Synthesis and evaluation of bioorthogonal pantetheine analogues for in vivo protein modification

Article abstract of DOI:10.1021/ja063217n

In vivo carrier protein tagging has recently become an attractive target for the site-specific modification of fusion systems and new approaches to natural product proteomics. A detailed study of pantetheine analogues was performed in order to identify su

Full text of DOI:10.1021/ja063217n

Published on Web 08/26/2006
Synthesis and Evaluation of Bioorthogonal Pantetheine
Analogues for in Vivo Protein Modification
Jordan L. Meier, Andrew C. Mercer, Heriberto Rivera, Jr., and Michael D. Burkart*
Contribution from the Department of Chemistry and Biochemistry, UniVersity of California,
San Diego, 9500 Gilman DriVe, La Jolla, California 92093-0358
Received May 8, 2006; E-mail: mburkart@ucsd.edu
Abstract: In vivo carrier protein tagging has recently become an attractive target for the site-specific
modification of fusion systems and new approaches to natural product proteomics. A detailed study of
pantetheine analogues was performed in order to identify suitable partners for covalent protein labeling
inside living cells. A rapid synthesis of pantothenamide analogues was developed and used to produce a
panel which was evaluated for in vitro and in vivo protein labeling. Kinetic comparisons allowed the
construction of a structure-activity relationship to pinpoint the linker, dye, and bioorthogonal reporter of
choice for carrier protein labeling. Finally bioorthogonal pantetheine analogues were shown to target carrier
proteins with high specificity in vivo and undergo chemoselective ligation to reporters in crude cell lysate.
The methods demonstrated here allow carrier proteins to be visualized and isolated for the first time without
the need for antibody techniques and set the stage for the future use of carrier protein fusions in chemical
biology.
domain down to just 11 amino acids,7 offering a fusion tag of
the size and flexibility to be competitive with contemporary
Introduction
Recent years have seen intense research effort focused toward
the development of new methods for the study and manipulation
of covalently modified proteins, with particular attention given
to in vivo methodologies.1 Fluorescent protein fusions2 and
antibody conjugates3 provide powerful tools for protein imaging
and manipulation. However drawbacks of these methods, such
as structural perturbations sometimes induced by large fusions
and general membrane impermeability of antibodies, have lead
researchers to devise methods for the site-specific modification
of proteins by small-molecule probes. Ideally these probes
should be low molecular weight, covalent in nature, and
possessed of fluorescence or affinity properties allowing for
facile imaging and manipulation. We recently introduced one
such technique, demonstrating cellular uptake and covalent
modification of carrier protein fusions by pantetheine analogues.4
These coenzyme A (CoA) precursors were shown to penetrate
the cell membrane and be transformed into fully formed CoA
derivatives via the endogenous CoA metabolic pathway, where-
upon they were transferred to a carrier protein by the promiscu-
ous phosphopantetheinyltransferase (PPTase; E.C. 2.7.7.7) Sfp
(Figure 2). This advance allows carrier protein labeling, a
technique first developed from cell lysates5 and since demon-
strated on the cell surface,6 to be performed within the cell,
opening the door for more sophisticated labeling systems. Recent
developments have seen the trimming of the carrier protein
tagging systems and further highlighting the importance of
techniques for the labeling of intracellular carrier proteins.
Several strategies for site-specific labeling of proteins in vivo
have been previously demonstrated. Examples include Bertozzi’s
manipulation of the sialic acid biosynthetic pathway for the
introduction of keto and azido functionalized cell-surface
glycoproteins,8 Cravatt’s introduction of azido/alkyne function-
alities by covalent irreversible inhibition of protein active sites,9
and Hsieh-Wilson’s chemoenzymatic introduction of a keto-
functionality for capture of O-GlcNAc-modified proteins.10 In
each of these examples the protein is not directly labeled with
a fluorescence or affinity tag, but rather a unique and biologi-
cally inert chemical functionality is introduced. This functional-
(5) (a) La Clair, J. J.; Foley, T. L.; Schegg, T. R.; Regan, C. M.; Burkart, M.
D. Chem. Biol. 2004, 11, 195-201. (b) Yin, J.; Liu, F.; Li, X.; Walsh, C.
T. J. Am. Chem. Soc. 2004, 126, 7754-7755. (c) Yin, J.; Liu, F.; Schinke,
M.; Daly, C.; Walsh, C. T. J. Am. Chem. Soc. 2004, 126, 13570-13571.
(6) (a) George, N.; Pick, H.; Vogel, H.; Johnsson, N.; Johnsson, K. J. Am.
Chem. Soc. 2004, 126, 8896-8897. (b) Yin, J.; Lin, A. J.; Buckett, P. D.;
Wessling-Resnick, M.; Golan, D. E.; Walsh, C. T. Chem. Biol. 2005, 12,
999-1006. (c) Vivero-Pol L.; George, N.; Krumm, H.; Johnsson, K.;
Johnsson, N. J. Am. Chem. Soc. 2005, 127, 12770-12771.
(7) Yin, J.; Straight, P. D.; McLoughlin, S. M.; Zhou, Z.; Lin, A. J.; Golan, D.
E.; Kelleher, N. L.; Kolter, R.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 15815-15820.
(8) (a) Mahal, L. K.; Yarema, K. J.; Bertozzi, C. R. Science 1997, 276, 1125-
1128. (b) Yarema, K. J.; Mahal, L. K.; Bruehl, R.; Rodriguez, E. C.;
Bertozzi, C. R. J. Biol. Chem. 1998, 273, 31168-31179. (c) Saxon, E.;
Bertozzi, C. R. Science 2000, 287, 2007-2010.
(9) (a) Speers, A. E.; Cravatt, B. F. Chem. Biol. 2004, 11, 535-546. (b) Speers,
A. E.; Adam, G. C.; Cravatt, B. F. J. Am. Chem. Soc. 2003, 125, 4686-
4687. (c) Alexander, J. P.; Cravatt, B. F. Chem Biol. 2005, 12, 1179-
1187.
(1) Bertozzi, C. R.; Prescher, J. A. Nat. Chem. Biol. 2005, 1, 13-21.
(2) (a) Tsien, R. Y. Annu. ReV. Biochem. 1998, 67, 509-544. (b) Lippincott-
Schwartz, J.; Patterson, G. H. Science 2003, 300, 87-91.
