Chemoenzymatic reversible immobilization and labeling of proteins without prior purification

Article abstract of DOI:10.1021/ja211308s

Site-specific chemical modification of proteins is important for many applications in biology and biotechnology. Recently, our laboratory and others have exploited the high specificity of the enzyme protein farnesyltransferase (PFTase) to site-specifically modify proteins through the use of alternative substrates that incorporate bioorthogonal functionality including azides and alkynes. In this study, we evaluate two aldehyde-containing molecules as substrates for PFTase and as reactants in both oxime and hydrazone formation. Using green fluorescent protein (GFP) as a model system, we demonstrate that the purified protein can be enzymatically modified with either analogue to yield aldehyde-functionalized proteins. Oxime or hydrazone formation was then employed to immobilize, fluorescently label, or PEGylate the resulting aldehyde-containing proteins. Immobilization via hydrazone formation was also shown to be reversible via transoximization with a fluorescent alkoxyamine. After characterizing this labeling strategy using pure protein, the specificity of the enzymatic process was used to selectively label GFP present in crude E. coli extract followed by capture of the aldehyde-modified protein using hydrazide-agarose. Subsequent incubation of the immobilized protein using a fluorescently labeled or PEGylated alkoxyamine resulted in the release of pure GFP containing the desired site-specific covalent modifications. This procedure was also employed to produce PEGylated glucose-dependent insulinotropic polypeptide (GIP), a protein with potential therapeutic activity for diabetes. Given the specificity of the PFTase-catalyzed reaction coupled with the ability to introduce a CAAX-box recognition sequence onto almost any protein, this method shows great potential as a general approach for selective immobilization and labeling of recombinant proteins present in crude cellular extract without prior purification. Beyond generating site-specifically modified proteins, this approach for polypeptide modification could be particularly useful for large-scale production of protein conjugates for therapeutic or industrial applications.

Full text of DOI:10.1021/ja211308s

Article
pubs.acs.org/JACS
Chemoenzymatic Reversible Immobilization and Labeling of Proteins
without Prior Purification
Mohammad Rashidian, James M. Song, Rachel E. Pricer, and Mark D. Distefano*
Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55454, United States
S
* Supporting Information
ABSTRACT: Site-specific chemical modification of proteins
is important for many applications in biology and biotechnol-
ogy. Recently, our laboratory and others have exploited the
high specificity of the enzyme protein farnesyltransferase
(PFTase) to site-specifically modify proteins through the use
of alternative substrates that incorporate bioorthogonal
functionality including azides and alkynes. In this study, we
evaluate two aldehyde-containing molecules as substrates for
PFTase and as reactants in both oxime and hydrazone
formation. Using green fluorescent protein (GFP) as a model system, we demonstrate that the purified protein can be
enzymatically modified with either analogue to yield aldehyde-functionalized proteins. Oxime or hydrazone formation was then
employed to immobilize, fluorescently label, or PEGylate the resulting aldehyde-containing proteins. Immobilization via
hydrazone formation was also shown to be reversible via transoximization with a fluorescent alkoxyamine. After characterizing
this labeling strategy using pure protein, the specificity of the enzymatic process was used to selectively label GFP present in
crude E. coli extract followed by capture of the aldehyde-modified protein using hydrazide−agarose. Subsequent incubation of the
immobilized protein using a fluorescently labeled or PEGylated alkoxyamine resulted in the release of pure GFP containing the
desired site-specific covalent modifications. This procedure was also employed to produce PEGylated glucose-dependent
insulinotropic polypeptide (GIP), a protein with potential therapeutic activity for diabetes. Given the specificity of the PFTase-
catalyzed reaction coupled with the ability to introduce a CAAX-box recognition sequence onto almost any protein, this method
shows great potential as a general approach for selective immobilization and labeling of recombinant proteins present in crude
cellular extract without prior purification. Beyond generating site-specifically modified proteins, this approach for polypeptide
modification could be particularly useful for large-scale production of protein conjugates for therapeutic or industrial applications.
applicable in principle, development of new methods for site-
specific modification of proteins that function under mild
conditions is an area of intense research.18 While a number of
reactions suitable for protein modification have been
developed,19−22 to date, the Cu(I)-catalyzed click reaction has
been the most widely employed bioorthogonal process.20
Although highly useful, that reaction employs Cu(I) which is
toxic to cells and can in some cases erode biological activity. To
address those issues, copper-free variations of the click reaction
have been developed that function based on inclusion of
electron-withdrawing substituents and/or ring strain into
alkyne-containing reagents. While highly promising, these
new reagents are not generally commercially available, are
difficult to synthesize, and manifest low aqueous solubility.23−25
As an alternative, oxime- and hydrazone-based reactions have
found wide application in the conjugation of biomolecules on
account of the absence of aldehyde or ketone groups in
proteins and their orthogonal reactivity with aminooxy or
hydrazine derivatives to give stable hydrazones or oximes.26−33
While reactions between aldehydes and ketones with alkoxy-
INTRODUCTION
■
Site-specific chemical modification of proteins is important for
many applications in biology and biotechnology. It can facilitate
studies of proteins with respect to their structure, folding, and
interaction with other proteins in both biochemical and cellular
investigations. In particular, in many biotechnology applica-
tions, the oriented (i.e., site-specific) covalent attachment of
proteins to surfaces is important because it ensures
homogeneous surface coverage and accessibility to the active
site of the protein.1−8 Protein immobilization is an important
first step for many applications including construction of
biosensors and protein microarrays, development of immuno-
assay methods, and employment of enzymes in biotechnology
procedures.9−11 Similarly, site-specific protein labeling is
essential for a variety of applications ranging from introduction
of fluorophores for biophysical studies to preparation of
protein−polymer conjugates for medical applications.12−17
Importantly, the structural sensitivity of polypeptides calls for
chemical transformations that proceed under mild conditions
and are compatible with all functional groups present therein.
However, such modification is challenging because of the large
number of reactive functional groups typically present in
polypeptides. Although many existing chemical reactions are
Received: December 2, 2011
Published: March 21, 2012
© 2012 American Chemical Society
8455
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 1. (A) Structures of farnesyl diphosphate, farnesyl aldehyde diphosphate (1), and formylbenzoyl-oxy geranyl diphosphate (2). (B) Schematic
representation of prenylation of a protein containing a CAAX-box positioned at its C-terminus (GFP−CVIA, 7) with aldehyde-containing analogue
2 to yield the prenylated product 9a. (C) ESI MS analysis of 7 with the deconvoluted mass spectrum shown in the inset. (D) ESI MS analysis of 9a
with the deconvoluted mass spectrum shown in the inset.
amines or hydrazides are generally slow, they can be
significantly accelerated by addition of aniline.27,34 This has
resulted in a number of exciting applications ranging from the
site-specific glycosylation of proteins29 to the fluorescent
labeling of bacteria and mammalian cells.35 Given the utility
of oxime and hydrazone formation, a number of methods have
been developed to introduce aldehydes and ketones into
proteins. Chemical approaches include transamination in the
presence of sodium glyoxylate and copper sulfate36 or using
pyridoxal-5-phosphate,37 while these have proved to be
powerful methods, they are not applicable to all N-termini
and cannot always be driven to completion. Enzymatic methods
include the action of formylglycine-generating enzyme38 or
nonsense suppression approaches that permit incorporation of
aldehydes into internal positions within proteins.39
Recently, our laboratory and others have exploited the high
specificity of the enzyme protein farnesyltransferase40 (PFTase)
to site-specifically modify peptides and proteins.32,41−43 In
nature, PFTase catalyzes the transfer of a farnesyl isoprenoid
group from farnesyl diphosphate (FPP, Figure 1) to a sulfur
atom present in a cysteine residue. That residue must be
located in a tetrapeptide sequence (denoted as a CAAX-box)
positioned at the C-terminus of a protein or peptide to be a
PFTase substrate. Interestingly, CAAX-box sequences such as
CVIA can be appended to the C-termini of many proteins,
rendering them efficient substrates for PFTase. Since PFTase
can tolerate many simple modifications to the isoprenoid
substrate,3,44−48 it can be used to introduce a diverse range of
functionality into proteins at their C-termini. Chemoselective
reaction with the resulting functionalized protein can then be
used for a wide range of applications. Since the “AAX” residues
from a CAAX-box sequence can be removed by treatment with
carboxypetidase after prenylation, net addition to the protein in
this labeling method can be limited to a single prenylcysteine
residue.49
In an initial communication,32 we reported that compound 1,
an aldehyde-containing analogue of FPP, can be incorporated
into a purified protein substrate using PFTase and that the
resulting aldehyde-functionalized protein can be immobilized or
fluorescently labeled via oxime formation. In this study, we
followed up on those initial observations by comparing the
properties of α,β-unsaturated aldehyde 1 with aryl aldehyde 2
in terms of their efficiency as PFTase substrates and reactants in
both oxime and hydrazone formation. Using green fluorescent
protein (GFP50) as a model system, we demonstrate that the
purified protein can be enzymatically modified with 1 or 2.
