Covalent Protein Labeling by Enzymatic Phosphocholination

Article abstract of DOI:10.1002/anie.201502618

We present a new protein labeling method based on the covalent enzymatic phosphocholination of a specific octapeptide amino acid sequence in intact proteins. The bacterial enzyme AnkX from Legionella pneumophila has been established to transfer functional phosphocholine moieties from synthetically produced CDP-choline derivatives to N-termini, C-termini, and internal loop regions in proteins of interest. Furthermore, the covalent modification can be hydrolytically removed by the action of the Legionella enzyme Lem3. Only a short peptide sequence (eight amino acids) is required for efficient protein labeling and a small linker group (PEG-phosphocholine) is introduced to attach the conjugated cargo.

Full text of DOI:10.1002/anie.201502618

Angewandte
Chemie
International Edition: DOI: 10.1002/anie.201502618
German Edition: DOI: 10.1002/ange.201502618
Protein Labeling
Covalent Protein Labeling by Enzymatic Phosphocholination**
Katharina Heller, Philipp Ochtrop, Michael F. Albers, Florian B. Zauner, Aymelt Itzen,* and
Christian Hedberg*
Dedicated to Professor Roger S. Goody on the occasion of his 71th birthday
Abstract: We present a new protein labeling method based on
the covalent enzymatic phosphocholination of a specific
octapeptide amino acid sequence in intact proteins. The
bacterial enzyme AnkX from Legionella pneumophila has
been established to transfer functional phosphocholine moi-
eties from synthetically produced CDP-choline derivatives to
N-termini, C-termini, and internal loop regions in proteins of
interest. Furthermore, the covalent modification can be hydro-
lytically removed by the action of the Legionella enzyme
Lem3. Only a short peptide sequence (eight amino acids) is
required for efficient protein labeling and a small linker group
(PEG-phosphocholine) is introduced to attach the conjugated
cargo.
an activated precursor nucleotide (coenzyme A) to a serine of
the undecapeptide sequence DSLEFIASKLA.[3] The Ppant is
chemically modified and combined with the PPTase-catalyzed
reaction to transfer functional Ppant to recombinantly
produced proteins containing the PPTase recognition sequen-
ce.[3b,c,4] If, however, the recognition sequence is replaced by
the 8 kDa acyl carrier protein (ACP), modification and
subsequent cleavage with PPTase and ACP hydrolase
(AcpH), respectively, is achievable.[4b] However, recently
the demodification of the 11 amino acid recognition sequence
by different AcpH has been reported.[5] We sought to develop
an alternative enzymatic protein labeling strategy utilizing
a defined nucleotide that would make it possible to conven-
iently attach and detach a label of choice at a short peptide
sequence. Here, we present data on the general applicability
of covalent enzymatic phosphocholination of proteins for
regioselective labeling and delabeling using synthetic CDP-
choline derivatives.
The recently discovered host cell phosphocholination by
Legionella pneumophila is a posttranslational modification in
which a phosphocholine moiety is enzymatically transferred
via the Legionella effector protein AnkX from a cytidine
diphosphate choline (CDP-choline) to a serine residue in the
switch II loop of the small GTPase Rab1 in the host cell
(Scheme 1A).[6] Interestingly, the Rab1 phosphocholine
modification is hydrolytically cleaved by the Legionella
phosphodiesterase Lem3 at a later stage of infection (Sche-
me 1A).[7] We have recently observed that AnkX recognizes
only the TITSSYYR peptide sequence of the switch II loop of
Rab1b, but does not necessarily require the discrete GTPase
structure.[7a]
S
ite-directed labeling strategies are essential for modifying
proteins with functionalities not defined by the genetic code.
It is generally desirable to introduce modifications by
regioselective chemical reactions to maintain protein activity
and integrity with minimal cross-reactivity.[1] An alternative
approach is the utilization of enzymes that specifically
recognize small amino acid sequences in a protein of
interest.[2] For instance, phosphopantetheinyl (Ppant) trans-
ferase (PPTase) physiologically transfers a Ppant moiety from
[*] M.Sc. K. Heller,[+] F. B. Zauner, Prof. Dr. A. Itzen
Center for Integrated Protein Science Munich
Technische Universität München, Department Chemistry
Lichtenbergstrasse 4, 85748 Garching (Germany)
E-mail: aymelt.itzen@tum.de
M.Sc. P. Ochtrop,[+] M.Sc. M. F. Albers, Prof. Dr. C. Hedberg
Chemical Biology Center (KBC), Institute of Chemistry
Umeå University, 90187 Umeå (Sweden)
We envisioned modification/synthesis of functionalized
CDP-choline derivatives carrying any label of choice (Sche-
me 1B), which could be transferred to fusion proteins of
interest, where the recognition sequence could be located
either at N-terminal, C-terminal, or in internal loop regions
(Scheme 1C).