(3) (a) Fritze, C. E.; Anderson, T. R. Methods Enzymol. 2000, 327, 3-16. (b)
Massoud, T. F.; Gambhir, S. S. Genes DeV. 2003, 17, 545-580.
(4) Clarke, K. M.; Mercer, A. C.; La Clair, J. J.; Burkart, M. D. J. Am. Chem.
Soc. 2005, 127, 11234-11235.
(10) (a) Hwan-Ching, T.; Khidekel, N.; Ficarro, S. B.; Peters, E. C.; Hsieh-
Wilson, L. C. J. Am. Chem. Soc. 2004, 126, 10500-10501. (b) Khidekel,
N.; Ficarro, S. B.; Peters, E. C.; Hsieh-Wilson, L. C. Proc. Natl. Acad.
Sci. U.S.A. 2004, 36, 13132-13137.
9
12174
J. AM. CHEM. SOC. 2006, 128, 12174-12184
10.1021/ja063217n CCC: $33.50 © 2006 American Chemical Society

In Vivo Bioorthogonal Carrier Protein Modification
A R T I C L E S
Figure 1. Structures of pantetheine analogues and biotin detection agents used in this study.
ity can then undergo reaction with exogenously delivered
reporters to label the protein of interest for detection and/or
isolation, depending on the nature of the reporter. Advantages
of this two-step labeling process include (i) better uptake of
smaller probes due to increased membrane permeability, (ii)
increased incorporation of probes into native biosynthetic
pathways due to a greater similarity to natural substrate, and
(iii) the ability to conjugate a protein to virtually any reporter
possessing reactivity with the bioorthogonal functionality.1,9
Here we present a full study of simplified pantetheine analogues
that harness the power of such bioorthogonal ligation reactions.
First we optimize the synthesis of simplified pantetheine
analogues via a one-step reaction with pantolactone. Next, the
specificity of the CoA biosynthetic pathway is probed by a small
panel of these simplified substrates. Finally, we validate the
utility of this strategy by demonstrating and comparing the
delivery of bioorthogonal chemical functionalities to carrier
proteins in vitro and in vivo and using the newly tagged carrier
proteins for two widely applied chemoselective ligations: the
reaction of ketones and hydroxylamines to form oximes and
the Cu(I)-catalyzed azide-alkyne [3 + 2] cycloaddition reaction
(“click” chemistry). This new ability to manipulate a bioor-
thogonally tagged carrier protein in vivo promises to be a
valuable tool for both new approaches to natural product
proteomics as well as the study of novel intracellular carrier
protein fusion systems.
the presumption was made that the identities of cystamine and
â-alanine were necessary for turnover by the CoA metabolic
pathway. However our own in vitro studies and recent work by
Lee12 questioned the degree of selectivity gained through
specific interactions between the â-alanine moiety of pantoth-
enate and PANK (used in this manuscript to denote all enzymes
with pantothenate kinase activity; E.C. 2.7.1.33), the first
enzyme in the CoA biosynthetic pathway and gatekeeper for
downstream metabolism.13 Further, we found that E. coli PANK
(CoaA) catalyzed the phosphorylation of pantetheine and
analogues with variations at the cystamine moiety almost as
well as pantothenate itself.14 Given this newly revealed permis-
siveness in the CoA biosynthetic pathway, we reasoned that
simplified analogues of pantetheine could be used for both in
vitro and in vivo applications. Elimination of the amide bond
between cystamine and â-alanine significantly simplifies syn-
thetic access to reporter-modified pantetheine analogues by
reducing overall molecule polarity and solubility issues, elimi-
nating time-consuming protection/deprotection steps of the 1,3-
diol, and replacing the multiple peptide coupling and purification
steps of previous syntheses4 with a simple one-step nucleophilic
ring-opening of pantolactone. With this in mind we chose to
mimic the aletheine moiety (N-(â-alanyl)-â-aminoethanethiol)
with more synthetically flexible poly(ethylene glycol) (PEG)
linkers. In addition to the synthetic utility of this substitution,
PEG spacers have the advantages of increasing the aqueous
solubility of small molecule probes and distancing reporter labels
from a labeled protein with the effect of both enhancing
secondary detection properties and reducing any negative effect
of reporter/protein interactions.15 We sought to incorporate these
Results and Discussion
Analogue Synthesis: Pantolactone-Ring Opening. In our
efforts to address CoA biosynthesis with novel analogues, we
had initially investigated the synthesis of pantetheine and
phosphopantetheine derivatives that could be assembled in a
manner analogous to peptide library synthesis.11 This neces-
sitated addressing the synthetic challenges associated with
pantolactone, namely the lability of the R-proton following
protection of the pantolactone secondary alcohol. At this point,
(11) Mandel, A. L.; La Clair, J. J.; Burkart, M. D. Org. Lett. 2004, 6, 4801-
4803.
(12) Virga, K. G.; Zhang, Y. M.; Leonardi, R.; Ivey, R. A.; Hevener, K.; Park,
H. W.; Jackowski, S.; Rock, C. O.; Lee, R. E. Bioorg. Med. Chem. 2006,
14, 1007-1020.
(13) Jackowski, S.; Rock, C. O. J. Bacteriol. 1981, 148, 926-932.
(14) Worthington, A. S.; Burkart, M. D. Org. Biomol. Chem. 2005, 4, 44-46.
(15) Kumar, V.; Aldrich, J. V. Org. Lett. 2003, 5, 613-616.
9
J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006 12175

A R T I C L E S
Meier et al.
Figure 2. General strategy for in vivo labeling of carrier protein by pantetheine analogues. Virtually any monoprotected amine (21) can be transformed into
a pantetheine analogue (25) by the three-step coupling/deprotection/ring-opening sequence. Cellular uptake and biosynthetic processing by CoaA, CoaD, and
CoaE yield the CoA analogue 28, which is then transferred to the carrier protein by a PPTase to yield bioorthogonally labeled carrier protein 28. After cell
lysis this carrier protein can now be conjugated to the reporter of choice via an appropriate chemoselective ligation reaction.
advantages into the design of an ideal, synthetically straight-
forward, biodetectible pantetheine analogue.
transition state of this reaction, protic solvents proved ideal for
nucleophilic ring-opening, with ethanol providing the best
balance of energy-absorbance and solubilization. Moving from
our model system to usefully functionalized amines, it was
shown that alkyne (41), PEG (44), and fluorophore (20)
containing pantetheine analogues could be synthesized in good
to moderate yields within 30 min using microwave assistance.
