Affinity-Based Tagging of Protein Families with Reversible Inhibitors: A Concept for Functional Proteomics

Article abstract of DOI:10.1002/anie.200352084

Kinases can be tagged with an engineered chemical probe that comprises a reversibly binding protein ligand, a photoreactive group (4-benzoylphenylalanine), and a fluorescent reporter group (carboxyfluorescein). This principle, which does not require irrev

Full text of DOI:10.1002/anie.200352084

Angewandte
Chemie
Functional Proteomics
Affinity-Based Tagging of Protein Families with
Reversible Inhibitors: A Concept for Functional
Proteomics**
Figure 1. Mechanism-based tagging of proteins with a) irreversible and
b) reversible inhibitors.
Miriam C. Hagenstein, Jan H. Mussgnug, Kirsten Lotte,
Regina Plessow, Andreas Brockhinke, Olaf Kruse,*and
Norbert Sewald*
that irreversible enzyme inhibitors are required, which
severely limits its applicability as irreversibly binding ligands
are not known for many classes of proteins. As separation of
protein mixtures by 2D-PAGE (2D polyacrylamide gel
electrophoresis) usually involves denaturing conditions, the
vast majority of protein families can only be addressed by
substantial modification of this approach.
Dedicated to Professor Hans-Dieter Jakubke
on the occasion of his 70th birthday
Proteomics, one of the most powerful techniques of the
postgenomic era, relies on the two-dimensional separation of
proteins.[1] However, as more than 10000 different proteins
are usually expressed by a cell at the same time, resolution
often remains incomplete. Sensitive detection and quantifi-
cation represent further problems, because conventional
staining methods are nonselective and sometimes difficult to
reproduce. Detection methods with fluorescent dyes are
highly sensitive, but quantification is problematic, because the
dye is covalently linked to the proteins in an unselective
manner prior to the separation step. Uniform labeling is not
possible, and extensive labeling may lead to fluorescence
quenching and solubility problems.[2] Consequently, selective
labeling with reporter groups (e.g. fluorescent labels, radio-
active tags, or biotin) on a mechanistic basis is desirable.
Selective labeling is possible when the reporter groups are
conjugated to irreversible enzyme inhibitors and, hence, form
covalent links to enzymes in the course of the enzyme
reaction. Activity-based profiling has proven to be amenable
for serine proteases with fluorophosphonates[3] and cysteine
proteases with epoxides[4] as the irreversible inhibitors
(Figure 1a). This intriguing concept suffers from the fact
We have rendered this strategy generally applicable for
protein profiling in proteomics. As a covalent bond between
the proteins and the reporter group is necessary to prevent
dissociation under the conditions of 2D electrophoresis, an
additional chemical step is required. Only a few chemical
reactions are appropriate for this kind of protein modifica-
tion; among them are photoreactions, which are well-known
in biochemistry and widely used for photoaffinity labeling.[5]
An engineered chemical probe was designed on this basis,
comprising a (semi-)specific, reversibly binding protein ligand
(inhibitor) linked to a reporter group and a reactive group
(photoaffinity label) (Figure 1b). This concept is suited for
many different classes of proteins and may facilitate the
discovery of new members of a protein family. Any method
capable of decreasing the amount of data and detecting
mechanistically related proteins in 1D- or 2D-PAGE will be
helpful for the retrieval of hitherto unknown proteins and for
the future development of proteome research. To prove the
principle of our concept, the class of isoquinolinesulfona-
mides of the H series (H-8, H-9, etc.)[6] was envisaged as the
first model inhibitors.
Isoquinolinesulfonamides competitively inhibit a broad
range of kinases, including protein kinases, by occupying the
ATP-binding sites. Kinases play a major role in many
regulatory mechanisms of living cells. Phosphorylation/
dephosphorylation reactions trigger important metabolic or
pathologic pathways and participate in internal and external
adaptation mechanisms. Recent advances in protein kinase
research rely on systematic analyses of completely sequenced
genomes, but the correlation between genome and proteome
remains difficult.