Oxime or hydrazone formation was then employed to
immobilize, fluorescently label, or PEGylate the resulting
aldehyde-functionalized proteins. Immobilization via hydrazone
formation was also shown to be reversible via transoximization
with a fluorescent alkoxyamine. After characterizing this
labeling strategy using pure protein, the specificity of the
8456
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 2. (A) Schematic representation of oxime and hydrazone ligations of 8a to yield 8b and 8c. (B) Fluorescence (right) and Coomassie blue
staining (left) images of a gel loaded with 8a labeled with alexafluor 6c and Texas red 6b via oxime and hydrazone ligations, respectively, showing
covalent attachment of fluorophores to the protein: lane 1, GFP−CVIA 7; lane 2, 8c; lane 3, 8b. ESI MS spectra of 8a (C) and hydrazone/oxime
ligation products 8b and 8c, showing full conversion for oxime (D) and ∼20% for hydrazone (E) ligations with the deconvoluted mass spectra
shown in the insets.
analysis. Corrections were applied to all of the rate calculations based
on the difference between the fluorescence intensity of the prenylated
product and the starting peptide. Assuming 100% conversion, the
difference corresponds only to the fluorescence of the total amount of
product. The slope was then divided by the fluorescence difference
followed by multiplying by the total concentration of peptide (2.4
μM), which then gives the rate of formation of product in μM/s. It
should be noted that the KM values reported here are actually apparent
KM values, since the measurements were performed at only a single
peptide concentration. The data were fit to a Michaelis−Menten
model using a nonlinear regression program to determine kcat and KM.
Enzymatic Synthesis of 4a and 5a. Enzymatic reactions (26
mL) contained Tris·HCl (50 mM, pH 7.5), MgCl2 (10 mM), KCl (20
mM), ZnCl2 (10 μM), DTT (5.0 mM), 3 (2.4 μM), PFTase (80 nM),
and either 1 or 2 (30−50 μM). To ensure complete disulfide
reduction of the peptide, all reagents except substrates and enzyme
were premixed and incubated for 2 h at 4 °C. With all reagents mixed,
the reaction was initiated by addition of enzyme and the resulting
mixture was incubated at 30 °C for 1 h. Reaction progress was
monitored by UV absorbance (λ = 340 nm, absorbance of the dansyl
chromophore) using analytical RP-HPLC. The following conditions
were employed: flow rate, 1 mL·min−1; 500 μL injection loop; gradient
0−100% B in 30 min; solvent A, NH4HCO3 (25 mM in H2O); solvent
B, CH3CN. After 1 h, the reaction was purified using a Waters Sep-Pak
enzymatic process was used to selectively label GFP present in
crude E. coli extract followed by capture of the aldehyde-
modified protein using hydrazide−agarose. Subsequent in-
cubation of the immobilized protein using a fluorescently
labeled or PEGylated alkoxyamine resulted in the release of
pure GFP containing the desired site-specific covalent
modifications. This procedure was also employed to produce
PEGylated glucose-dependent insulinotropic polypeptide51
(GIP), a protein with potential therapeutic activity for
diabetes.52
EXPERIMENTAL SECTION
■
Enzymatic Studies of FPP Analogues 1 and 2 Using a
Continuous Fluorescence Assay. Enzymatic reaction mixtures
contained Tris·HCl (50 mM, pH 7.5), MgCl2 (10 mM), KCl (20
mM), ZnCl2 (10 μM), 2.4 μM N-dansyl-GCVIA (3), 0.04% (w/v) n-
dodecyl-ß-D-maltoside, 80 nM PFTase, and varying concentrations of
either 1 or 2 (0−50 μM) in a final volume of 250 μL. Reaction
mixtures were equilibrated at 30 °C for 5 min, initiated by addition of
PFTase, and monitored for an increase in fluorescence (λex = 340 nm,
λem = 505 nm) for approximately 10 min. The initial rates of formation
of products were obtained as slopes in IU/min using least-squares
8457
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Plus reversed-phase C18 Environmental Cartridge. The cartridge was
first washed with solvent B (10 mL) and then equilibrated with solvent
A (20 mL). The crude enzymatic reaction mixture was applied to the
cartridge, and a gradient elution was performed in the following
sequence: 10 mL of solvent A, 10 mL of solvent C (20% solvent B,
80% solvent A), 10 mL of solvent D (40% solvent B, 60% solvent A),
and 10 mL of solvent E (60% solvent B, 40% solvent A). Fractions (1
mL per tube) were collected, and product elution was monitored using
a hand-held UV lamp. The green-fluorescent product was clearly
visible, the brightest fraction was selected, and its purity was confirmed
by RP-HPLC. LC-MS analysis of the purified products gave ions of
913.5 and 979.4 as the predominant species, which are consistent with
[M + H]+ for 4a and 5a, respectively.
Oxime Ligation between Peptide−Aldehyde 4a and 5a and
Aminooxy Alexafluor-488 (6c). Coupling reactions contained 3−5
μM 4a or 5a, 200 μM alexafluor-488 (6c), PB (0.1 M, pH 7.0), and
aniline (100 mM) in a final volume of 500 μL. Reactions were
performed at room temperature and initiated by addition of aniline
(100 mM). LC-MS analysis of the reaction mixture after 3−4 h gave
ions of 1384.6 and 1450.6 as the predominant species, which are
consistent with [M + H]+ for 4c and 5c, respectively.
Hydrazone Ligation between Peptide−Aldehydes 4a and 5a
and Texas Red Hydrazide (6b). Coupling reactions contained 3−5
μM 4a or 5a, 200 μM Texas red (6b), PB (0.1 M, pH 7.0), and aniline
(100 mM) in a final volume of 500 μL. Reactions were performed at
room temperature and initiated by addition of aniline (100 mM). LC-
MS analysis of the reaction mixture after 1 h gave ions of 758.46 and
791.45 as the predominant species, consistent with [M + 2H]2+ for
hydrazones 4b and 5b, respectively.
Enzymatic Incorporation of Compounds 1 and 2 into GFP−
CVIA (7).53 Enzymatic reaction mixtures (10 mL) contained Tris·HCl
(50 mM, pH 7.5), MgCl2 (10 mM), KCl (30 mM), ZnCl2 (10 μM),
DTT (5.0 mM), 7 (2.4 μM), either 1 or 2 (30−50 μM), and PFTase
(80−200 nM). After incubation at 30 °C for 2 h for 1 and overnight
for 2, the respective reaction mixtures were concentrated using an
Amicon Centriprep centrifugation device (10 000 MW cutoff). Next,
the excess of 1 or 2 was removed through a NAP-5 (Amersham)
column using Tris·HCl (50 mM, pH 7.5) as the eluant. The
subsequent protein concentration was calculated by UV absorbance at
488 nm (ε = 55 000 M−1·cm−1).
Coupling Reaction between Aldehyde-Labeled GFP−CVIA
(8a and 9a) with Alexafluor-488 (6c). Alexafluor-488 (6c) (3.2 μL
of 3.2 mM solution in DMSO) was added to 42 μL of 8a or 9a (stock
solution of 60 μM in PB). PB (2 M, pH 7.0, 2.5 μL) was added, and
the reaction was initiated by adding aniline (100 mM) and allowed to
proceed for 3−5 h at room temperature. The mixture was then
purified by a NAP-5 column to remove excess dye. LC-MS analysis of
the sample showed only oxime-ligated protein, and no free aldehyde
was detected, indicating a complete reaction in both cases.
Immobilization of 9a onto Hydrazide Agarose Beads.