E-mail: christian.hedberg@umu.se
Prof. Dr. C. Hedberg
Max Planck Institute of Molecular Physiology
Department of Chemical Biology
Otto-Hahn-Strasse 11, 44227 Dortmund (Germany)
[+] These authors contributed equally to this work.
First, we investigated several AnkX (949 amino acids)[6a,8]
and Lem3 (570 amino acids)[7] constructs for their perfor-
mance in protein expression and enzymatic activity (Fig-
ure S1). Constructs AnkX1–800 and full-length Lem3 were
found to be superior with respect to expression, purification,
and desired activity, and were therefore retained for establish-
ing our protein labeling strategy. In an alanine scan of the
AnkX recognition sequence, we established that modifications
up- and downstream of octapeptide TITSSYYR were fully
accepted, and modifications (marked in bold) within TITS-
SYYR attenuated acceptor capability. Changes within TSSYY
[**] This work was supported by a priority program (SPP 1623) and
a collaborative research center (SFB 1035, project B05) of the
German Research Foundation (DFG) as well as by the Knut and
Alice Wallenberg Foundation (KAW 2013.0187), Sweden. M.F.A.
thanks the Max Planck Society for a fellowship from the Interna-
tional Max Planck Research School (IMPRS). We are indebted to
Sascha Gentz, Dr. Matthias Müller (MPI-Dortmund), Prof. Stephan
Sieber, Prof. Johannes Buchner, Dr. Martin Haslbeck, and Dr. Sabine
Schneider (TU Munich) for useful discussions and the provision of
materials.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201502618.
Angew. Chem. Int. Ed. 2015, 54, 10327 –10330
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
10327

.
Angewandte
Communications
Figure 1. Phosphocholination of AnkX model substrates. A) Rab1b3–174
and various protein model substrates (SUMO, MBP) equipped with
the octapeptide recognition sequence at N- and C-termini are moni-
tored by western blotting against phosphocholine (for analysis see
Figure S6). B) Negative control corresponding to Figure 1A with
SUMO-TITSAYYR. Anti-His6 blotting served as a loading control.
C) Enzymatic phosphocholination of DrrA340–533 at internal loop regions
that have been extended with the AnkX octapeptide recognition motif,
monitored by western blotting against phosphocholine. D) Catalytic
efficiencies (kcat/KM) of the phosphocholination reaction on the various
protein substrates relative to the corresponding reaction on the native
substrate Rab1b3–174. Ctrl: loading control of individual experiments
with His6-MBP-PC-Rab1b3–174; C: intensity of quantitatively phospho-
cholinated protein.
Scheme 1. A) Protein phosphocholination and dephosphocholination
by the Legionella enzymes AnkX and Lem3. B) Design of a CDP-
choline derivative bearing a label at one of the methyl positions on the
quaternary amine. C) Design of protein substrates (protein of interest,
POI) equipped with the octapeptide recognition sequence at N- and C-
termini and at internal loop regions.
completely abolished phosphocholination, suggesting that the
octapeptide sequence TITSSYYR is the minimum conserved
recognition motif for efficient phosphocholination by AnkX
(Figure S2). In order to investigate whether the octapeptide
sequence could be phosphocholinated by AnkX1–800 as protein
fusions, we recombinantly attached the TITSSYYR octapep-
tide sequence at the N- and C-termini of the maltose binding
protein (MBP) and the small ubiquitin-like modifier (SUMO)
and carried out heterologous expression in E. coli (Figure S3
and Supporting Information). Phosphocholination of the
model proteins (50 mm) was investigated with a catalytic
amount of AnkX (1 mm) in the presence of 1 mm CDP-choline
and analyzed by ESI-MS after 24 h of incubation (Figure S4).