[*] Prof. Dr. N. Sewald, Dipl.-Chem. M. C. Hagenstein
Department of Chemistry, Organic and Bioorganic Chemistry
University of Bielefeld
Universitätsstrasse 25, 33615 Bielefeld (Germany)
Fax:(+49)521-106-8094
E-mail: norbert.sewald@uni-bielefeld.de
Dr. O. Kruse, Dipl.-Biol. J. H. Mussgnug
Department of Biology, Molecular Cell Physiology
University of Bielefeld
Universitätsstrasse 25, 33615 Bielefeld (Germany)
Fax: (+49)521-106-6410
E-mail: olaf.kruse@uni-bielefeld.de
Our initial experiments are directed towards the detection
of plant kinases. Despite the progress in understanding
detailed functions of plant protein kinases, it is still very
difficult to assign the vast amount of recently identified kinase
genes by systematic genomics. Serine/threonine kinases can
be divided into more than twelve groups based on the
sequence relationships and are of special interest in plants. In
particular, chloroplasts are reported to contain several serine/
threonine kinases that are functionally involved in adaptation
mechanisms as biological responses to changes in environ-
mental conditions (e.g. light).[7–9] The sensitivity of protein
Dipl.-Chem. K. Lotte, Dipl.-Chem. R. Plessow, Dr. A. Brockhinke
Department of Chemistry, Physical Chemistry
University of Bielefeld
Universitätsstrasse 25, 33615 Bielefeld (Germany)
[**] This project was supported by the University of Bielefeld and the
Fonds der Chemischen Industrie. The authors thank Professor Dr. J.
Mattay for providing access to photochemical equipment, Dr. K.
Niehaus and N. Kuepper for MALDI TOF MS analysis, Dr. M. Letzel
for ESIMS analysis, and Dr. K. Stembera for helpful discussions
regarding SPR analysis.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2003, 42, 5635 –5638
DOI: 10.1002/anie.200352084
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5635

Communications
phosphorylation of the light-harvesting proteins (LHC) and
secondary amino functions with a Boc group gives the
orthogonally protected linker 3. Hydrazinolysis of the Dde
groups yields 4, which reacts with 5-isoquinolinesulfonyl
chloride to give 5. The isoquinolinesulfonamide 5 is then
coupled to Fmoc-protected l-4-benzoylphenylalanine (Fmoc-
Bpa-OH). The free amine 6 is obtained by cleavage of the
Fmoc group and then coupled to the isomeric mixture of 5-
and 6-carboxyfluorescein. Any other dye or any other
reporter group can be used. In the final step, the Boc
groups are removed to yield the target molecule 7.
The kinase-binding ability of modified H-9 7 was proved
by surface plasmon resonance experiments (Figure 2). The
conjugate 8, an immobilizable analogue of 7 (Scheme 1), was
synthesized by coupling 6 to Fmoc-6-aminohexanoic acid,
followed by complete deprotection.