Hydrazide agarose beads (Thermo Scientific, hydrazide loading 16
μmol/mL) (300 μL) were washed with PB (0.1 M, pH 7.0, 3 × 500
μL). PB (30 μL, 1 M, pH 7.0) was added to the beads followed by
addition of 9a (200 μL, 87 μM). Immobilization was initiated by
adding aniline (2 μL, 100 mM). For controls, GFP−CVIA (7) was
added instead of 9a. The solution was centrifuged; then the GFP UV-
absorbance of the supernatant was measured as a function of time
(488 nm, ε = 55 000 M−1·cm−1). After 2 h, the solution was
centrifuged and the beads were washed thoroughly with PB (0.3 M,
pH 7.3, 3 × 300 μL) and KCl (1 M, 3 × 300 μL) to remove
nonspecifically bound proteins and stored in pH 7.5 Tris buffer at 4
°C.
of 3.2 mM solution in DMSO) was added to 42 μL of 8a or 9a (stock
solution of 60 μM in PB). PB (2 M, pH 6.7, 2.5 μL) was added, and
the reaction was initiated by adding 100 mM aniline and allowed to
proceed for 5−6 h at room temperature. The mixture was then
purified using a NAP-5 column to remove excess dye. LS-MS analysis
of the sample showed only oxime-ligated protein, and no free aldehyde
was detected, indicating a complete reaction in both cases.
Coupling Reaction between Aldehyde-Labeled GFP−CVIA
(8a and 9a) with Texas Red Hydrazide (6b). Texas red hydrazide
(6b) (7 μL of 1.9 mM solution in DMSO) was added to 100 μL of 8a
and 9a (stock solution of 40 μM in PB). Reaction was initiated by
adding 100 mM aniline and allowed to proceed for 1 h at room
temperature. The mixture was then purified using a NAP-5 column to
remove excess 6b. LC-MS analysis of the sample showed the presence
of both hydrazone-ligated proteins 8b and 9b and free aldehydes 8a
and 9a. The ratios of free aldehydes to their respective hydrazone
products were ∼4, indicating only ∼20% completion within this range
of reactant concentrations. The mass spectrum in Figure 2E, which
corresponds to conversion of compound 8a to 8b, was made by
superimposing the two separate mass spectra of both the free aldehyde
protein 8a and the hydrazone-ligated protein 8b present in the product
mixture, based on their relative intensities. The aldehyde- and
hydrazone-functionalized proteins have different retention times and
thus show two different peaks in the corresponding LC chromato-
grams.
FRET Studies Between GFP−Aldehyde 9a and Texas Red
Hydrazide (6b). Texas Red hydrazide (6b) (7 μL of 1.9 mM solution
in DMSO) was added to 90 μL of 9a (stock solution of 60 μM in PB,
0.1 M, pH 6.7). Reaction was initiated by adding 0.9 μL of aniline (100
mM) followed by vortexing the solution and allowing it to proceed for
1 h at room temperature. The mixture was then purified using a NAP-
5 column to remove excess 6b. The protein solution was diluted to 1
nM, and its fluorescence was measured. For the first control, the same
amount of GFP−CVIA (7) fluorescence was measured and compared
with that of 9b. As a second control, the ligated protein (9b) was
heated for a few minutes to denature the protein and the resulting
fluorescence was measured to verify that FRET required both the
protein and the Texas red fluorophores.
FRET Studies between GFP−Aldehyde 9a and Aminooxy−
TAMRA (6d). Aminooxy−TAMRA (6d) (100 μM) was added to 90
μL of 9a (stock solution of 50 μM in Tris·HCl (100 mM, pH 7.0).
Reaction was initiated by adding 0.9 μL of aniline (100 mM) followed
by vortexing the solution and allowing it to proceed for 3 h at room
temperature. The mixture was then purified using a NAP-10 column to
remove excess 6d. The collected solution appeared red and not green,
suggesting that efficient FRET was occurring between the GFP and
the TAMRA fluorophores. The fluorescence of the solution was
measured and compared with two other controls. For the first control,
the same amount of GFP−CVIA (7) fluorescence was measured and
compared with that of 9e. As a second control, the same amount of
TAMRA fluorescence was measured. The two controls confirmed that
FRET was occurring between the protein and the TAMRA
fluorophores.
Crude Prenylation, Immobilization, and Subsequent Label-
ing and Release of GFP−CVIA. A pellet of cells expressing GFP−
CVIA was suspended in buffer (20 mM Tris·HCl pH 7.5, 1 mM
EDTA), sonicated, and clarified by centrifugation. The GFP
concentration present in the crude soluble protein mixture was
calculated by UV absorbance at 488 nm. Next, prenylation was
performed by adding PFTase (200 nM), 2 (50 μM), Tris·HCl (50
mM, pH 7.5), MgCl2 (10 mM), KCl (30 mM), ZnCl2 (10 μM), and
DTT (5.0 mM) to a solution of 7 to achieve a final concentration of
2.0 μM in the crude mixture. After overnight incubation at 30 °C, the
reaction mixture was filtered and concentrated using an Amicon
Centriprep centrifugation device (10 000 MW cutoff). Next, excess 2
was removed through a NAP-5 (Amersham) column using Tris·HCl
(50 mM, pH 7.5) as the eluting solvent. The subsequent GFP
concentration in the crude mixture was calculated by UV absorbance
at 488 nm and determined to be 30 μM. Immobilization was
performed as described above. The beads were washed thoroughly
Release of Immobilized GFP from Beads Using Hydroxyl-
amine. The GFP beads were incubated in PB (0.3 M, pH 7.0) with
hydroxylamine (200 mM) and aniline (100 mM), and the resulting
mixture was vortexed. The solution was centrifuged, and then the UV
absorbance of the GFP in the supernatant was measured as a function
of time (488 nm, ε = 55 000 M−1·cm−1).
Coupling Reaction between Aldehyde-Labeled GFP−CVIA
(8a and 9a) with Alexafluor-488 (6c). Alexafluor-488 (6c) (3.2 μL
8458
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
with PB (0.3 M, pH 7.3) and KCl (3 × 300 μL, 1 M) to remove
nonspecifically bound proteins followed by incubation with aminooxy
fluorophore 6c (1 mM) and aniline (100 mM) overnight with
constant agitation. The supernatant was then analyzed via SDS-PAGE
and in-gel fluorescence analysis to confirm the labeling and release of
the protein from the beads.
Coupling Reaction between Aldehyde-Labeled GFP−CVIA
(9a) with Aminooxy PEG (10). Aminooxy PEG (10) (1.5 mg, MW
10 kDa) was added to 100 μL of 9a (stock solution of 10 μM in 50
mM Tris·HCl). PB (pH 7) was added to a final concentration of 0.1
M. Reaction was initiated by adding 100 mM aniline and allowed to
proceed for 1 h at room temperature. SDS-PAGE analysis of the
sample was used to confirm covalent attachment of the PEG 10 to
aldehyde 9a. Excess 10 and PB were removed using a zip-tip protocol
followed by MALDI MS analysis of the sample to characterize the
product and demonstrate that no free aldehyde was present, indicating
complete reaction.
PEGylation from Immobilized GFP Beads. Immobilization was
performed as described above. Beads were washed thoroughly with PB
(0.3 M, pH 7.3, 3 × 300 μL) and KCl (1 M, 3 × 300 μL) to remove
nonspecifically bound proteins. Next, the beads were incubated with
aminooxy PEG 10 (2 mM) and aniline (100 mM) overnight while
vortexing the solution. SDS-PAGE analysis of the supernatant
indicated the successful PEGylation and release of the aldehyde−
GFP from the hydrazide beads.
Enzymatic Prenylation of GIP-CVIM (12a) with Aldehyde
Substrate 2. Enzymatic reaction mixtures (10 mL) contained
Tris·HCl (50 mM, pH 7.5), MgCl2 (10 mM), KCl (30 mM), ZnCl2
(10 μM), DTT (5.0 mM), 12a (2 μM), 2 (50 μM), and PFTase (200
nM). After incubation at 30 °C overnight, significant precipitate was
present in solution, suggesting possible precipitation of GIP was
occurring upon prenylation. The precipitate was separated from the
solution by centrifugation at 12 000g for 15 min, washed with 25 mM
(NH4)2CO3 buffer to remove excess 2, and centrifuged again. MALDI-
MS analysis of the precipitate (dissolved in H2O, 0.5% TFA, v/v)
confirmed the prenylation of GIP with 2, while the solution showed
neither starting GIP 12a nor the prenylated material 12b.
Coupling Reaction between Aldehyde-Labeled GIP (12b)
with Aminooxy PEG (13). A small amount of the precipitate 12b
was dissolved in H2O containing 0.5% TFA (v/v), and aminooxy PEG
(13) was added to a final concentration of 200 μM. Reaction was
allowed to proceed for 2 h at room temperature. Excess 13 was
removed using a zip-tip protocol. MALDI-MS analysis of the sample
was employed to characterize the product and demonstrate that no
free aldehyde was present, indicating complete reaction.
