Isotope-coded affinity tags with tunable reactivities for protein footprinting

Article abstract of DOI:10.1002/anie.200803378

(Chemical Equation Presented) Another foot forward: Isotope-coded affinity tag (ICAT) reagents are versatile probes for footprinting dynamic protein structure. To render ICAT footprinting effective, a series of ICAT reagents with alkylation rates that span five orders of magnitude was designed. With this toolkit, an unprecedented range of Cys residues can be footprinted, including those deeply buried in the protein core and those involved in protein-protein interactions of modest affinity.

Full text of DOI:10.1002/anie.200803378

Angewandte
Chemie
DOI: 10.1002/anie.200803378
Protein Dynamics
Isotope-Coded Affinity Tags with Tunable Reactivities for Protein
Footprinting**
Eric S. Underbakke, Yimin Zhu, and Laura L. Kiessling*
Fundamental to our understanding of biological systems is the
ability to define interaction surfaces and conformational
changes in dynamic protein complexes, yet this objective
remains a challenge. Protein footprinting is a powerful
solution to this problem. Methods such as hydrogen–deute-
rium exchange,[1] limited proteolysis,[2] and radiolytic cleav-
age[3] assess changes in protein surface exposure by measuring
the susceptibility of the polypeptide backbone to modifica-
tion. A complementary approach is to use side-chain modi-
fication. Because of their unique nucleophilicity, cysteine
(Cys) residues at select positions can report on solvent
accessibility and local chemical environment.[4] The reactivity
of a Cys residue can be measured to define interaction
surfaces at individual amino acid resolution. Moreover,
methods have been developed for rapid generation of a
library of Cys variants for a protein of interest.[4a]
Figure 1. Cys alkylation rates are determined by labeling with a heavy
13C ICAT for time tAlk and then with a light 12C ICAT. After labeling,
samples are digested with trypsin, modified peptides are enriched on
affinity resin, and the resulting mixtures analyzed by mass spectrome-
try. Relative abundance of the heavy and light labels at various tAlk
define alkylation rate profiles.
In principle, Cys footprinting is amenable to mapping
accessibility changes for diverse applications, including mon-
itoring folding, ligand binding, or protein interaction inter-
faces. For such applications, the accessibility of a residue must
be correlated with its rate of reaction. Unfortunately, the
intrinsic reaction rate of Cys residues can vary widely.[4a,5]
Surface or deeply buried residues can undergo alkylation at
rates too high or low, respectively, to yield useful information.
As a consequence, a limited number of Cys substitutions can
serve as reporters. We reasoned that electrophilic footprinting
reagents with diverse reactivities could increase both the
utility and information content of Cys footprinting. Here, we
report the design and synthesis of isotope-coded affinity tag
(ICAT) reagents with a range of alkylation rates. We
demonstrate that these reagents can be used to map the
accessibility of protein residues in diverse environments. Most
notably, they can be used to footprint a protein–protein
interaction.
and light ICATreagents can be used to measure relative label
incorporation at a given position using mass spectrometry.
The affinity tag allows selective enrichment of ICAT-labeled
peptides from a complex mixture. Traditional footprinting
approaches are limited by protein size and sample complexity,
because they are incompatible with high background signals.
Because of its capacity for extensive sample refinement, we
envisioned that the ICAT approach would have tremendous
potential for investigating protein–protein interactions in
complex milieu.
Protein footprinting requires an ICATreagent that is both
small and water-soluble. A carbohydrate-based affinity tag
that binds tightly to boronate affinity resin, confers excellent
water solubility, and serves as the isotope-coding portion of
the reagent has been described.[4a] This ICAT reagent
appeared to meet our needs because it had been used to
detect changes in Cys accessibility in folded proteins.
Unfortunately, we found that it had several drawbacks.
Specifically, it possesses a chloroaromatic substituent, which
complicates sample analysis by mass spectrometry. Addition-
ally, we found its production to be inefficient and low-
yielding. To address these issues, we used glucamine as our
key building block. From this readily accessible precursor, a
series of ICAT reagents (1–4) was assembled (Scheme 1).