It was proven that protein kinase A (PKA) and creatine
kinase[15] bind in a concentration-dependent manner (Fig-
ure 2a). Binding could be reduced or even totally eliminated
by preincubation of the corresponding protein with a series of
soluble ligands such as H-9 (1a; Figure 2b), 7, and 8.[15] These
experiments clearly prove that conjugation of H-9 (1a) with
Bpa and reporter groups does not significantly affect binding
to the model kinases. The KD value for the interaction of PKA
with immobilized 8[16] was calculated to be (2 Æ 3) ” 10À7 m,
which is similar to the KD value reported for 1a and PKA
(1.9 ” 10À6 m).[12] Moreover, unspecific binding of a series of
non-kinase proteins (hemoglobin, ovalbumin, bovine serum
photosystem II proteins to inhibitors of kinase C and cAMP-
dependent kinases has been described.[10,11] The isoquino-
linesulfonamide-type inhibitor H-9 (1a)[12] inhibits the phos-
phorylation of LHCII proteins at concentrations of about
50–100 mm, presumably by interacting with the ATP-binding
site of membrane-associated threonine kinases. No unambig-
uous proof of a kinase in Chlamydomonas reinhardtii
thylakoid membranes existed before a kinase that is assumed
to be involved in LHC phosphorylation was identified only
recently by a genome-based approach.[13]
The conjugate 7, which consists of H-9 with a polar linker,
a photoreactive group, and a fluorophore, was designed as an
engineered chemical probe to identify kinases after covalent
tagging and protein separation by SDS-PAGE (SDS = sodium
dodecyl sulfate). The linker attachment was chosen on the
basis of the crystal structure of a protein kinase–inhibitor
complex (with H-8, 1b) in such a way that inhibitor binding
would only be minimally affected by the conjugation.[14] 4-
Benzoylphenylalanine was selected as the photoreactive
group, because it can be activated reversibly by excitation–
relaxation cycles, and the triplet state exclusively inserts into
À
C H bonds within a 3.1-Š distance of the oxygen atom of the
carbonyl group. It is stable in common protic solvents and
leads to efficient single-site modification.[5]
The synthesis of 7 starts with protection of the primary
amino groups in 2 with Dde (Scheme 1). Protection of the
Scheme 1. Synthesis of the chemical probe 7, which comprises the kinase inhibitor moiety H-9 (1a), the photoreactive group 4-benzoylphenylala-
nine, and the fluorescent reporter group 5-/6-carboxyfluorescein. Conjugate 8 is a derivative of 7 that can be immobilized to surfaces, for example,
on a surface plasmon resonance sensor chip. Reagents and Conditions: a) 2-acetyldimedone, DMF, 74%; b) Boc2O, CH2Cl2, 89%; c) N2H4·H2O,
CH2Cl2, 45%; d) 5-isoquinolinesulfonyl chloride·HCl, NEt3, CHCl3, 63%; e) EDC·HCl, NMM, Fmoc-Bpa-OH, HOBt·H2O, CH2Cl2, 68%; f) 4-(amino-
methyl)piperidine, CH2Cl2, 85%; g) EDC·HCl, NMM, 5-/6-carboxyfluorescein (isomeric mixture), HOBt·H2O, CH2Cl2/DMF (1:1), 45%; h) TFA
(30% in CH2Cl2), 99%; i) EDC·HCl, NMM, Fmoc-6-aminohexanoic acid, HOBt·H2O, CH2Cl2, 52%; j) 4-(aminomethyl)piperidine, CH2Cl2, 92%;
k) TFA (30% in CH2Cl2), 98%. Dde=2-(4,4-dimethyl-2,6-dioxocyclohexylidene)ethyl; DMF=N,N-dimethylformamide, Boc=tert-butoxycarbonyl,
EDC=1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, NMM=N-methylmorpholine, Fmoc=9-fluorenylmethyloxycarbonyl, HOBt=1-hydroxy-1H-
benzotriazole.
5636
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 5635 –5638

Angewandte
Chemie
Figure 3. Specific photoaffinity labeling of different kinases with the
H-9 derivative 7. Left: silver stain; right: fluorescence image. M: pre-
stained protein marker; Lane 1: hexokinase; lane 2: creatine kinase;
lane 3: 3PGA phosphokinase; lane 4: cAMP-dependent protein
kinase A.
(Figure 4A, lanes 2 and 4). The drastic decrease in fluores-
cence of this preincubated preparation (Figure 4A, lane 4)
relative to the reference sample (Figure 4A, lane 3) clearly
demonstrates that photoaffinity labeling with 7 occurs
specifically, as the SDS-denatured protein was not tagged.