Prenylation of GIP (12a) in Crude E. coli Extract. E. coli extract
containing GIP-CVIM (12a) was subjected to enzymatic prenylation
by incubating it in the presence of PFTase (200 nM), 2 (50 μM),
Tris·HCl (50 mM, pH 7.5), MgCl2 (10 mM), KCl (30 mM), ZnCl2
(10 μM), and DTT (5.0 mM). After overnight incubation, the
precipitate was separated from the solution by centrifugation at 12
000g for 15 min, washed with 25 mM (NH4)2CO3 buffer to remove
excess 2, and then centrifuged again. MALDI-MS analysis of the
precipitate, performed as described above, was used to confirm the
prenylation of GIP with 2.
Immobilization and Subsequent PEGylation and Release of
Resin-Bound GIP. Immobilization was performed as described above
for GFP except that GIP was dissolved in H2O containing 0.5% TFA
(v/v). No aniline catalyst was added to the solution in this case. After
incubation for 1 h, the beads were washed thoroughly with H2O (3 ×
300 μL) to remove nonspecifically bound proteins from the beads.
Next, the beads were incubated with aminooxy PEG 13 (∼400 μM) in
H2O containing 0.5% TFA overnight with constant agitation of the
solution. Excess 13 was removed using a zip-tip protocol. MALDI MS
analysis of the sample was employed to confirm the presence of the
desired product and to assess the purity of the PEGylated GIP (14).
General Procedure for MALDI Analysis of Protein Samples.
A zip-tip (C4 column) was first washed with 10 μL of solvent A
(CH3CN containing 0.1% TFA; v/v) and then equilibrated with
solvent B (H2O containing 0.1% TFA; v/v). Sample (10 μL) was then
adsorbed onto the C4 matrix via repeated cycles of aspiration and
ejection (5−10 cycles) using a pipettor. Next, the zip-tip was washed 5
× 10 μL with solvent B and the proteins eluted with 2 μL of a mixture
of solvent A and B (75:25). Next, 0.7 μL of the eluted material was
added to a MALDI plate, and 0.7 μL of matrix was added on top of the
sample plate to form crystals. A saturated solution of sinapinic acid
(3,5-dimethoxy-4-hydroxy-cinnamic acid) was used as the matrix.
RESULTS AND DISCUSSION
■
Comparison of Alkyl and Aryl Aldehydes As Sub-
strates for PFTase. To examine the ability of PFTase to be
used in a protein modification strategy employing oxime and
hydrazone formation, we first wanted to explore the range of
aldehydes that could be accepted as alternative substrates for
PFTase. Thus, compound 2, containing an aryl aldehyde, was
designed and synthesized in five steps from geraniol (Scheme
S2, Supporting Information). In brief, THP-protected geraniol
was initially oxidized at C-8 to a terminal alcohol,54 followed by
acylation with formylbenzoic acid using EDC as the coupling
reagent. The THP group was removed, and the alcohol was
converted to the corresponding allylic bromide using CBr4 and
PPh3. Subsequent displacement with [(n-Bu)4N]3HP2O7
followed by purification via ion-exchange chromatography
and RP-HPLC yielded product in which the desired aldehyde
was almost completely transformed to the corresponding
carboxylic acid; therefore, a direct phosphorylation strategy
using (HNEt3)2HPO4 and CCl3CN as the activating reagent
was employed. Subsequent purification by RP-HPLC produced
the desired aldehyde analogue 2 in 5.4% overall yield, whose
structure was confirmed by 1H NMR, 31P NMR,55 and HR-ESI-
MS. Aldehyde 1 was prepared in six steps starting from farnesol
as previously described32 with several modifications that
significantly improved the overall yield to 1.3%. Despite these
improvements, synthesis of 2 was significantly more efficient
primarily due to the selectivity in the SeO2 oxidation step.
Preparation of 1 proceeds via THP-protected farnesol, whereas
synthesis of 2 uses THP−geraniol. Selective oxidation of the
alkene at C-6 (over the electron-poor C-2 alkene) in geraniol is
facile compared to preferential oxidation of the alkene at C-10
in farnesol due to competing reaction with the C-6 olefin,
which exhibits comparable reactivity. Hence, the reaction
cannot be driven to completion, resulting a significantly
reduced yield (compare 56% for geraniol oxidation to 23%
for farnesol oxidation).
Initially, prenylation reactions containing N-dansyl-GCVIA
(3), 2, and PFTase were monitored by HPLC and LC-MS/MS.
As observed previously with 1, a new species with longer
retention time appeared in the reaction mixture containing 2.
LC-MS analysis of that compound gave an [M + H]+ peak at
979.4 Da, consistent with the proposed structure of peptide 5a
(Supporting Information). Next, kinetic analysis of the
incorporation of analogue 2 by PFTase was performed using
a continuous fluorescence-based enzyme assay as previously
carried out with 1. Varying concentrations of 2 were incubated
with the fluorescent peptide substrate, N-dansyl-GCVIA, and
PFTase; the rates of those enzymatic reactions were
determined and shown to obey saturation kinetics. Steady-
state kinetic parameters for prenylation reactions with the two
aldehyde analogues are summarized in Table 1 with additional
details provided in the Supporting Information section (Figure
S2). Comparison of the catalytic efficiencies for these
alternative substrates indicates that both compounds have
reduced efficiency relative to FPP, manifesting kcat/KM values of
8459
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
than 1 and 4 h, respectively. Concentration by ultra-
centrifugation followed by size-exclusion chromatography to
remove unreacted substrates yielded aldehyde-functionalized
GFP−CVIA 8a and 9a. Reaction completion was confirmed by
LC-MS analysis (Figure 1C and 1D) in which none (in the case
of 8a) or very small amounts (in the case of 9a) of free GFP−
CVIA (7) could be detected in comparison to the large peaks
for prenylated GFPs. Deconvolution of the LC-MS data from
the purified protein products showed species at 27 559.0 and
27 625.5 Da, consistent with the structures of aldehyde−GFPs
8a and 9a. In general, LC-MS analysis of GFP and its congeners
has proved to be quite powerful for studying these reactions. As
noted above, in a preliminary communication32 we had shown
that aldehyde−GFPs 8a could be derivatized to produce oxime-
linked products. Here, it was desired to expand those
experiments to include hydrazone formation and compare the
relative reactivity between the two different aldehyde donors,
8a and 9a, containing α,β-unsaturated and aryl−aldehydes,
respectively. To fluorescently label those aldehyde-functional-
ized proteins, we chose Texas red hydrazide (6b) and
Alexaflour-488 aminooxy (6c) for their excellent quantum
yields and high visible light absorption. Thus, aldehyde−GFPs
8a and 9a were incubated separately with alkoxyamine 6c at pH
7 and room temperature. Kinetic analysis, performed via LC-
MS measurements, showed that the reaction required 3−4 h to
proceed to completion. At that point, no detectable unmodified
protein−aldehydes (8a and 9a) were observed. Gratifyingly, the
deconvoluted MS data indicated the presence of species at 28
032.0 and 28 120.5 Da, consistent with the proposed oximes 8c
and 9c. In-gel fluorescence analysis performed under
denaturing conditions confirmed covalent attachment of
aminooxy 6c to the aldehyde-containing proteins (Figure
2B). Unprenylated GFP−CVIA (7) failed to show any labeling
with alkoxyamine 6c, further confirming that the ligations
require the presence of the enzymatically introduced aldehyde
functionality and that the ligation reaction is truly bioorthog-
onal. Overall, the oxime ligation reactions appear to be highly
efficient since no unligated aldehyde−GFPs (8a or 9a) were
observed upon LC-MS analysis (Figure 2C and 2D) of the
ligation reaction mixtures.
Table 1. Steady-State Kinetic Parameters of Substrates, and
HPLC Retention Times for Prenylated Peptide Products
(kcat
/
Rt
(min)
a
a
compound
kcat (s−1
0.52
)
KM (μM)
1.71
KM)rel
FPP
1
1
0.133 0.003
0.015 0.001
1.87 0.17
1.02 0.16
0.23
0.05
2
N-dansyl-GC(Far)
VIA
21.5
4a
5a
18.9
18.6
a
Vrel refers to kcat/KM with respect to FPP
0.23 and 0.05, respectively (relative to FPP). We found that
decreases in kcat constituted the major reason for the
diminished catalytic efficiency of the analogues; kcat for
aldehyde 1 was 4-fold lower, while kcat for aldehyde 2 was
35-fold lower (compared to kcat for FPP). No significant
differences were observed in the KM values for the different
analogues. Thus, in summary, while 1 is the superior alternative
substrate, 2 is easier to prepare making these two compounds
functionally interchangeable.