Specifically, the 12C compounds were generated in one step
from commercially-available glucamine by acylation. The
heavy reagents were synthesized from 13C-labeled glucose.
Reductive amination with benzylamine afforded N-benzyl-
glucamine,[7] which was subjected to palladium-catalyzed
hydrogenolysis to yield 13C-labeled glucamine.[8] Acylation
of heavy glucamine afforded the 13C-ICAT reagents 1–4 in
good overall yields.
Our approach relies on the use of electrophilic ICAT
reagents (Figure 1).[4a] ICAT reagents are trifunctional mol-
ecules composed of an electrophilic moiety, an affinity tag,
and a stable isotope (e.g., 12C or 13C) mass tag.[6] Pairs of heavy
[*] E. S. Underbakke, Dr. Y. Zhu, Prof. L. L. Kiessling
Departments of Chemistry and Biochemistry
University of Wisconsin, Madison, Madison, WI 53706 (USA)
Fax: (+1)608-265-0764
E-mail: Kiessling@chem.wisc.edu
[**] This research was supported by the NIH (GM55984). E.S.U. thanks
the Molecular Biosciences Training Grant for support (GM07215).
We thank R. E. Grant for synthetic assistance and Dr. E. E. Carlson
for helpful discussions.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200803378.
Angew. Chem. Int. Ed. 2008, 47, 9677 –9680
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9677

Communications
Figure 2. The intrinsic rates of alkylation for ICAT reagents 1–4. ICAT
reagent pairs were designed to afford a broad range of alkylation rates.
Intrinsic alkylation rates (kint) were determined using representative
Cys variants of CheW under denaturing conditions (4m guanidinium
chloride). Data were fit to a first-order exponential decay to derive rate
constants. Error bars represent the standard deviation of three
independent rate determinations.
Scheme 1. Conditions: a) BnNH2, borane·pyridine, (CH3)3CCO2H,
MeOH/H2O (1:1), 408C, 81%. b) Pd(OH)2/C, (CH3)3CCO2H, MeOH,
558C, 96%. c) 4-Maleimidobutyric acid, HBTU, HOBt, DIEA, DMF, RT,
4 h, 56%. d) (RCH2CO)2O, DMF, 08C, 12 h. Yields: 2, 81%; 3, 89%; 4,
86%. Dots denote 12C for light ICAT and 13C for heavy ICAT.
HBTU=O-(benzotriazol-1-yl)-tetramethyluronium hexafluorophos-
phate, HOBt=1-hydroxybenzotriazole, DIEA=diisopropylethylamine.
ICAT reagents 1–4 are based on a common protein
modification strategy: the alkylation of Cys residues with
maleimide or a-haloacetamide derivatives. We anticipated
that a direct comparison of the intrinsic reaction rates of the
relevant electrophiles would provide fundamental guidelines
for protein footprinting and protein modification, in general.
Accordingly, we determined the intrinsic alkylation rates (kint)
of compounds 1–4 by footprinting representative Cys variants
of a denatured protein. These conditions were chosen to
provide rates representative of fully solvent-exposed Cys
residues.
Our choice of footprinting target, the adaptor protein
CheW, was driven by an interest in the role of protein–protein
interactions in Escherichia coli chemotactic signal trans-
duction.[9] Several Cys variants (Q37C, T59C, and V64C) of
CheW were produced, and footprinting reactions were
performed under denaturing conditions. Alkylation was
initiated with heavy (13C) ICAT and quenched at various
timepoints with dithiothreitol; samples were then counter-
labeled with light (12C) ICAT (see Supporting Information).