Figure 2. Surface plasmon resonance sensorgrams: a) concentration-
dependent binding of cAMP-dependent protein kinase A (PKA) to
immobilized 8: 1) 120 nm, 2) 60 nm, 3) 30 nm, 4) 15 nm, 5) 7.5 nm;
KD =(2Æ3)ꢀ10À7 m. b) Competition with H-9 (1a): 6) 24 nm PKA,
7) 24 nm PKA+19 nm H-9, 8) 24 nm PKA+184 nm H-9, 9) 24 nm
PKA+1.85 mm H-9. RU=Response Unit.
albumin, and a-glucosidase) to immobilized 8 was shown to
be negligible.
Excitation–emission spectroscopy (EES) was used to
study the effects of potential radiation-free fluorescence
resonance energy transfer (FRET) in 7, particularly between
the benzophenone triplet state and the neighboring fluores-
cein.[15,17] We found the fluorescence quantum yield to be very
low for both 1a and 4-benzoylphenylalanine (FFl ꢀ 10À4).
By comparing normalized emission spectra of 7 and of the
analogous compound without a photoreactive group, no
significant FRET could be found at lexc = 350 nm
(FFRET410À2). Energy transfer between H-9 and fluorescein
was found to be negligible in all cases (FFRET410À4). There-
fore, quenching and chemical reactions are the dominant
channels of loss after photoactivation at this wavelength.
In the next step, purified proteins were used for tagging
experiments with 7. Four commercially available kinases were
compared to prove the specificity of the photoaffinity labeling
with the photoreactive and fluorescent inhibitor 7. Three
kinases (Figure 3, lanes 1, 2, and 4) were fluorescently tagged
by photoaffinity labeling with 7 to a high extent, whereas the
fourth (Figure 3, lane 3), although present in a higher
concentration, exhibited a very low binding tendency for 7.
The specificity of photoaffinity labeling of a particular kinase
is highlighted by the fact that in the case of the hexokinase
and the cAMP-dependent protein kinase A assays, only the
catalytic unit is tagged, whereas the regulatory unit remains
unlabeled.
Figure 4. Selective kinase tagging by photoaffinity labeling with deriva-
tive 7. a) Comparison of photoaffinity labeling efficiency of native crea-
tine kinase with labeling of denatured creatine kinase. Lane 1: silver
stain, native creatine kinase; Lane 2: silver stain, denatured creatine
kinase; Lane 3: fluorescence image, native creatine kinase; Lane 4:
fluorescence image, denatured creatine kinase. b) Retrieval of added
creatine kinase from thylakoid membrane preparations after photoaf-
finity labeling. Lane 5: silver stain; Lane 6: silver stain, with additional
150 ng creatine kinase; Lane 7: fluorescence image; Lane 8: fluores-
cence image, with additional 150 ng creatine kinase (see arrow).
In a second approach creatine kinase was added to
isolated thylakoid proteins prior to labeling with 7. The results
demonstrate that the H-9 derivative 7 targeted several
thylakoid membrane proteins predominantly in a molecular
mass range between 40 and 80 kDa (Figure 4B, lane 7). The
creatine kinase that was added to the thylakoid membrane
preparation could be found again (Figure 4B, lane 8). The
subsequent identification of the labeled protein band at
45 kDa by MALDI TOF MS fingerprint analysis[15] as the
supplemented creatine kinase clearly demonstrates the func-
tionality of the tagging system in combination with mass
To prove the feasibility of our concept for the affinity-
based identification of putative kinases, creatine kinase was
treated with SDS prior to incubation and photoreaction with 7
Angew. Chem. Int. Ed. 2003, 42, 5635 –5638
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5637

Communications
spectrometry.[18] The identification and characterization of the
other labeled proteins is currently underway.
[16] Compound 8 was immobilized on a BIAcore CM5 sensor chip by
activation of surface-bound carboxy groups with EDC/NHS. It
can be assumed that amide bond formation proceeds predom-
inantly through the primary amino group.