Preparation and Reactivity of PFTase-Mediated
Aldehyde-Functionalized Peptides. In our earlier work
with 1, experiments with the aldehyde-functionalized peptide
4a focused on oxime-forming reactions. Here, we sought to
expand the scope of possible chemistry to include hydrazone
formation as well. Accordingly, large-scale (26 mL) reactions
containing N-dansyl-GCVIA (3), 1 or 2, and PFTase were
performed and the products isolated after purification via solid-
phase extraction. The resulting material was subsequently used
to evaluate ligation reactions between Texas red hydrazide 6b/
alexafluor-488 aminooxy 6c and aldehyde-containing peptides
4a and 5a in the presence of aniline.27 Kinetic analysis of oxime
formation showed that in the range of 2−4 μM 4a and 5a
ligations at pH 7 were essentially complete within 3−4 h. LC-
MS analysis of the reaction mixture resulted in [M + H]+ peaks
being observed at 1384.6 and 1450.6 Da, consistent with
production of oximes 4c and 5c, respectively (Supporting
Information). In contrast, hydrazone formation with 4a and 5a
in the same concentration range of reagents showed only 30−
50% completion but within 30−60 min (a significantly shorter
time frame). LC-MS analysis of the reaction mixtures resulted
in [M + 2H]2+ peaks observed at 758.46 and 791.45 Da,
consistent with formation of hydrazones 4b and 5b,
respectively (Supporting Information). Overall, these experi-
ments with aldehyde-containing peptides 4a and 5a suggest
that hydrazone ligations have the advantage over oxime-
forming reactions of reaching equilibrium at a higher rate but at
the cost of lower conversion to the conjugated products due to
their lower association constants.
Aldehyde-functionalized GFPs 8a and 9a were also each
incubated with hydrazide 6b at pH 7 and room temperature
under the same conditions employed in the aforementioned
oxime ligations. LC-MS analysis of aldehyde-functionalized
GFP-containing reactions after 1 h showed approximately 20%
conversion of aldehydes 8a and 9a to their respective
hydrazones 8b and 9b (Figure 2C and 2E); more extensive
reaction times did not result in the appearance of additional
hydrazone product, suggesting that the reaction had reached
equilibrium within 1 h. These results are in good agreement
with those from hydrazone ligations for aldehyde-functionalized
peptides 4a and 5a described above.
Application to FRET Analysis of Labeled GFP. To
demonstrate the utility of this method for applications beyond
simple protein labeling, we next investigated the ability of Texas
red-labeled GFP (9b) to undergo fluorescence resonance
energy transfer (FRET). After performing the ligation reaction
between aldehyde 9a and Texas red hydrazide 6b at room
temperature for 1 h, excess fluorophore was removed via size-
exclusion chromatography. A strong fluorescent signal at 640
nm (emission wavelength of Texas red) was observed upon
excitation at 488 nm (excitation wavelength of GFP), indicative
of FRET between Texas red and GFP due to their close
Preparation and Reactivity of PFTase-Mediated
Aldehyde-Functionalized Proteins. With the ability of
aldehyde analogues 1 and 2 to be incorporated by PFTase
and their subsequent derivatization via oxime and hydrazone
ligations established in a peptide model system, we next
evaluated the utility of the aldehyde analogues for selective
protein modification. Accordingly, aldehydes 1 and 2 were
incubated with GFP−CVIA (7) in the presence of PFTase for 2
h and overnight at 30 °C, respectively. Those reaction times
were based on our earlier observations that peptide substrate 3
could be prenylated with aldehyde analogues 1 and 2 in less
8460
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 3. (A) Schematic representation of the fluorescent labeling of 9a via hydrazone ligation. Conjugated protein was expected to show FRET
■
●
▲
between Texas red and GFP−CVIA. (B) Excitation spectra obtained by monitoring at 640 nm: ( ) 9b, ( ) denatured 9b, and ( ) 7. All three
samples had equal concentrations of the chromophores.
Figure 4. (A) Schematic representation of the fluorescence labeling of 9a via oxime ligation. Conjugated protein was expected to show FRET
■
●
▲
between TAMRA and GFP. (B) Emission spectra obtained by excitation at 488 nm: ( ) 9e, ( ) GFP−CVIA (7), and ( ) 6d. All three samples
had equal concentrations of the chromophores. (C) Molecular model of GFP−TAMRA 9e conjugate.
proximity, resulting from covalent attachment of the
fluorophore to the protein (Figure 3). When the hydrazone-
ligated protein was denatured, no FRET was observed upon
excitation at 488 nm (Figure 3, red spectrum) and only a small
background peak was observed upon excitation of GFP that had
not been modified with Texas red (Figure 3, green spectrum),
further confirming that FRET was occurring between the
aldehyde-functionalized protein and the fluorophore. While the
above results appeared promising, the FRET efficiency could
not be calculated from the experimental data due to incomplete
hydrazone ligation reaction. Hence, aminooxy−TAMRA 6d
was ligated with aldehyde−protein 9a. As expected, LC-MS
analysis of the oxime formation reaction mixture showed a peak
at 28 111 Da consistent with the structure of TAMRA-labeled
8461
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
GFP 9d and showed no 9a, in good agreement with the high
efficiency observed in previous oxime ligation reactions.
Emission spectra of 9e, monitored at 488 nm excitation,
showed FRET, while the same amount of GFP−CVIA (7) and
fluorophore 6d showed a substantially larger emission band at
510 nm and a smaller band at 580 nm, respectively. Energy was
transferred from donor (GFP) to acceptor (TAMRA) with an
efficiency greater than >96%, and the distance was calculated to
be 37 Å (Figure 6B), consistent with a distance of 35 Å
calculated for the model GFP−TAMRA (Figure 6C) and
measured from the GFP chromophore to the TAMRA
fluorophore (see the Supporting Information for a description
of the modeling).
Reversible Immobilization of Purified Aldehyde-
Functionalized GFP Using Hydrazide-Modified Agarose
Beads. Next, we examined two additional applications for the
aldehyde-functionalized proteins described here. First, their
utility in protein immobilization was examined (Figure 5).
Figure 6. Immobilization onto and subsequent release of 9a from
hydrazide-functionalized agarose beads: (A) immobilization reaction
mixture in the presence of aniline, (B) release of 9d from agarose
beads via oxime ligation with hydroxylamine in the presence of aniline
for ∼3 h, and (C) control immobilization reaction containing
unmodified GFP−CVIA 7. Immobilization reaction was carried out
in the presence of protein (54 μM), aniline (100 mM), and PB (100
mM, pH 7). Release of hydrazone−GFP 9d from agarose beads was
carried out in the presence of hydroxylamine (200 mM), aniline (100
mM), and PB (200 mM, pH 7). Bright-field images are on the top, and
fluorescent microscope images are on the bottom. Scale bars in the
lower right-hand corners represent 200 μm.
in approximately 3 h 80% of the immobilized GFP was released
from the beads, and accordingly, the beads became significantly
less fluorescent (Figures 6B and 7). For comparison, the
Figure 5. (A) Schematic representation of immobilization of 9a onto
hydrazide-functionalized agarose beads to yield 9d. (B) Kinetic
analysis of immobilization of 9a onto hydrazide-functionalized agarose
beads. Reaction was carried out at room temperature in the presence
of 100 mM aniline and excess beads. UV absorbance of GFP in the
supernatant was measured at different times showing >95%
immobilization in ∼45 min. Data was fit to a simple exponential
process.
Hydrazide-functionalized agarose beads were incubated with
aldehyde−GFP 9a at room temperature in the presence of 100
mM aniline. The immobilization reaction was followed by
monitoring the UV absorbance at 488 nm of the supernatant as
a function of time. Results from those measurements showed
that equilibrium was reached in approximately 45 min; in that
time, the beads became highly fluorescent (Figure 6A); less
fluorescent beads were observed in the absence of aniline
catalyst, and no fluorescent beads were seen using GFP lacking
the aldehyde moiety (Figure 6C). On the basis of the amount
of aldehyde−GFP 9a remaining in the supernatant, the
efficiency of covalent immobilization was calculated to be
greater than 95% (Figure 5B), an impressive result for site-
specific protein immobilization. Next, oxime ligation using
hydroxylamine in the presence of aniline was employed to
remove the covalently immobilized hydrazone−GFP 9d.