The kint values determined for iodoacetamide 2 (ranging from
3.4–2.0mÀ1 sÀ1) are similar to a value determined independ-
ently (Figure 2).[4a] The rates of reaction of bromoacetamide 3
were approximately two-fold lower (1.7–1.0mÀ1 sÀ1), while the
rates of reaction of chloroacetamide 4 were approximately
100-fold lower (0.017–0.011mÀ1 sÀ1). We also devised a reagent
that reacts more rapidly; the maleimide-ICAT 1 provided kint
values ranging from 780 to 750mÀ1 sÀ1. Thus, the kint values of
our ICAT kinetic series cover approximately five orders of
magnitude, ranging from the extremely rapid 1 to the
relatively slow 4.
Figure 3. The rates of alkylation of Cys residues in different environ-
ments. a) Cys variants of CheW were generated at residues with
different degrees of solvent accessibility. b) The five Cys variants of
CheW were footprinted in the native, folded state (268C, pH 8.0).
Footprinting timecourses were fit to a first-order exponential decay.
Due to incomplete alkylation, I68C data are connected by a dotted line
approximating the fit. Error bars represent the standard deviation of
three independent timecourse experiments.
To survey the range of reactivity of Cys in a folded
protein, we footprinted CheW variants in which Cys residues
were engineered into diverse environments (Figure 3a). In
principle, such substitutions can perturb native protein
folding and function, but Cys replacement is generally
conservative.[10] Indeed, Cys substitution was well tolerated
at the five residues probed; the phosphotransfer activity of
chemotaxis signaling complexes reconstituted with these
CheW variants was preserved (data not shown). These
CheW Cys variants were pooled and footprinted with
bromoacetamide reagent 3, the ICAT reagent with inter-
mediate reactivity. Alkylation rates ranged from 1.0mÀ1 sÀ1
for the highly solvent-exposed Q37C to less than 0.001mÀ1 sÀ1
for the core residue I68C (Figure 3b). In the I68C CheW
9678 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9677 –9680

Angewandte
Chemie
variant, the rate of alkylation was too low to fit to a kinetic
model; therefore, this alkylation rate represents an upper
limit. Overall, the selected CheW residues exhibit reactivities
ranging over four orders of magnitude. These reaction rate
differences are consistent with previous studies in which Cys
alkylation rates within native, folded proteins can span three
to six orders of magnitude.[4a,5]
Using standard reagents, researchers have noted that
buried residues are problematic for footprinting.[12] The
extremely long reaction times required are impractical, and
they also risk protein instability and reagent cross-reactivity.
We envisioned that the rapidly-alkylating reagent 1 could
overcome the resistance of buried residues to alkylation. To
test this hypothesis, we focused on Ile68, which resides in the
core of CheW (Figure 4a).[11a] As expected, attempts to
regions often alkylate rapidly. Additionally, protein binding
typically decreases Cys reactivity less than five-fold,[4c] and
these subtle changes are below the detection limits of current
methods.[13] Indeed, applications of Cys alkylation to define
protein–protein interactions are rare. In one instance, iodo-
acetamide alkylation was used to map a high affinity anti-
body–antigen interaction,[4a] but this method failed with a
weaker protein–protein interaction (KD ꢀ 1 mm).[13] In another
instance, a weak protein–protein interaction was detected,
but only when an extremely high concentration (> 1 mm) of
the protein binding partner was employed.[4c] We reasoned
that an effective approach would be to modulate the rate of
alkylation, so that subtle rate changes in the reaction of
surface residues could be amplified and detected. Our
fundamental studies of Cys alkylation rates suggested that
ICATreagent, 4, with a relatively low kint could yield valuable
results.
To test this strategy, we examined a protein–protein
complex involved in signal transduction: the interaction of
CheA with CheW. CheW mediates a functionally important
bridging interaction between the transmembrane chemo-
receptors and the kinase CheA.[14a] Ternary complex forma-
tion couples histidine kinase activity to the ligand occupancy
of chemoreceptors.[14] The structure of the Thermotoga
maritima CheW with a portion of CheA has been deter-
mined,[11] thereby providing a context for interpreting foot-
printing results. The dissociation constant of this complex (KD
ꢀ 17 mm)[14a] is in the range typical for those of dynamic
protein–protein interactions involved in signal transduction.