[17] A. Brockhinke, R. Plessow, P. Dittrich, K. Kohse-Hꢀinghaus,
Appl. Phys. B 2000, 71, 755 – 763.
Notably, the selectivity of the labeling process with 7 is
also shown by the fact that the LHC proteins (Figure 4B,
proteins with molar masses < 33 kDa) are not labeled. To
exclude the possibility that other photochemical processes
(e.g. photooxidation) prevail over the desired CH insertion of
the benzophenone triplet, photoaffinity control experiments
with a conjugate that lacks the inhibitor moiety (linker-Bpa-
carboxyfluorescein) and with carboxyfluorescein alone were
carried out.[15]
[18] M. Hippler, J. Klein, A. Fink, T. Allinger, P. Hoerth, Plant J.
2001, 28, 595 – 606.
In summary, the engineered chemical probe 7, which
comprises the reversibly binding protein ligand H-9, a
fluorescent reporter group, and a photoreactive group,
allows affinity-based tagging of kinases. Selective detection
of kinases in functional proteomics is now possible. The
concept is generally applicable to affinity-based tagging of
proteins and will be exploited in the first instance for
plastidial plant serine/threonine kinases by using new tech-
niques[18] for thylakoid protein separation by 2D gel electro-
phoresis followed by isolation and further identification. The
presence of basic groups in 7 may change the pI value of the
protein after the tagging step. This could lead to a decoupling
of tagged and untagged proteins in the isoelectric focusing
domain of 2D gel electrophoresis. Further experiments to
improve this shortcoming are currently in progress. However,
the concept may, in its current state, be applied to the
discovery of new kinases.
Received: June 10, 2003
Revised: September 15, 2003 [Z52084]
Keywords: fluorescent probes · inhibitors · kinases ·
.
photoaffinity labeling · proteomics
[1] F. Lottspeich, Angew. Chem. 1999, 111, 2630 – 2647; Angew.
Chem. Int. Ed. 1999, 38, 2476 – 2492.
[2] Proteome Research: Two-Dimensional Gel Electrophoresis and
Identification Methods (Ed.: T. Rabilloud), Springer, Heidel-
berg, 2000.
[3] Y. Liu, M. P. Patricelli, B. F. Cravatt, Proc. Natl. Acad. Sci. USA
1999, 96, 14694 – 14699.
[4] D. Greenbaum, K. F. Medzihradszky, A. Burlingame, M. Bogyo,
Chem. Biol. 2000, 7, 569 – 581.
[5] G. Dormµn, G. N. Prestwich, Trends Biotechnol. 2000, 18, 64 – 77.
[6] N. Ono-Saito, I. Niki, H. Hidaka, Pharmacol. Ther. 1999, 82,
123 – 131.
[7] A. Gal, H. Zer, I. Ohad, Physiol. Plant. 1997, 100, 869 – 885.
[8] J. F. Allen, Biochim. Biophys. Acta 1992, 1098, 275 – 335.
[9] O. Kruse, Naturwissenschaften 2001, 88, 284 – 292.
[10] Z. Ishak, I. R. White, P. A. Millner, Biochem. Soc. Trans. 1992,
20, 8S.
[11] J. Farchaus, R. A. Dilley, W. A. Cramer, Biochim. Biophys. Acta
1985, 809, 17 – 26.
[12] H. Hidaka, M. Inagaki, S. Kawamoto, Y. Sasaki, Biochemistry
1984, 23, 5036 – 5041.
[13] N. Dep›ge, S. Bellfiore, J.-D. Rochaix, Science 2003, 299, 1572 –
1575.
[14] R. A. Engh, A. Girod, V. Kinzel, R. Huber, D. Bossemeyer, J.
Biol. Chem. 1996, 271, 26157 – 26164.
[15] See Supporting Information.
5638
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 5635 –5638