Interestingly a very recent report also presented ethanol as the
solvent of choice for this transformation under thermal condi-
tions; however without the addition of any base these large-
scale syntheses suffered from very long reaction times (72-
120 h).18 Accordingly our ideal microwave reaction conditions
were also tested under simple reflux. Triethylamine proved to
be a sufficient base, as replacement with Hunig’s base showed
no significant effect on the reaction outcome. Stronger bases
were avoided. Again it was found that reflux of (S)-(-)-R-
methylbenzylamine with excess pantolactone and triethylamine
provided pantetheine analogues with no apparent racemization
in excellent yields in 7-12 h. This alternative synthesis provides
To quickly access a large selection of analogues it was
deemed appropriate to first revise the current methodology for
pantothenamide synthesis. Previous protocols calling for the
base-promoted nucleophilic ring opening of pantolactone by an
amine could be subject to racemization or hampered by long
reaction times (>24 h).12,16,17 To address these problems we
turned to microwave-assisted organic synthesis. By using (S)-
(-)-R-methylbenzylamine we were able to test a variety of
conditions for their ability to open pantolactone with a fairly
1
hindered chiral nucleophile and analyze enantiopurity by H
NMR (Table 1, see Supporting Information for 1H NMR data).
The study showed pantolactone to be surprisingly robust to
a variety of conditions, and reaction times could be reduced
nearly 50-fold compared to previous preparations with a
retention of optical purity. As expected from the hypothesized
(16) Dueno, E. E.; Chu, F.; Kim, S. I.; Jung, K. W. Tetrahedron Lett. 1999, 40,
1843-1846.
(17) (a) Michelson, A. M. Biochim. Biophys. Acta 1964, 93, 71-77. (b) Moffatt,
J. G.; Khorana, H. G. J. Am. Chem. Soc. 1961, 83, 663-675.
(18) Krause, B. R. et al. Synth. Commun. 2006, 36, 365-391.
9
12176 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006

In Vivo Bioorthogonal Carrier Protein Modification
A R T I C L E S
Table 1. Data Table for One-Step Synthesis of Pantetheine Analogues via Nucleophilic Ring-Opening of Pantolactone
a Microwave-assisted. b Thermal condition.
another avenue for analogue preparation in cases where the
reporter or linker is sensitive to decomposition under microwave
conditions.19
Synthesis of Bioorthogonal and Fluorescent Pantetheine
Analogues. The general strategy for chemoenzymatic synthesis
of CoA analogues is depicted in Figure 2. First a monoprotected
amine (depicted in this example by N-Boc ethylenediamine 21)
is conjugated to the biodetectible tag of choice by standard
peptide coupling conditions. After deprotection, this amine can
be conjugated to either pantolactone 24 through nucleophilic
ring opening or pantothenic acid via EDAC mediated coupling.
The newly formed pantetheine analogue is then processed via
(19) Microwave-assisted conjugation of 7-dimethylaminocoumarin-4-acetic acid
containing amines to pantolactone resulted in the formation of unidentified
decomposition products. For these couplings the classical condition (MeOH,
NEt3, reflux) was used.
9
J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006 12177

A R T I C L E S
Meier et al.
Scheme 1. Synthesis of CoaA Probesa
a (a) EDAC (2 equiv), DIPEA (2 equiv), DMF, rt 12 h; (b) PPh3 (1.2 equiv), 10:1 THF/H2O, rt 12-24 h; (c) 4 M HCl/dioxanes, rt 24 h; (d) Pd(PPh3)4
(cat.), PPh3 (2 equiv), dimedone (7 equiv), rt 12 h; (e) pantolactone (3 equiv), NEt3 (3 equiv), MeOH, reflux 24-72 h.
stepwise conversion by CoaA, a phosphopantetheine adeny-
lytranserase (CoaD; E.C. 2.7.7.3), and a dephospho-CoA kinase
(CoaE; E.C. 2.7.1.24) to form CoA analogue 28, which is
subsequently transferred to a conserved residue of the carrier
protein by a PPTase to produce reporter-modified crypto-carrier
protein 30.
carbonyl and H-bond donating nitrogen of the natural substrate
amide with weak H-bond accepting ether oxygens. Their
synthesis made use of a common orthogonally protected diamine
48 (see Supporting Information), which underwent differential
deprotection to give monoprotected diamines 32b and 33d.
Subsequent EDAC mediated conjugation to dye 31, azido/Boc
deprotection by standard conditions, and nucleophilic ring
opening of pantolactone afforded enzyme probes 15 and 16.
Our interest in incorporating the advantageous properties of PEG
spacers in our library of pantetheine analogues lead us to
synthesize PEG-linked-pantoic acid conjugate 17 in a 41%
overall yield through an analogous dye conjugation/deprotection/
nucleophilic ring-opening sequence starting from previously
described N-Alloc diaminoethylene glycol 34e. Compound 18
was chosen to test the limits of linker length in CoA biosynthesis
and was easily attainable from the commercially available
monoazido/monoamino terminal nonaethylene glycol 32a through
a similar series of reactions.
Analogues 12-18 were synthesized by this route in order to
test the permissibility of CoA biosynthesis toward unnatural
pantetheine analogues, particularly the effect of changes in the
â-alanine/cystamine region (Scheme 1). We chose three parallel
amino-protecting group strategies (azide, Boc, Alloc) based on
the commercial availability (32a), simple synthesis from
literature preparations (33c, 34e), and orthogonal protecting
group traits (32b, 33d) of the specified diamines (Scheme 1).
This strategy allowed compound 12 to be synthesized in two
steps from N-Boc ethylenediamine conjugated 7-dimethylami-
nocoumarin-4-acetic acid (DMACA) 36c.20 Acid-catalyzed
deprotection afforded the free amine, which performed nucleo-
philic ring opening of pantolactone to afford the final product
in 85% yield. Compound 13 was synthesized as previously
described,4 while dansylated pantetheine 14 was synthesized by
an analogous route. Compounds 15 and 16 were chosen to probe
the effect of replacement of the strong H-bond accepting
Compounds 1-8 sought to create a wide spectrum of
pantetheine analogues incorporating some of the most commonly
used bioorthogonal tags such as ketones, azides, and alkynes.