Hydroxylamine (200 mM) was incubated with 9d in the
presence of aniline (100 mM) at room temperature, and the
UV absorbance at 488 nm of the supernatant was measured as a
function of time. In this case, analysis of the results showed that
Figure 7. (A) Schematic representation of the release of immobilized
GFP 9d to yield 9e from agarose beads via oxime ligation with
hydroxylamine. (B) Kinetic analysis of the release of 9e from agarose
beads by oxime ligation. Reaction was carried out at room temperature
in the presence of 100 mM aniline and 200 mM of hydroxylamine. UV
absorbance of GFP in the supernatant was measured over time, which
showed approximately 80% release of 9e in 3 h. Analysis of the
hydrolytic stability of 9d in the absence of hydroxylamine and aniline
showed no detectable release of GFP on the same time scale. Data was
fit to a simple exponential decay process.
hydrolytic stability of immobilized GFP in the absence of
hydroxylamine and aniline was also analyzed, and the results
showed that the hydrazone bond in pH 7.5 Tris buffer was
completely stable for 48 h with no detectable release of GFP.56
This achievement highlights a significant advantage of this
chemistry over click chemistry and other irreversible methods
since it can be used to efficiently covalently immobilize proteins
8462
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 8. Chemoenzymatic site-specific tagging of proteins by aldehyde−FPP analogs by PFTase followed by capture of the aldehyde-functionalized
protein in the crude cell lysate via hydrazide functionalized beads. Prenylation in the crude extract was confirmed by LC-MS analysis. Immobilized
protein was then released into the solution or fluorescently labeled by addition of hydroxylamine or an aminooxy−fluorophore using aniline as the
catalyst. SDS-PAGE analysis: lane 1, crude E. coli lysate containing 9a visualized by Coomassie blue staining; lane 2, 9c released from hydrazide beads
after treatment with 6c and visualized by Coomassie blue staining; lane 3, 9c released from hydrazide beads after treatment with 6c and visualized by
gel fluorescence analysis.
onto solid surfaces and then release them under mild
conditions without protein denaturation. Addition of aniline
catalyzes the hydrolysis of hydrazone to hydrazide and
aldehyde. Since oxime formation has a larger equilibrium
constant than hydrazone formation,57 the presence of
hydroxylamine and aniline drives the equilibrium from
hydrazone toward oxime formation and free hydrazide.
that of the starting GFP (due to addition of the aminooxy
moiety) consistent with release of GFP. In-gel fluorescence
analysis (Figure 8, lane 3) suggested that the released protein
was labeled with the fluorophore 6c; LC-MS analysis of the
released protein provided additional evidence for formation of
9c.
Application to Protein PEGylation. Attachment of
polyethylene glycol (PEG) chains to proteins is the most
widely used method for improving the pharmacokinetics of
polypeptide-based therapeutic agents.58−60 Current methods
for PEGylation are generally nonselective and can result in a
mixture of protein−PEG positional isomers with variable
biological activity.61 Site-specific methods offer a useful
alternative approach for circumventing this problem of
heterogeneity. Given our success in being able to incorporate
a fluorescent label into a protein via the capture and release
strategy described above, we decided to evaluate the utility of
this approach for preparation of a PEGylated protein. Thus,
aldehyde-functionalized GFP 9a was first prepared from
purified GFP 7 using PFTase as described above and treated
with aminooxy-PEG-10 000 10 to produce the protein−PEG
conjugate 11. Analysis of that material by MALDI-MS (Figure
9) showed an increase in molecular mass from 27.6 (for 9a) to
38 kDa for 11; the broader peak observed for 11 is consistent
with attachment of a polydisperse polymer to a monodisperse
protein. It is also important to note that no species resulting
from addition of multiple PEG chains were observed, consistent
with the selective nature of the chemistry employed here.
Analysis of the PEGylation reaction mixture by SDS PAGE
revealed a decrease in the electrophoretic mobility of 11
(Figure 10, lane 3) compared to the starting protein 9a (Figure
10, lane 2). As noted in the MALDI MS, a wider band was
observed for 11 relative to 9a, again consistent with the
polydisperse nature of the protein−PEG conjugate. With
production of the PEGylated product clearly established, we
next focused on generating the same material from 9a that had
not been purified chromatographically. Thus, 7 was prenylated
with 2 using PFTase in crude E. coli extract followed by capture
using hydrazide-functionalized agarose. After washing the
Enzymatic Modification, Immobilization, and Label-
ing in Crude Extract. An important feature of the labeling
method described here is that it uses an enzymatic process for
introduction of aldehyde groups into proteins. Due to the
specificity of that biocatalytic process and the fact that there are
no endogenous proteins in E. coli that contain a C-terminal
CAAX box sequence, we reasoned that it should be possible to
selectively functionalize proteins present in crude extract
without purification. Additionally, once modified, it should
also be possible to immobilize aldehyde-containing proteins
and release them with an alkoxyamine that includes a
fluorophore or PEG chain. In that way, a single protein present
in E. coli crude extract could be modified, immobilized, and
labeled without purification. To explore this, E. coli cells
expressing GFP−CVIA were grown, lysed, and subjected to
enzymatic prenylation using PFTase and substrate 2. LC-ESI/
MS analysis of the reaction mixture was employed to confirm
introduction of the aldehyde functionality into GFP−CVIA 7 in
the crude cell lysate. The reaction mixture was then
concentrated, and excess 2 was removed via size-exclusion
column chromatography (NAP-5 column). Aldehyde−GFP 9a
was then selectively immobilized from the crude cell lysate onto
hydrazide-functionalized beads using aniline as the catalyst.
Immobilization was followed by measuring the GFP absorbance
present in solution and judged to be complete within 45 min, at
which time the beads became highly fluorescent and the
supernatant solution became almost colorless. Next, the beads
were washed to remove any nonspecifically bound proteins and
then treated with aminooxy fluorophore 6c in the presence of
100 mM aniline overnight. SDS-PAGE analysis of the
supernatant solution showed a single band (Figure 8, lane 2)
migrating with an apparent mass of 29 kDa slightly higher than
8463
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 9. (A) Generation of site-specifically C-terminal PEGylated
GFP from pure 9a. (B) MALDI analysis of PEGylated GFP 11. Lower
panel is the MALDI spectrum of pure PEG 10, middle panel is the
MALDI spectrum of pure 9a, and top panel is the MALDI spectrum of
the oxime PEGylated GFP 11, which confirms complete conversion.
Reaction was performed using 9a (10 μM) and 10 (100 μM) for 2 h.
Excess of 10 was removed via a zip-tip protocol prior to MALDI
analysis.
Figure 10. Use of PFTase-catalyzed protein modification for site-
specific PEGylation from purified protein or crude cell lysate. (A)
Generation of site-specific C-terminal PEGylated protein from pure
9a. (B) PEGylation and release of immobilized 9d from hydrazide
beads using PEG 10. (C) SDS PAGE analysis of PEGylated GFP (11)
from purified 9a or from immobilized protein 9d. In the case of the
crude cell lysate, 7 was chemoenzymatically and site-specifically tagged
by aldehyde-containing analog 2 via PFTase-catalyzed reaction,
followed by capture of the resulting aldehyde-functionalized protein
from the lysate using hydrazide functionalized beads. Immobilized
protein was then released back into solution and simultaneously site-
specifically PEGylated by addition of aminooxy-PEG 10 using aniline
as a catalyst. SDS-PAGE analysis: lane 1, crude E. coli lysate containing
9a; lane 2, purified 9a; lane 3, 11 produced by PEGylation of pure 9a
with 10; lane 4, 11 prepared from 9d (obtained using purified 9a) and
subsequently released with 10; lane 5, 11 prepared from 9d (obtained
using 9a present in crude lysate) and subsequently released with 10.
material to remove nonspecifically bound proteins, the desired
PEGylated protein (11) was eluted via treatment with 10 in the
presence of aniline. SDS PAGE analysis showed the presence of
a single band (Figure 10, lane 5) that comigrated with the
authentic product prepared from pure 9a (Figure 10, lane 3).
