinct quaternary
contacts to cooperative operator binding by Mnt repressor, Proc.
Natl. Acad. Sci. USA 98 (2001) 2301–2305.
[10] M.E. Milla, B.M. Brown, R.T. Sauer, P22 Arc repressor:
enhanced expression of unstable mutants by addition of polar
C-terminal sequences, Protein Sci. 2 (1993) 2198–2205.