Cys residues were engineered at a CheW position predicted to
be buried by CheA binding (T46C) and one that should be
unaffected (Q37C).[11b] These CheW Cys variants were pooled
and footprinted in the presence and absence of a His6-tagged,
Cys-free CheA variant (CheA*). Using the ICAT affinity tag,
the alkylated products of protease digestion were enriched
prior to analysis by mass spectrometry. Because of the high
background from the macromolecular binding partner, this
enrichment step was essential. As predicted, the t1/2 at residue
Q37C showed no significant change in the presence of CheA*
when footprinted with either reagent 2 or 4 (Supporting
Information). In native CheW, Thr46 is surface-exposed; it
therefore undergoes rapid alkylation in the presence of
reagent 2 (t1/2 = 40 s, Figure 2c). When CheA* was present,
the alkylation rate appeared to low, but the difference was not
obvious. In contrast, when reagent 4 is used for footprinting,
the addition of CheA results in an increase in the alkylation
t1/2 that exceeds two hours. Thus, the ICAT reagent 4 reveals
that CheW residue Thr46 is buried upon CheA binding. This
finding demonstrates the advantages of our approach for
footprinting protein–protein interactions.
Figure 4. Footprinting CheW Cys residues with ICAT reagents that
span a broad range of reactivity. a) The structure of T. maritima CheW
bound to a truncation of CheA[11b] was used to identify an E. coli CheW
residue buried by folding (Ile68), engaged in a protein–protein
interaction (Thr46), or unaffected (Gln37). b) Timecourse for the
alkylation of buried CheW residue I68C (pH 8.0, 268C) with iodoacet-
amide-based ICAT 2 or the more reactive agent 1. c) CheA*-induced
differences in alkylation rates of CheW surface residue T46C probed
with 2 or 4 (pH 8.0, 268C). Error bars represent the standard deviation
of three independent footprinting timecourses. Error bars smaller than
the timepoint symbol are not shown.
footprint this residue with either iodoacetamide 2 or bro-
moacetamide 3 were unsuccessful (Figure 3b, 4b). In con-
trast, alkylation with reagent 1 was complete within two
hours, and a measurable alkylation half-life (t1/2) of 11 min
was obtained. Thus, reagent 1 provides the means to monitor
changes in the accessibility of a deeply buried residue.
Previous attempts to use Cys alkylation to detect protein–
protein interactions have highlighted the difficulty of detect-
ing dynamic interactions using the standard alkylating agent
iodoacetamide.[13] Surface residues that define interaction
The ICAT reagents described herein are valuable new
tools for protein footprinting. With this toolkit and informa-
tion about the relative reactivities of each reagent, highly
solvent-exposed residues as well as deeply buried residues can
serve as reporters. The utility of these reagents is highlighted
by our demonstration that protein footprinting can be applied
to complexes of modest affinity. This finding is significant
because it is the transient, dynamic protein complexes that are
of interest in biological processes. We anticipate that our
Angew. Chem. Int. Ed. 2008, 47, 9677 –9680
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9679

Communications
15153 – 15163; f) B. P. Tu, J. C. Wang, Proc. Natl. Acad. Sci. USA
1999, 96, 4862 – 4867.
[5] M. H. Jacob, D. Amir, V. Ratner, E. Gussakowsky, E. Haas,
Biochemistry 2005, 44, 13664 – 13672.
results will expand the application of ICAT footprinting to
map and monitor diverse protein interaction surfaces.
Received: July 11, 2008
Revised: September 3, 2008
Published online: October 31, 2008
[6] a) R. Aebersold, D. R. Goodlett, Chem. Rev. 2001, 101, 269 –
296; b) S. P. Gygi, B. Rist, S. A. Gerber, F. Turecek, M. H. Gelb,
R. Aebersold, Nat. Biotechnol. 1999, 17, 994 – 999.
[7] T. Yoshida, Y. C. Lee, Carbohydr. Res. 1994, 251, 175 – 186.