PEGylation of Glucose-Dependent Insulinotropic
Polypeptide (GIP). The incretin, glucose-dependent insulino-
tropic polypeptide (GIP), is secreted from intestinal K-cells in
response to nutrient ingestion and acts to augment insulin
secretion in the pancreas. GIP has been proposed as a potential
therapeutic agent for treatment of type 2 diabetes based on its
stimulation of insulin secretion in the presence of elevated
glucose levels;62,63 however, efforts to bring GIP forward as a
drug have been hampered due to its short circulating half-life.
Recently, a modified form of GIP functionalized with a C-
terminal mini-PEG group has shown resistance to proteolytic
degradation while preserving biological activity in an obese rat
model system.52 Accordingly, having established the utility of
our method for C-terminal site-specific modification described
above with a model protein, GFP, we decided to demonstrate
its utility for preparing a PEGylated form of GIP, a polypeptide
with clear therapeutic potential. Thus, purified GIP−CVIM
(12a), a form of GIP engineered to contain a C-terminal CAAX
box (in this case CVIM64) was prenylated with analog 2 under
conditions established above for GFP and subsequently
PEGylated using a small aminooxy-functionalized PEG
containing three ethylene glycol units (13). This shorter
PEGylation reagent was employed since it is similar in length to
what has previously been shown to be effective for increasing
GIP stability in serum. MALDI MS analysis (Figure 11)
confirmed successful prenylation and PEGylation of GIP; as
noted above with GFP, both the enzymatic prenylation and the
subsequent chemical PEGylation proceed with essentially
complete conversion. Next, we employed the capture and
release strategy developed above for GFP to prepare PEGylated
GIP without prior purification. GIP−CVIM (12a), present in
crude E. coli extract, was prenylated with 2, and the resulting
aldehyde-functionalized polypeptide 12b was captured on
hydrazide beads. The beads were washed extensively and
then treated with aminooxy−PEG 13, resulting in oxime
formation and release into solution. MALDI MS analysis of the
eluted material showed only the presence of PEGylated−GIP
(14), indicating a high degree of specificity in the capture and
release (Figure 12). Thus, this general method allows facile and
effective purification of site-specifically PEGylated GIP from the
crude cell extract. Overall, these experiments conclusively
demonstrate how a protein, present in crude extract, can be
selectively modified, labeled with a fluorophore or PEG
polymer, and released in pure form via a simple process that
requires no significant chromatographic steps. Given the
specificity of the PFTase-catalyzed reaction coupled with the
ability to introduce a CAAX-box onto almost any protein, this
method shows great potential as a general approach for
selective immobilization and labeling of recombinant proteins
present in crude cellular extract without prior purification.
8464
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
Figure 11. (A) Schematic representation of prenylation of glucose-dependent insulinotropic polypeptide (GIP) containing a CAAX-box positioned
at its C-terminus (GIP-CVIM, 12a) with aldehyde-containing analogue 2 to yield the prenylated product 12b, which is then site-specifically
PEGylated using a short chain aminooxy-PEG (13). (B) MALDI MS analysis of prenylation and PEGylation of GIP 12a. MALDI MS spectra (from
the top to the bottom) of oxime PEGylated GIP 14, the prenylated aldehyde labeled GIP 12b, and pure 12a, respectively.
Figure 12. Use of PFTase-catalyzed protein modification for site-specific PEGylation of GIP 12a from crude cell lysate. (A) Chemoenzymatic site-
specific tagging of GIP 12a by aldehyde−FPP analog 2 in the crude cell lysate via PFTase followed by capture of the aldehyde-functionalized
polypeptide 12b via hydrazide functionalized beads. Immobilized polypeptide was then released back into solution and simultaneously site-
specifically PEGylated by addition of aminooxy-PEG 13. (B) MALDI analysis of the released material confirmed formation and release of the pure
PEGylated GIP (14) into the solution.
Beyond generating site-specifically modified proteins, this
approach could greatly reduce the cost of producing PEGylated
polypeptides for therapeutic applications due to the streamlined
nature of the process.
permit efficient release of proteins; covalent immobilization
using hydrazone ligation of an aldehyde-containing protein can
be followed by subsequent oxime formation to release the
polypeptide without denaturation. Using synthetically modified
alkoxyamines, a variety of new functionality ranging from
fluorescent groups to PEG chains can be appended onto
proteins. A second key feature of this approach concerns the
enzymatic method for aldehyde incorporation. By capitalizing
on the selectivity of the enzymatic process, the initial protein
functionalization can be performed using unpurified protein
substrates. The resulting modified protein can then be captured
via hydrazone formation and released via oxime formation to
produce a variety of pure, site-specifically modified protein
conjugates. Such a streamlined approach for polypeptide
modification could be particularly useful for large-scale
production of protein conjugates for therapeutic or industrial
applications. It should also be noted that the ligation chemistry
described herein and the Cu(I)-catalyzed click reaction are
CONCLUSION
■
In this work, we demonstrated that PFTase can be used to
introduce aldehyde functionality near the C-terminus of a
protein and that the resulting aldehyde-functionalized proteins
can then be modified in a plethora of ways via aniline-catalyzed
hydrazone or oxime ligation under mild conditions. We show
that if the concentration of the aldehyde-functionalized protein
is relatively high (>50 μM), hydrazone ligation is more efficient
compared with oxime ligation due to its faster kinetics, whereas
if the protein concentration is in the low micromolar range,
oxime ligation is more advantageous due to its larger
equilibrium constant. An important feature of the chemistry
reported here is its reversible nature that can be harnessed to
8465
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
(19) Chalker, J. M.; Wood, C. S. C.; Davis, B. G. J. Am. Chem. Soc.
2009, 131, 16346−16347.
orthogonal. This opens up the possibility of performing
multiple modifications on proteins using different bioorthog-
onal chemistries. Given that CAAX-box sequences can be
appended to the C-terminus of almost any protein, the method
reported here should be useful for a variety of applications in
protein chemistry.
(20) Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed.
2001, 40, 2004−2021.
(21) Blackman, M. L.; Royzen, M.; Fox, J. M. J. Am. Chem. Soc. 2008,
130, 13518−13519.
(22) Song, W.; Wang, Y.; Qu, J.; Madden, M. M.; Lin, Q. Angew.
Chem., Int. Ed. 2008, 47, 2832−2835.
ASSOCIATED CONTENT
(23) Codelli, J. A.; Baskin, J. M.; Agard, N. J.; Bertozzi, C. R. J. Am.
Chem. Soc. 2008, 130, 11486−11493.
■
S
* Supporting Information
(24) Baskin, J. M.; Prescher, J. A.; Laughlin, S. T.; Agard, N. J.;
Chang, P. V.; Miller, I. A.; Lo, A.; Codelli, J. A.; Bertozzi, C. R. Proc.
Natl. Acad. Sci. 2007, 104, 16793−16797.
Synthetic procedures and additional data and figures. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(25) Lallana, E.; Fernandez-Megia, E.; Riguera, R. J. Am. Chem. Soc.
2009, 131, 5748−5750.
AUTHOR INFORMATION
(26) Baskin, J. M.; Dehnert, K. W.; Laughlin, S. T.; Amacher, S. L.;
Bertozzi, C. R. Proc. Natl. Acad. Sci. 2010, 107, 10360−10365.
(27) Cordes, E. H.; Jencks, W. P. J. Am. Chem. Soc. 1962, 84, 826−
831.
■
Corresponding Author
E-mail: diste001@umn.edu.
Notes
(28) Dirksen, A.; Yegneswaran, S.; Dawson, P. E. Angew. Chem., Int.
The authors declare no competing financial interest.
Ed. 2010, 2023−2027.
(29) Zeng, Y.; Ramya, T. N. C.; Dirksen, A.; Dawson, P. E.; Paulson,
J. C. Nat. Methods 2009, 6, 207−209.
ACKNOWLEDGMENTS
■
(30) Khidekel, N.; Ficarro, S. B.; Peters, E. C.; Hsieh-Wilson, L. C.
Proc. Natl. Acad. Sci. 2004, 101, 13132−13137.
We thank Dr. Joseph Dalluge for helpful discussions regarding
mass spectrometry. This work was supported by the National
Institutes of Health (GM058842 and GM084152), the
University of Minnesota, and the Minnesota Supercomputer
Institute.
(31) Yi, L.; Sun, H.; Wu, Y.-W.; Triola, G.; Waldmann, H.; Goody, R.
S. Angew. Chem., Int. Ed. 2010, 49, 9417−9421.
(32) Rashidian, M.; Dozier, J. K.; Lenevich, S.; Distefano, M. D.
Chem. Commun. 2010, 46, 8998.