Particular attention was paid to the azido moiety, where
analogues replacing the â-Ala (1), cystamine (2), and thiol (5)
regions of natural pantetheine with terminal azide moieties were
(20) La Clair, J. J.; et al. ChemBioChem 2006, 7, 409-416.
9
12178 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006

In Vivo Bioorthogonal Carrier Protein Modification
A R T I C L E S
Scheme 2. Synthesis of â-Ala-Linked Bioorgthogonal Pantetheine Analoguesa
a (a) 41 or 42, EDAC (2 equiv), HOBt (2.5 equiv), DIPEA (2 equiv), DMF, rt 12 h; (b) 1 M HCl, 1:1 THF/H2O, rt 1 h; (c) 42, EDAC (2 equiv), HOBt
(2.5 equiv), DIPEA (2 equiv), DMF, rt 12 h; (d) PPh3 (2 equiv), 10:1 THF/H2O, rt 12 h; (e) NHS ester 43a or 43b, NEt3 (2 equiv), rt 8 h; (f) 1 M HCl, 1:1
THF/H2O rt 1 h.
Scheme 3. Synthesis of PEG-Linked Bioorthogonal Pantetheine
N-Alloc protected diaminoethylene glycol 44 to give a common
intermediate, followed by Pd(PPh3)4 mediated deprotection and
Analoguesa
coupling of the purified free amine to an keto/azido acid
activated as the succinimidyl ester. Finally, analogue 8 was
synthesized in one step through microwave-assisted nucleophilic
ring-opening of pantolactone by monoazido/monoamino termi-
nated nonaethylene glycol 32a.
In Vitro Pathway Incorporation: Kinetics with CoaA. As
mentioned earlier, phosphorylation of the primary hydroxyl
group of pantothenate by the first protein in the CoA biosyn-
thesis pathway, CoaA, is believed to be the rate-limiting step
in vivo.13 Due to its role as the gatekeeper of CoA biosynthesis,
we assayed each of our new analogues for kinetic activity with
CoaA (Table 2).
The assay was performed as previously described using the
prototypical bacterial pantothenate kinase, CoaA from E. coli.15
The reaction of CoaA with the natural substrate pantothenate
gave values that conformed to those previously reported in the
literature.14,21 Pantothenate mimics 2 and 3 show kcat and Km
values closely approaching those of the natural substrate. As
seen in our previous studies pantetheine is also a substrate for
CoaA, and pantetheine-resembling substrates 4, 5, 13, and 14
show turnovers near equal to (4, 13, 14) or greater than (5) that
of pantetheine. Compounds 5, 13, and 14 are processed
Pd(PPh3)4 (0.05 equiv), PPh3, dimedone; (c) NHS ester 43a or 43b, NEt3
a (a) Pantolactone (3 equiv), NEt3 (3 equiv), EtOH, 160 °C 0.5 h; (b)
particularly efficiently by CoaA. Conversely, compounds with
PEG linkers between the pantoic acid moiety and the bioor-
thogonal terminus were poor substrates for CoaA. The length
of the linker region was a strong factor in determining substrate
suitability, with the longest compounds 8 and 18 proving such
poor substrates that kinetic data could not be generated. Shorter
PEG linked compounds (7 and 17) were viable substrates in
the assay but turned over at a rate 2-5-fold less than that for
derivatives containing â-Ala/diamine linkers. PEG linked pan-
toic acid conjugate 6 differs only from 7 by the exchange of an
azide for an acyl substitutent but shows markedly decreased
kinetic activity.
(2 equiv), rt 4 h.
synthesized, in addition to experimentation with short (7) and
long (8) PEG-linked azido-pantetheine analogues. Compound
1 was synthesized in one step from the nucleophilic ring-opening
of pantolactone by 2-azidoethanamine (42). Compounds 2 and
3 (Scheme 2) were synthesized by standard peptide coupling
of p-methoxybenzylideneacetal-protected (PMB) pantothenate
39 and the corresponding alkynyl and azido-amines, followed
by acidic deprotection of the 1,3-diol. Compounds 4 and 5 again
utilized 39 as a starting material, which was coupled to
2-azidoethanamine, deprotected to the corresponding amine, and
coupled to the corresponding N-hydroxysuccinimidyl keto/azido
(43a,b) ester. PEG linked pantetheine conjugates 6 and 7
(Scheme 3) were constructed in a similar fashion by nucleophilic
ring-opening of pantolactone under microwave-conditions by
To isolate and investigate the effect of H-bond accepting
heteroatoms in the linker region, pantetheine analogues 15 and
(21) Strauss, E.; Begley, T. P. J. Biol. Chem. 2002, 277, 48205-48209.
9
J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006 12179

A R T I C L E S
Meier et al.
Table 2. Kinetic Parameters of E. coli CoaA with Natural
Substrates and Pantetheine Analogues and Summary of in Vivo/in
Vitro Resultsa
k
cat/Km
(s-1 M-
103)
)
in
in
1
1
compound
k
cat (min-
)
Km
(µ
M)
(×
vitro
vivo
pantothenate 31.27 ( 0.58 28.56 ( 1.77 18.25 ( 0.68 na
na
na
-
pantetheine14 19.2 ( 0.1
91 ( 10
3.53 ( 0.44
na
1
2
3
4
5
6
7
8
12
13
14
15
16
17
18
22.33 ( 1.54 692.3 ( 89.13 0.54 ( 0.07
++
31.45 ( 0.75 32.85 ( 2.54 15.96 ( 0.76 ++ ++
28.02 ( 0.42 43.53 ( 1.99 10.73 ( 0.32 ++ ++
20.64 ( 0.45 62.65 ( 3.89 5.49 ( 0.24
40.91 ( 1.12 71.89 ( 6.98 9.48 ( 0.52
1.60 ( 0.06 0.196 ( 0.09 136.4 ( 10.63 ++
++
++ ++
+
-
-
-
+
6.16 ( 0.44 53.76 ( 11.16 1.91 ( 0.27
na na na
11.56 ( 0.57 37.24 ( 5.82 5.17 ( 0.51
++
+
++
19.16 ( 1.41 28.40 ( 6.92 11.25 ( 1.65 ++ ++
17.26 ( 0.62 27.33 ( 3.28 10.53 ( 0.76 ++
-
-
-
-
-
1.36 ( 0.05 0.76 ( 0.32
1.00 ( 0.03 0.14 ( 0.18
0.03 ( 0.04
0.12 ( 0.01
++
++
++
+
10.06 ( 0.43 51.18 ( 6.42 3.28 ( 0.28
na
na
na
a Kinetic data with CoaA and summary of in vivo/in vitro labeling
efficiency of carrier proteins using reporter modified pantetheine analogues.