[8] R. C. Bernotas, R. V. Cube, Synth. Commun. 1990, 20, 1209 –
1212.
[9] a) J. E. Gestwicki, L. L. Kiessling, Nature 2002, 415, 81 – 84;
b) A. C. Lamanna, G. W. Ordal, L. L. Kiessling, Mol. Microbiol.
2005, 57, 774 – 785; c) M. J. Borrok, E. M. Kolonko, L. L.
Kiessling, ACS Chem. Biol. 2008, 3, 101 – 109.
[10] Cys footprinting can be applied without determining that Cys
substitution preserves function, but there are caveats in inter-
preting the data that result. Specifically, disruptive Cys substi-
tutions can appear as surface residues, as they will most likely
result in an unstructured protein region.
[11] a) Y. Li, Y. F. Hu, W. Y. Fu, B. Xia, C. W. Jin, Biochem. Biophys.
Res. Commun. 2007, 360, 863 – 867; b) S. Y. Park, P. P. Borbat, G.
Gonzalez-Bonet, J. Bhatnagar, A. M. Pollard, J. H. Freed, A. M.
Bilwes, B. R. Crane, Nat. Struct. Mol. Biol. 2006, 13, 400 – 407.
[12] J. A. Silverman, P. B. Harbury, J. Mol. Biol. 2002, 324, 1031 –
1040.
Keywords: alkylation · footprinting · isotope-coded affinity tag ·
.
mass spectrometry · protein modification
[1] a) C. A. Hughes, J. G. Mandell, G. S. Anand, A. M. Stock, E. A.
Komives, J. Mol. Biol. 2001, 307, 967 – 976; b) M. M. G. Krishna,
L. Hoang, Y. Lin, S. W. Englander, Methods 2004, 34, 51 – 64;
c) D. L. Smith, Y. Z. Deng, Z. Q. Zhang, J. Mass Spectrom. 1997,
32, 135 – 146.
[2] a) S. Dokudovskaya, R. Williams, D. Devos, A. Sali, B. T. Chait,
M. P. Rout, Structure 2006, 14, 653 – 660; b) M. Zhong, L. Lin,
N. R. Kallenbach, Proc. Natl. Acad. Sci. USA 1995, 92, 2111 –
2115; c) A. Scaloni, N. Miraglia, S. Orru, P. Amodeo, A. Motta,
G. Marino, P. Pucci, J. Mol. Biol. 1998, 277, 945 – 958.
[3] a) G. Z. Xu, M. R. Chance, Anal. Chem. 2005, 77, 4549 – 4555;
b) R. T. Liu, J. Q. Guan, O. Zak, P. Aisen, M. R. Chance,
Biochemistry 2003, 42, 12447 – 12454.
[4] a) J. A. Silverman, P. B. Harbury, J. Biol. Chem. 2002, 277,
30968 – 30975; b) J. H. Ha, S. N. Loh, Nat. Struct. Biol. 1998, 5,
730 – 737; c) O. Abramczyk, M. A. Rainey, R. Barnes, L. Martin,
K. N. Dalby, Biochemistry 2007, 46, 9174 – 9186; d) C. P. John-
son, H. Y. Tang, C. Carag, D. W. Speicher, D. E. Discher, Science
2007, 317, 663 – 666; e) J. L. Apuy, Z. Y. Park, P. D. Swartz, L. J.
Dangott, D. H. Russell, T. O. Baldwin, Biochemistry 2001, 40,
[13] A. S. Burguete, P. B. Harbury, S. R. Pfeffer, Nat. Methods 2004,
1, 55 – 60.
[14] a) J. A. Gegner, F. W. Dahlquist, Proc. Natl. Acad. Sci. USA
1991, 88, 750 – 754; b) J. A. Gegner, D. R. Graham, A. F. Roth,
F. W. Dahlquist, Cell 1992, 70, 975 – 982.
9680 www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 9677 –9680