(33) Park, S.; Yousaf, M. N. Langmuir 2008, 24, 6201−6207.
(34) Dirksen, A.; Dirksen, S.; Hackeng, T. M.; Dawson, P. E. J. Am.
Chem. Soc. 2006, 128, 15602−15603.
REFERENCES
■
(1) Zhu, H.; Bilgin, M.; Bangham, R.; Hall, D.; Casamayor, A.;
Bertone, P.; Lan, N.; Jansen, R.; Bidlingmaier, S.; Houfek, T.; Mitchell,
T.; Miller, P.; Dean, R. A.; Gerstein, M.; Snyder, M. Science 2001, 293,
2101−2105.
(2) Luk, Y.-Y.; Tingey, M. L.; Dickson, K. A.; Raines, R. T.; Abbott,
N. L. J. Am. Chem. Soc. 2004, 126, 9024−9032.
(3) Weinrich, D.; Lin, P.-C.; Jonkheijm, P.; Nguyen, U. T. T.;
(35) Rayo, J.; Amara, N.; Krief, P.; Meijler, M. M. J. Am. Chem. Soc.
2011, 133, 7469−7475.
(36) Dixon, H. B. F. J. Protein Chem. 1984, 3, 99−108.
(37) Gilmore, J. M.; Scheck, R. A.; Esser-Kahn, A. P.; Joshi, N. S.;
Francis, M. B. Angew. Chem., Int. Ed. 2006, 45, 5307−5311.
(38) Rush, J. S.; Bertozzi, C. R. J. Am. Chem. Soc. 2008, 130, 12240−
12241.
Schroder, H.; Niemeyer, C. M.; Alexandrov, K.; Goody, R.;
̈
(39) Wang, L.; Schultz, P. G. Angew. Chem., Int. Ed. 2005, 44, 34−66.
(40) Turek, T. C.; Gaon, I.; Distefano, M. D.; Strickland, C. L. J. Org.
Chem. 2001, 66, 3253−3264.
Waldmann, H. Angew. Chem., Int. Ed. 2010, 49, 1252−1257.
(4) Zhong, M.; Fang, J.; Wei, Y. Bioconjugate Chem. 2010, 21, 1177−
1182.
(41) Duckworth, B. P.; Zhang, Z.; Hosokawa, A.; Distefano, M. D.
ChemBioChem 2007, 8, 98−105.
(5) Lin, P.-C.; Ueng, S.-H.; Tseng, M.-C.; Ko, J.-L.; Huang, K.-T.; Yu,
S.-C.; Adak, A. K.; Chen, Y.-J.; Lin, C.-C. Angew. Chem., Int. Ed. 2006,
45, 4286−4290.
(6) Goldstein, D. C.; Thordarson, P.; Peterson, J. R. Aust. J. Chem.
2009, 62, 1320.
(42) Gauchet, C.; Labadie, G. R.; Poulter, C. D. J. Am. Chem. Soc.
2006, 128, 9274−9275.
(43) Duckworth, B. P.; Chen, Y.; Wollack, J. W.; Sham, Y.; Mueller, J.
D.; Taton, T. A.; Distefano, M. D. Angew. Chem. 2007, 119, 8975−
8978.
(7) Wong, L. S.; Khan, F.; Micklefield, J. Chem. Rev. 2009, 109,
4025−4053.
(44) Labadie, G. R.; Viswanathan, R.; Poulter, C. D. J. Org. Chem.
2007, 72, 9291−9297.
(8) Lin, P.-C.; Weinrich, D.; Waldmann, H. Macromol. Chem. Phys.
2010, 211, 136−144.
(45) Subramanian, T.; Liu, S.; Troutman, J. M.; Andres, D. A.;
Spielmann, H. P. ChemBioChem 2008, 9, 2872−2882.
(46) Kale, T. A.; Hsieh, S. J.; Rose, M. W.; Distefano, M. D. Curr.
Top. Med. Chem. 2003, 3, 1043−1074.
(9) Wu, B.-Y.; Hou, S.-H.; Huang, L.; Yin, F.; Zhao, Z.-X.; Anzai, J.-I.;
Chen, Q. Mater. Sci. Eng., C 2008, 28, 1065−1069.
(10) Yamaguchi, H.; Miyazaki, M.; Honda, T.; Briones-Nagata, M. P.;
Arima, K.; Maeda, H. Electrophoresis 2009, 30, 3257−3264.
(11) Lee, Y.; Lee, E. K.; Cho, Y. W.; Matsui, T.; Kang, I.-C.; Kim, T.-
S.; Han, M. H. Proteomics 2003, 3, 2289−2304.
(12) Keppler, A.; Pick, H.; Arrivoli, C.; Vogel, H.; Johnsson, K. Proc.
Natl. Acad. Sci. 2004, 101, 9955−9959.
(47) Mu, Y.; Gibbs, R. A.; Eubanks, L. M.; Poulter, C. D. J. Org.
Chem. 1996, 61, 8010−8015.
(48) Placzek, A. T.; Gibbs, R. A. Org. Lett. 2011, 3576−3579.
(49) Wollack, J. W.; Silverman, J. M.; Petzold, C. J.; Mougous, J. D.;
Distefano, M. D. ChemBioChem 2009, 10, 2934−2943.
(50) Yakhnin, A. V.; Vinokurov, L. M.; Surin, A. K.; Alakhov, Y. B.
Protein Expression Purif. 1998, 14, 382−386.
(13) Yin, J.; Liu, F.; Li, X.; Walsh, C. T. J. Am. Chem. Soc. 2004, 126,
7754−7755.
(14) Chen, I.; Ting, A. Y. Curr. Opin. Biotechnol. 2005, 16, 35−40.
(15) Torchilin, V. P.; Lukyanov, A. N. Drug Discovery Today 2003, 8,
259−266.
(51) Inagaki, N.; Seino, Y.; Takeda, J.; Yano, H.; Yamada, Y.; Bell, G.
I.; Eddy, R. L.; Fukushima, Y.; Byers, M. G.; Shows, T. B.; Imura, H.
Mol. Endocrinol. 1989, 3, 1014−1021.
(16) Discher, D. E. Science 2002, 297, 967−973.
(17) Rizzi, S. C.; Hubbell, J. A. Biomacromolecules 2005, 6, 1226−
1238.
(52) Gault, V.; Kerr, B.; Irwin, N.; Flatt, P. Biochem. Pharmacol. 2008,
75, 2325−2333.
(53) Duckworth, B. P.; Xu, J.; Taton, T. A.; Guo, A.; Distefano, M. D.
Bioconjugate Chem. 2006, 17, 967−974.
(18) Kurpiers, T.; Mootz, H. D. Angew. Chem., Int. Ed. 2009, 48,
1729−1731.
8466
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467

Journal of the American Chemical Society
Article
(54) Gaon, I.; Turek, T.; Distefano, M. Tetrahedron Lett. 1996, 37,
8833−8836.
(55) Lenevich, S.; Distefano, M. D. Anal. Biochem. 2011, 408, 316−
320.
(56) Kalia, J.; Raines, R. T. Angew. Chem., Int. Ed. 2008, 47, 7523−
7526.
(57) Dirksen, A.; Hackeng, T. M.; Dawson, P. E. Angew. Chem., Int.
Ed. 2006, 45, 7581−7584.
(58) Jevsevar, S.; Kunstelj, M.; Porekar, V. G. Biotechnol. J. 2010, 5,
113−128.
(59) Veronese, F. M.; Mero, A.; Pasut, G. In PEGylated Protein Drugs:
Basic Science and Clinical Applications; Veronese, F. M., Ed.; Birkhauser
̈
Basel: Basel, 2009; pp 11−31.
(60) Veronese, F. M. Biomaterials 2001, 22, 405−417.
(61) Roberts, M. J.; Bentley, M. D.; Harris, J. M. Adv. Drug Delivery
Rev. 2002, 54, 459−476.
(62) Flatt, P. R. Diabet. Med. 2008, 25, 759−764.
(63) Elahi, D.; McAloon-Dyke, M.; Fukagawa, N. K.; Meneilly, G. S.;
Sclater, A. L.; Minaker, K. L.; Habener, J. F.; Andersen, D. K. Regul.
Pept. 1994, 51, 63−74.
(64) Friday, B. B.; Adjei, A. A. Biochim. Biophys. Acta 2005, 1756,
127−144.
8467
dx.doi.org/10.1021/ja211308s | J. Am. Chem. Soc. 2012, 134, 8455−8467