SUMO constructs (TITSSYYR-SUMO and SUMO-TITS-
SYYR) were phosphocholinated at the N- and C-termini
(Figure S4) and MBP at its N-terminus (but also minor
modification of C-terminally tagged MBP), verifying that
phosphocholination can also be introduced into model pro-
teins. The modifications of the AnkX substrates were vali-
dated independently by detecting protein phosphocholination
with an anti-phosphocholine antibody by western blotting
(Figure 1A–C). In this assay, also phosphocholination of MBP
at the C-terminus was detected. By using the identical assay
we could show that phosphocholine transfer is dependent on
the correct peptide sequence and acceptor amino acid (Fig-
ure 1A), since the SUMO-TITSAYYR substrate (lacking the
serine that is targeted by AnkX) failed to undergo AnkX-
mediated phosphocholination (Figure 1B). Including AnkX
recognition motifs into internal loop regions of target proteins
could significantly expand the application of protein labeling
by phosphocholination. We therefore introduced the TITS-
SYYR octapeptide sequence into the GEF domain of the
[9]
enzyme DrrA (amino acids 340–533, DrrA340–533
)
(Fig-
ure S3): 1) E425-TITSSYYR-S426 (DrrA-A), 2) T455-TITS-
SYYR-P456 (DrrA-B), 3) N509-TITSSYYR-V510 (DrrA-C).
Next, we investigated the phosphocholination by AnkX1–
of these proteins via Western blotting using the anti-
800
phosphocholine antibody (Figure 1C) and ESI-MS (Fig-
ure S4). The phosphocholination of internal loop regions on
DrrA was readily detected, thus suggesting that the AnkX
recognition motif TITSSYYR can be used for phosphocholi-
nation in internal loops. Full enzymatic phosphocholination
of peptide-tagged SUMO, MBP, and DrrA proteins could be
achieved with extended reaction times, thus demonstrating
that the enzymatic reaction does not halt at an intermediary
labeling state (Figure S4). Finally, we derived time-course
data of the AnkX phosphocholination reaction using quanti-
tative densitometry of the western blots. The apparent
catalytic efficiencies (kcat/KM) of the reaction were calculated
to compare the different AnkX substrates (Figure 1D and
Figures S5 and S6). Not surprisingly, the artificial substrates
(i.e. peptide-tagged SUMO, MBP, and DrrA) perform less
well than the native Rab1b protein substrate, which may be
attributed to increased conformational flexibility (e.g.
SUMO, MBP) and/or structural restrictions (e.g. internal
loops of DrrA).
The development of phosphocholination by AnkX as
a protein labeling strategy depends on the synthesis and
10328 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 10327 –10330

Angewandte
Chemie
utilization of CDP-choline derivatives permitting the transfer
of the label along with the derivatized phosphocholine group.
The crystal structure of AnkX1–484 in complex with CDP-
choline shows the quaternary choline ammonium group
positioned in a deep, but solvent-accessible pocket,[8] thus
suggesting that any label should be connected to the
quaternary ammonium function by a flexible hydrophilic
linker of sufficient length (Figure S7). We generated synthetic
CDP-choline derivatives containing a fluorescein label con-
nected to the amine group via a polyethylene glycol (PEG)
linker of varying length (1–5, Figure 2A) (for synthesis
details, see Figure S7).[10]
Next, we investigated qualitatively whether the CDP-
choline derivatives 1–5 are substrates for AnkX (Figure 2B).
Rab1b was used as the acceptor substrate (50 mm) and
phosphocholination was monitored after the addition of the
CDP-choline derivatives (1 mm) in the presence of catalytic
amounts of AnkX1–800 (0.5 mm) by ESI-MS (Figure S8) and in-
gel fluorescence after SDS-PAGE (Figure 2B). First, we
probed the active site of AnkX with fluorescent derivatives 1–
3; 1 and 2 were transferred by AnkX1–800, but not derivative 3,
where the ammonium/amine function was replaced by
oxygen, suggesting the importance of the positive charge for
substrate recognition. This is in line with the observation that
the choline group forms salt bridges with E226 and D265 and
a cation–p interaction with F107 of AnkX.[8] Next, we focused
on the PEG spacer using derivatives 1, 4, and 5, decreasing the
spacer length stepwise from six PEG units to only two. To our
surprise, derivatives 1 (PEG6) and 4 (PEG4) were transferred
with an apparently similar efficiency. Only derivative 5
(PEG2), where the spacer clearly is too short, was not
transferred. Conveniently, the attachment of PC-PEGx-fluo-
rescein moieties to the protein substrates resulted in a pro-
nounced shift in electrophoretic mobility and thus permitted
quantification of its incorporation using gel densitometry of
the Coomassie-stained SDS-PAGE bands. Both 1 and 4 were
quantitatively transferred to Rab1b, indicating that full
modification of protein substrates is achieved using AnkX-
mediated phosphocholine transfer. Nucleotide 2 was a worse
substrate and modified only roughly 50% of the protein in
Figure 2. AnkX-catalyzed modification using CDP-choline derivatives.