Kinetic values for the natural substrate pantothenate and pantetheine are
given for comparison. In vivo/in vitro labeling is based on data from gel
shift, fluorescent gel, Western blot, and MALDI-MS data (see Figures 3,4,5,
Supporting Information) and represented semiquantitatively as follows: ++
for excellent labeling of carrier protein, + for low but detectable labeling,
and - for no detectable labeling of carrier protein.
Figure 3. In vivo and in vitro activity of pantetheine analogue panel. (a)
Analogues were assayed for gel shift after reaction with CoA biosynthesis
enzymes (CoaA/D/E), PPTase (Sfp), and carrier protein (E. coli ACP).
Conversion of apo-ACP to reporter modified crypto-ACP causes a change
in the mobility of the protein on native PAGE. (b) In vitro modification of
the carrier protein VibB by reaction with fluorescent pantetheine analogues,
CoA biosynthesis enzymes (CoaA/D/E), and Sfp. (c) In vivo modification
of carrier protein by incubation of fluorescent pantetheine analogues with
E. coli overexpressing VibB and the PPTase Sfp. (d) In vitro modification
of VibB by reaction with bioorthogonal pantetheine analogues, CoA
biosynthesis enzymes (CoaA/D/E), and Sfp. Labeled carrier protein is
visualized by chemoselective ligation to the appropriate biotin reporter
(9-11) followed by SDS-PAGE, blotting onto nitrocellulose, and incubation
with streptavidin-linked alkaline phosphatase. (e) In vivo modification
of carrier protein by incubation of bioorthogonal pantetheine analogues
with E. coli overexpressing VibB and the PPTase Sfp. Visualization as in
(d).
16 were synthesized. These compounds were designed to replace
the amide bond between â-Ala and cystamine of pantetheine
with a single ether oxygen, allowing investigation of subtle
substrate-enzyme interactions in the CoaA active site.12 To our
surprise these compounds were extremely poor substrates. While
compounds with the aforementioned short PEG linkers (7, 17)
showed Km values 2-fold higher than that for natural substrate
pantothenate, and compounds with traditional â-Ala/diamine
linkers showed values that were either the same (13, 14) or
2-fold higher (4, 5), the Km values for 15 and 16 were 100-fold
lower. Turnover for these compounds was proportionately low,
indicative of tight binding. To test if these compounds were
acting as inhibitors of CoaA, a competitive kinetic assay was
set up using pantotenate as the substrate. The results (see
Supporting Information) show that 15 acts as a noncompetitive
inhibitor, suggesting that it may bind the allosteric regulation
site of CoaA.23 Investigation of these compounds as potential
inhibitor scaffolds is ongoing. Azide 1, which omitted entirely
the â-Ala/carbon diamine or PEG linkages of other analogues,
showed turnover within the range of the natural substrate;
however the Km was 20-fold higher than that for pantothenate
suggesting deletion of an interaction involved in active site
binding. Another interesting pantoic acid analogue 12, which
shortened the pantoic acid-reporter linker length to four carbons
and reversed the carbonyl and amide â-Ala/cystamine linkage
of natural pantetheine, showed lower turnover and catalytic
efficiency than analogues containing the natural â-Ala/pantoic
acid linkage (5, 13, 14).
causes a change in the mobility of the protein on a nondenaturing
polyacrylamide gel.12 To assay each compound for activity
throughout the entire co-opted CoA pathway, we ran covalently
modified carrier protein on a native PAGE gel and compared
the mobility to apo-ACP (Figure 3a). Pantetheine analogues
were reacted as previously reported4 with the enzymes CoaA-E
to create CoA analogues, followed by the addition of the PPTase
Sfp and apo-ACP. As seen in Figure 3a, all of the compounds
tested demonstrated some change in mobility with relation to
apo-ACP. For most compounds full conversion to crypto-ACP
is obtained; however analogues with long PEG linkers (8, 18)
give multiple bands on the gel that suggest apo-protein remains.
To confirm the results from this assay, we subjected the
reaction mixtures to matrix-assisted laser desorption/ionization
mass spectrometry (MALDI-MS). The MALDI-MS data (Figure
4 and Supporting Information) confirm that all pantetheine
analogues in our panel are indeed converted into CoA deriva-
tives and transferred onto a carrier protein in vitro. Apo-ACP
(Figure 4a) shows a characteristic peak with a mass of 8505
Da. Compound 13 was reacted with the CoaA biosynthesis
In Vitro Pathway Incorporation: Gel Shift. The conversion
of apo-ACP to holo-ACP or reporter-modified crypto-ACP
(22) Mercer, A. C.; La Clair, J. J.; Burkart, M. D. ChemBioChem 2005, 8, 1335-
1337.
(23) Ivey, R. A.; Zhang, Y.; Virga, K. G.; Hevener, K.; Lee, R. E.; Rock, C.
O.; Jackowski, S.; Park, H. J. Biol. Chem. 2004, 279, 35622-35629.
9
12180 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006

In Vivo Bioorthogonal Carrier Protein Modification
A R T I C L E S
the expected mass change of 604 mass units. PEG analogues
17 and 18 (Figure 4c,d) also show the expected mass shifts of
559 units and 916 units, respectively, corresponding to the
formation of a crypto-carrier protein. Only in the case of the
PEG-linked derivatives 8 and 18 does the conversion appear
significantly incomplete, supporting the results of the gel shift
assay.