240 min. We also obtained apparent catalytic efficiencies (kcat
/
A) Structures of synthesized CDP-choline derivatives carrying PEG-
fluorescein and containing different functional groups (1: quaternary
amine, 2: tertiary amine, 3: oxygen) or having different PEG linker
lengths (1, 4, 5). Fluorescein labeling of Rab1b3–174 (B) and TITSSYYR-
SUMO (C) with CDP-choline derivatives. The time course of phospho-
cholination was monitored through in-gel fluorescence of SDS-PAGE
gels. Successful phosphocholination was dependent on the linker
length (1, 4) and the presence of positive charge at the position of the
quaternary amine (1, 2, 4). D) Catalytic efficiencies for the modification
reaction with the various nucleotide derivatives were plotted relative to
the native reaction.
KM) from the time course experiments (Figure 2B,D). Here,
substrate 4 (PEG4) performed significantly better than
1 (PEG6), indicating that further optimization of the linker
length and linker chemistry could lead to additional improve-
ment of the substrate properties of CDP-choline derivatives.
In addition to Rab1b, we investigated the transfer of CDP-
choline derivatives 1, 2, and 4 to the TITSSYYR-SUMO
construct by detecting modified proteins with ESI-MS and in-
gel fluorescence (SDS-PAGE) (Figure 2C). The substrate
profile was identical to that of Rab1b, thus confirming that the
protein modification is dependent only on the presence of the
AnkX-recognition sequence TITSSYYR. However, the
extent of PC-PEGx-fluorescein attachment to TITSSYYR-
SUMO differed among the nucleotide substrates. The band
shift corresponding to the attachment of 1 indicated only
a slight incorporation into the proteins, whereas 4 modified
approximately 70% of TITSSYYR-SUMO in 180 min (Fig-
ure 2C). This is not surprising since the native AnkX substrate
Rab1b is a more efficient recipient of the modification than
the artificially generated peptide substrates. Still, the combi-
nation of TITSSYYR-tagged proteins with CDP-choline
derivatives (e.g. 4) allows for efficient labeling using AnkX
enzyme on a reasonable timescale.
Next we asked whether Lem3 could be utilized to detach
functional phosphocholine derivatives from modified target
proteins and whether the peptide sequence is important for
Angew. Chem. Int. Ed. 2015, 54, 10327 –10330
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 10329

.
Angewandte
Communications
recognition by Lem3.[7,11] Kinetic investigations revealed that
alanine substitutions except for the first threonine in the
octapeptide sequence (TITSSYYR) affected the Lem3 activ-
ity by 10–50% (Figure S9). We subjected fluorescently
phosphocholinated Rab1b3–174 and TITSSYYR-SUMO
(modified with constructs 1, 2, and 4) to Lem3-mediated
enzymatic dephosphocholination. In-gel fluorescence after
SDS-PAGE, and differences in the electrophoretic mobility of
modified versus unmodified proteins (Figure 3A,B) revealed
that Rab1b3–174 and TITSSYYR-SUMO could be effectively
dephosphocholinated by Lem3. It appears like the internal
loop region of Rab1b is the most efficient substrate for Lem3,
although TITSSYYR-SUMO reacts on a comparable time-
scale. This is reflected in similar catalytic efficiencies of Lem3
for the different PC-PEGx-fluorescein modified protein
substrates (Figure 3A,B).
In summary, we present a covalent enzymatic protein
labeling and delabeling method based on a small peptide
recognition motif. The labeling method is applicable to N-
terminal, C-terminal, and internal loop recognition motifs of
only eight amino acids in proteins of interest. The recognition
sequence is small compared to those of other labeling
techniques that require N- or C-terminal attachment of
entire protein domains (e.g. ACP,[4b] fluorescent proteins,[12]
SNAP-tag,[13] CLIP-tag,[14] etc.) also permitting internal label-
ing. The method is fully compatible with current existing
chemical labeling methodologies, since the recognition
sequence contains no cysteine or other reactive amino acids
and therefore can be used easily for double labeling
approaches. The possibility of conveniently detaching labels
using Lem3-mediated hydrolysis makes it possible to recover
protein samples and use them subsequently for another
labeling strategy. Given the importance of selective protein
labeling methods in biosciences, enzymatic phosphocholina-
tion for protein modification holds the potential to increase
the scope of possible labeling strategies in complex biological
systems.
Keywords: enzymes · nucleotides · phosphocholination ·
protein modifications
How to cite: Angew. Chem. Int. Ed. 2015, 54, 10327–10330
Angew. Chem. 2015, 127, 10467–10471
[1] a) D. M. Patterson, L. A. Nazarova, J. A. Prescher, ACS Chem.
Biol. 2014, 9, 592 – 605; b) E. M. Sletten, C. R. Bertozzi, Angew.