In Vitro Pathway Incorporation: Biodetectability. Having
confirmed that the pantetheine analogues were suitable substrates
for the CoA biosynthesis enzymes and PPTase/carrier protein
reaction, we next investigated the biodetectability of each
analogue. The fluorescent analogues (12-18) were detected by
UV visualization on PAGE gels as previously described.4
Bioorthogonally tagged pantetheine analogues 1-8 were de-
tected by chemoselective ligation to the appropriate alkoxyamine/
azide/alkyne functionalized biotin followed by PAGE and
visualization by western blotting and incubation with strepta-
vidin-conjugated alkaline phosphatase. Keto-pantetheine com-
pounds 4 and 6 were reacted overnight with biotin hydroxy-
lamine (9) at room temperature, while pantetheine analogues
with azide and alkyne functionalities were reacted with the
corresponding alkynyl/azido (10/11) biotin following the pro-
cedure of Alexander.9c Inspection of the fluorescent gels (Figure
3b) and western blots (Figure 3d) confirmed the results of the
gel shift assay and mass spectral data, indicating biodetectible
covalent modification of carrier proteins in vitro for compounds
1-8 and 12-18.
In Vivo Pathway Incorporation. Next we tested our library
of pantetheine analogues for integration into the E. coli CoA
pathway using our in vivo assay.4 E. coli overexpressing the
carrier protein VibB and the PPTase Sfp were incubated with
1 mM of each compound in 1 mL of culture. After 4 h of growth
cells were pelleted, washed, and lysed. Lysate from these
cultures was run on SDS-PAGE gels, and detection was carried
out as described for the in vitro studies. Compounds 2, 3, 5,
12, and 13 demonstrated detectible modification of VibB in vivo
(Figure 3c,e). Keto-pantetheine analogue 4 was also detectable
but showed much weaker labeling than the similarly linked
azido-analogue 5 (Figure 3e). The compounds most active in
vivo show a strong correlation with the CoaA kinetic profile.
To verify the results of the gels and blots, samples of crude
lysate were assayed by MALDI-MS (Figure 5). For MALDI-
MS analysis doubly transformed E. coli containing plasmids
for the carrier protein Fren and the PPTase Sfp were used. As
seen in Figure 5a compound 13 was taken up by the cell,
processed into a CoA analogue and attached onto Fren. The
apo peak can be seen at 8570 mass units. Holo-carrier protein
is also visible at 8910 Da. This peak arises from the fact that
natural CoA is available in the cell and readily ligated by
PPTases to the overexpressed apo-Fren. Carrier protein modified
with 13 can be seen at 9179 Da giving the expected 609 mass
unit change. Similarly mass spectral analysis of cell lysate after
incubation of bioorthogonal pantetheine analogue 2 with Fren/
Sfp overexpressing E. coli shows an observable 315 Da shift
of the known apo peak indicating formation of an alkyne-
modified crypto-carrier protein. Subjection of the same crude
cell lysate to click reaction conditions with biotin reporter 11
resulted in another 317 Da mass shift, indicating successful
formation of biotinylated Fren via a Cu(I)-catalyzed [3 + 2]
cycloaddition process.
Figure 4. In vitro labeling of ACP. Apo-ACP (a) is reacted with pantetheine
analogues, CoA biosynthetic enzymes (CoaA/D/E), and the PPTase Sfp.
Fluorescent compounds 13 (b), 17 (c), and 18 (d) are all shown to be
converted to CoA analogues and modify ACP by gel and mass spectral
analysis.
enzymes, ACP, and Sfp as described above, and the reaction
mix was analyzed by MALDI without further purification. As
seen in Figure 4b, reaction with the CoA analogue of 13 causes
9
J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006 12181

A R T I C L E S
Meier et al.
of the chemoenzymatic approach to synthesis of unnatural CoA
analogues. As mentioned above this finding greatly simplifies
the synthetic task of constructing CoA derivatives for in vitro
applications which utilize carrier protein tagging.6 This allows
access to virtually any reporter labeled-CoA analogue from a
monoprotected amine in three steps, one of which can be
expedited using microwave technology. Additionally the subtle
substrate preferences Sfp has been shown to exhibit in systems
incorporating multiple carrier proteins can be used for selective
coding based on differential phosphopantetheinylation by un-
natural CoA analogues,22 adding another layer of complexity
to in vitro and cell-surface carrier protein fusion systems.
Detailed analyses of the kinetic data demonstrate several
important relationships. In general compounds containing a
similar â-Ala linker region to that of natural pantothenate show
the best kinetic profiles. In all compounds of the panel the
pantoic acid region (C1-N5, numbering from the terminal
primary hydroxyl of pantetheine) was conserved. Lee et al.
indicated low binding of PANK inhibitors which lacked a
proton-donating amide NH at the N8 position, pointing to a
model of the PANK-ADP-pantothenate ternary complex in
which the C-7 acid of pantothenate acts as an H-bond donor
toward two key residues.12,23 Our results verify the importance
of this interaction. Compounds 1 and 3 contain the same
2-azidoethanamine-derived terminal azide, differing only in the
â-Ala linkage of 3. This change results in a 10-fold increase in
Km and near 20-fold increase in catalytic efficiency, indicating
much better binding of the substrate with the H-bond donating
â-Ala linker. Pantetheine analogues containing ethylene glycol
based linkers incapable of acting as efficient H-bond donors
(6-8, 17-18) showed similarly poor kinetics compared with
those containing â-Ala linkers. Compounds 15 and 16 provide
perhaps the strongest evidence of the importance of this
interaction, losing almost all substrate activity when reducing
the strength of the electron pair donor and removing the H-bond
donating NH completely from the pantetheine analogue. Inter-
estingly these compounds show markedly different kinetics than
PEG-linked pantetheine 17, which differs by only two atoms,
demonstrating the limitations of this model in predicting effects
caused by alternate variables such as the reduced rotation around
an sp2 hybridized carbon at C8 and addition of an extra H-bond
accepting heteroatom further down the linker. The kinetics of
analogue 12, in which the connectivity of the amide at C8 and
N9 of the natural substrate pantetheine is reversed, indicate that
there is some flexibility in the pocket around this position.
Compound 12 shows good binding but slow turnover, with a
catalytic efficiency poorer than that of any of the compounds
with the H-bond donator in its natural position (2-5, 13-14)
but better than that of every compound in which the H-bond
donating NH is absent (1, 6-8, 15-18). While the general trend
toward an H-bond donor effect is large, other interactions
resulting from the proximity of the aromatic coumarin-reporter
molecule to the active site in this analogue must also be
considered.
Figure 5. In vivo carrier protein labeling. The carrier protein Fren is labeled
in vivo by incubation of E. coli overexpressing Fren and Sfp with (a) a
fluorescent pantothenate analogue 13 and (b) a bioorthogonally tagged
analogue 2. Click reaction of alkyne-modified crypto-carrier protein with
biotin reporter 11 affords triazole-linked biotinylated carrier protein (c),
resulting in the expected shift in mass and allowing protein visualization
by western blot.
Insights into CoaA Substrate Specificity from Kinetic, in
Vivo, and in Vitro Analyses. Comparisons of the kinetic, in
vitro, and in vivo assay results for different members of the
panel yield important insights into the structure-activity
relationships between E. coli PANK and pantetheine analogues.
Despite the poor performance of compounds 1, 6-8, 12, and
15-18 in the CoaA kinetic assay, all were shown to be
converted to CoA analogues and loaded onto the carrier protein
ACP by Sfp in vitro. This is both a testament to the incredible
efficiency of Sfp in transferring unnatural CoA derivatives to
apo-carrier proteins and a further demonstration of the utility
Several other trends which may be important for future design
of carrier protein tags are of interest. In general, analogues
terminating in alkynes show better kinetic parameters than those
of azides, with azides showing better kinetic parameters than
those of ketones. For PEG-based pantetheine derivatives, chain
length proved an important factor, with longer chains showing
9
12182 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006

In Vivo Bioorthogonal Carrier Protein Modification
A R T I C L E S
negligible activity with CoaA and poor in vivo protein tagging.
Appending a different dye to the end of the pantetheine had no
statistically significant impact on CoaA kinetics; however
substitution of DMACA with dansyl lead to a complete loss of
in vivo activity, suggesting a possible lack of a viable
membrane-transport mechanism for dansyl-pantetheine ana-
logues or an intracellular degradation process. It has been shown
previously that DMACA is an excellent dye for in vivo
applications.20 The kinetics of â-Ala containing bioorthogonal
pantetheines (2-5) compared with â-Ala containing panteth-
eines in which a fluorescent reporter that was directly appended
(13-14) showed slightly better turnover and similar catalytic
efficiency. On the whole the binding site of CoaA beyond the
â-Ala moiety appears to be quite promiscuous, with little kinetic
effect observed on substitution of the hexanediamine linker of
13 with a ethylenediamine linker or substitution of the ethyl-
enediamine linker of 14 with a short (8-atom) PEG linker.24
Perhaps the most important conclusion that can be drawn from
the assay results is that the in vivo activity of a pantetheine
analogue has a direct correlation with kinetic activity with CoaA.
Analogues 2, 3, 4, 5, 12, and 13 were shown to be biodetectible
in E. coli by Western blot analysis and fluorescence visualiza-
tion. These show a kcat between 11.5 and 40.9 min-1 and kcat/
Km of 5.2 × 103-16.0 × 103 compared with the values of 19.2
min-1 (kcat) and 3.5 × 103 s-1 M-1 (kcat/Km) for unmodified
pantetheine (Table 2). In order for a pantetheine analogue to
be processed into a CoA analogue in vivo, not only must an
organism contain a promiscuous PANK with pantetheine kinase
activity but also the pantetheine analogue must have comparable
or better kinetics with PANK than pantetheine and (ideally)
natural substrate pantothenate. Given that enterococci produce
far more pantothenate than they require for primary metabo-
lism,25 any modified pantetheine analogue must make extremely
efficient use of CoaA in order for CoA conversion and
subsequent protein labeling to occur at detectible levels.
Although the structural requirements for the import of panteth-
eine analogues have not been studied in detail, E. coli has been
shown to have a pantothenate import system, the pantothenate
permease (panF) symporter,26 which also may exert some
selectivity in the import of analogues and thus influence the
ability of pantetheine analogues to be integrated into the CoA
pathway in vivo. However previous studies showed that while
overexpression of the panF gene resulted in elevated pantoth-
enate uptake, a concurrent increase in CoA production was not
observed, indicating PANK activity as the principal regulator
of CoA biosynthesis. With that in mind these kinetic parameters
should prove useful for the future design and assay of in vivo
carrier protein tags.
alternative methodology negates the purification steps necessary
in the production of maleimide-CoA analogues from com-
mercial sources, allows almost any variance of the chemical
identity of the linker region, and provides an economical
substitute for producing large amounts of CoA analogues in
cases where large quantities of the desired CoA-maleimide-
reporter conjugate may be prohibitively expensive. In addition
the expansion of analogues with bioorthogonal reporters allows
for increased detection and sensitivity. However despite these
subtle advances, it is in the prospect of in vivo labeling that
these tools become particularly important. The covalent modi-
fications described herein have practical value in the study of
in vivo activity of proteins and a place among the ever increasing
myriad of proteomic techniques used to study them.
Of particular importance are the chemoselective ligation
reactions demonstrated by the ketone, azide, and alkyne protein
labels. These tools allow carrier proteins to be visualized and
isolated for the first time without the expense and complication
of antibody techniques. While our survey of bioorthogonal
coupling partners was not exhaustive, the functionalities intro-
duced should be applicable to other published methods. For
instance, one can easily envisage analogues 3 and 5 being
modified by Bertozzi’s covalent Staudinger ligation with a
reporter-conjugated triarylphosphine analogue for in vivo ap-
plications in which more stringent Cu(I)-catalyzed click chem-
istry conditions are not ideal.27 It was to our disappointment
that the PEG-incorporating pantetheine analogues proved nona-
menable to in vivo protein labeling, most likely due to deletion
of a crucial H-bonding interaction. Yet while PEG analogues
are often useful for distancing reporter labels from the protein
of interest for downstream modification, most likely they are
not a necessity in this instance by virtue of the 4′-phosphopan-
tetheine moiety. Indeed the 4′-phosphopantetheine is commonly
believed to be appended to carrier proteins as a means to
distance substrates and products from the protein core, a concept
reinforced by recent structural studies of the fatty acid syn-
thase.28
The system used in this study was limited to the E. coli CoA
pathway and relied on the overexpression of a carrier protein
PPTase pair to facilitate detection. However it has already been
demonstrated that AcpS, a widely expressed PPTase, has the
ability to transfer CoA analogues chemoenzymatically generated
from panthothenamide type substrates to apo-ACP.12 Immediate
work will aim at lowering the detection limit to native protein
levels in E. coli, a reasonable goal given the demonstrated ability
of bioorthogonally labeled carrier protein to undergo ligation
to affinity agents readily amenable to enrichment strategies. Also
of interest to the growth of this methodology is the promiscuity
of CoA biosynthetic enzymes in species other than E. coli. Little
is known about the interspecies substrate specificity of CoaD
and CoaE, although studies of the mechanism of inhibition of
N-alkylpantothenamides have shown these enzymes capable of
catalyzing the formation of unnatural CoA analogues which
modify the fatty acid ACP in both E. coli and S. aureus,21,29
despite a disparity in sequence homology between the PANK
Conclusions
The molecules described here have varied applications and
provide insight valuable in the expanding field of proteomics.