Chem. Int. Ed. 2009, 48, 6974 – 6998; Angew. Chem. 2009, 121,
7108 – 7133.
[2] M. Rashidian, J. K. Dozier, M. D. Distefano, Bioconjugate
Chem. 2013, 24, 1277 – 1294.
[3] a) R. H. Lambalot, A. M. Gehring, R. S. Flugel, P. Zuber, M.
LaCelle, M. A. Marahiel, R. Reid, C. Khosla, C. T. Walsh, Chem.
Biol. 1996, 3, 923 – 936; b) J. Yin, P. D. Straight, S. M. McLough-
lin, Z. Zhou, A. J. Lin, D. E. Golan, N. L. Kelleher, R. Kolter,
C. T. Walsh, Proc. Natl. Acad. Sci. USA 2005, 102, 15815 – 15820;
c) Z. Zhou, P. Cironi, A. J. Lin, Y. Q. Xu, S. Hrvatin, D. E.
Golan, P. A. Silver, C. T. Walsh, J. Yin, ACS Chem. Biol. 2007, 2,
337 – 346; d) Z. Zhou, A. Koglin, Y. Wang, A. P. McMahon, C. T.
Walsh, J. Am. Chem. Soc. 2008, 130, 9925 – 9930.
[4] a) J. Yin, A. J. Lin, D. E. Golan, C. T. Walsh, Nat. Protoc. 2006, 1,
280 – 285; b) N. M. Kosa, R. W. Haushalter, A. R. Smith, M. D.
Burkart, Nat. Methods 2012, 9, 981 – 984.
[5] N. M. Kosa, K. M. Pham, M. D. Burkart, Chem. Sci. 2014, 5,
1179 – 1186.
[6] a) S. Mukherjee, X. Y. Liu, K. Arasaki, J. McDonough, J. E.
Galan, C. R. Roy, Nature 2011, 477, 103 – 106; b) C. A. Hardi-
man, C. R. Roy, Mbio 2014, 5, e01035-13; c) X. X. Pan, A.
Lührmann, A. Satoh, M. A. Laskowski-Arce, C. R. Roy, Science
2008, 320, 1651 – 1654; d) M. P. Müller, H. Peters, J. Blümer, W.
Blankenfeldt, R. S. Goody, A. Itzen, Science 2010, 329, 946 – 949.
[7] a) P. R. Goody, K. Heller, L. K. Oesterlin, M. P. Müller, A. Itzen,
R. S. Goody, EMBO J. 2012, 31, 1774 – 1784; b) Y. H. Tan, R. J.
Arnold, Z. Q. Luo, Proc. Natl. Acad. Sci. USA 2011, 108, 21212 –
21217.
[8] V. Campanacci, S. Mukherjee, C. R. Roy, J. Cherfils, EMBO J.
2013, 32, 1469 – 1477.
[9] S. Schoebel, L. K. Oesterlin, W. Blankenfeldt, R. S. Goody, A.
Itzen, Mol. Cell 2009, 36, 1060 – 1072.
[10] a) C. M. Crane, J. Kaiser, N. L. Ramsden, S. Lauw, F. Rohdich,
W. Eisenreich, W. N. Hunter, A. Bacher, F. Diederich, Angew.
Chem. Int. Ed. 2006, 45, 1069 – 1074; Angew. Chem. 2006, 118,
1082 – 1087; b) J. G. Moffatt, H. G. Khorana, J. Am. Chem. Soc.
1961, 83, 649 – 658.
[11] M. F. Albers, C. Hedberg, J. Org. Chem. 2013, 78, 2715 – 2719.
[12] N. C. Shaner, P. A. Steinbach, R. Y. Tsien, Nat. Methods 2005, 2,
905 – 909.
[13] A. Keppler, S. Gendreizig, T. Gronemeyer, H. Pick, H. Vogel, K.
Johnsson, Nat. Biotechnol. 2002, 21, 86 – 89.
[14] A. Gautier, A. Juillerat, C. Heinis, I. R. Correa, M. Kindermann,
F. Beaufils, K. Johnsson, Chem. Biol. 2008, 15, 128 – 136.
Figure 3. Enzymatic dephosphocholination by Lem3. Lem3 dephospho-
cholination of protein substrates Rab1b3–174 (A) and TITSSYYR-
SUMO (B) preparatively modified with fluorescent phosphocholine
derivatives, analyzed by in-gel fluorescence of SDS-PAGE (gray scale)
and Coomassie staining (blue). Yellow star: fluorescein.
Received: March 21, 2015
Published online: July 3, 2015
10330 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 10327 –10330