For in vitro applications and in vivo cell-surface labeling of
carrier protein fusions, all pantetheine analogues studied in this
manuscript are efficiently converted into CoA analogues and
tethered to the protein via the four-step enzymatic pathway. This
(27) Kohn M, Breinbauer R. Angew. Chem., Int. Ed. 2004, 43, 3106-3116.
(28) (a) Simon Jenni, S.; Leibundgut, M.; Maier, T.; Ban, N. Science 2006,
311, 1258-1262. (b) Maier, T.; Jenni, S.; Ban, N. Science 2006, 311, 1263-
1267.
(29) Leonardi, R.; Chohnan, S.; Zhang, Y.; Virga, K. G.; Lee, R. E.; Rock, C.
O.; Jackowski, S. J. Biol. Chem. 2005, 280, 3314-3322.
(24) This comparison refers to kinetic data compiled for alternatively linked
dansyl and DMACA-pantetheine analogues in ref 14.
(25) Rock, C. O.; Calder, R. B.; Karim, M. A.; Jackowski, S. J. Biol. Chem.
2000, 275, 1377-1383.
(26) Jackowski, S.; Alix, J. H. J. Bacteriol. 1990, 172, 3842-3848.
9
J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006 12183

A R T I C L E S
Meier et al.
enzymes of the different species.30 Studies in our own lab have
shown the mammalian form of CoaD and CoaE, which exists
as a bifunctional fusion, efficiently processes 4′-phosphopan-
tetheine analogues into CoA analogues in vitro.14 One system
where this approach may be limited is in a subset of pathogenic
bacteria which contain only the recently discovered CoaX type
pantothenate kinase. This PANK isoform has been shown to
be resistant to inhibition by N-alkylpantothenamides, possibly
indicating an inability of pantetheine analogues to act as
competitive substrates.31 To date most studies of pantetheine
analogues have focused on their properties of inhibitors; the
new amenability of carrier protein systems to fusion tagging
strategies and our demonstration of protein modification in living
systems provide strong incentive for the reexamination of the
substrate properties of this class of small molecules.
vivo.32 Our increasing understanding of the kinetic parameters
and structural limitations of pantetheine analogues sets the stage
for the future use of 4′-phosphopantetheine analogue labeling
in chemical biology.
Acknowledgment. This manuscript is dedicated to Bill
Fenical on the occasion of his 65th birthday. Funding was
provided by the University of California, San Diego, Department
of Chemistry and Biochemistry, NSF CAREER Award 0347681
and NIH RO1GM075797. J.M. was supported by NIH Training
Grant T32DK007233. H.R. was supported as an NIH NIGMS
PREP scholar. We thank Joseph Noel and Thomas Baiga of
the Salk Institute for Biological Studies for the use of a
microwave reactor and Jessica Alexander of the Scripps
Research Institute for helpful discussions.
Supporting Information Available: Abbreviations used are
given in ref 33. General procedures for microwave assisted ring-
opening of pantolactone; experimental details for the synthesis
of compounds 1-18; kinetic, in vitro, and in vivo assay details;
Here we have performed a detailed investigation of panteth-
eine analogues to identify suitable partners for covalent protein
labeling inside living cells. A rapid synthesis of pantothenamide
analogues was developed for this purpose and used to produce
a panel which was evaluated for in vitro and in vivo protein
labeling. Kinetic comparisons allowed the construction of
structure-activity relationships to pinpoint the linker, dye, and
bioorthogonal reporter of choice for protein labeling. Finally
bioorthogonal pantetheine analogues were shown to target carrier
proteins with high specificity in vivo and undergo chemose-
lective ligation to reporters in crude cell lysate. The paucity of
site-specific protein labeling tools represents a major obstacle
to the routine application of such small-molecule probes in
1
full author listings for refs 18 and 20; and H and 13C NMR
spectra of all final compounds and synthetic intermediates. This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA063217N
(32) Chen, I.; Ting, A. Y. Curr. Opin. Biotechnol. 2005, 16, 35-40.
(33) Abbreviations: CoA, Coenzyme A; PPTase, phosphopantetheinyltrans-
ferase; O-GlcNAc, O-linked â-N-acetylglucosamine; â-Ala, â-alanine;
PANK, pantothenate kinase; CoaA, ecoli PANK; CoaD, E. coli phospho-
pantetheine adenylytransferase; CoaE, E. coli dephospho-CoA kinase PEG,
poly(ethylene glycol); Pantolactone, (D)-(-)-pantolactone; Boc, tert-butyl
carbamate; Alloc, allyl-carbamate; DMACA, 7-dimethylaminocoumarin-
4-acetic acid; CP, carrier protein; ACP, fatty acid synthase acyl carrier
protein from E. coli; VibB, vibriobactin synthase carrier protein from Vibrio
cholerae; Fren, frenolicin synthase carrier protein from Streptomyces
roseofulVus; panF, pantothenate permease.
(30) Choudhry, A. E.; Mandichak, T. L.; Broskey, J. P.; Egolf, R. W.; Kinsland,
C.; Begley, T. P.; Seefeld, M. A.; Ku, T. W.; Brown, J. R.; Zalacain, M.;
Ratnam, K. Antimicrob. Agents Chemother. 2003, 47, 2051-2055.
(31) Brand, L. A.; Strauss, E. J. Biol. Chem. 2005, 280, 20185-20188.
9
12184 J. AM. CHEM. SOC. VOL. 128, NO. 37, 2006