Redox-based reagents for chemoselective methionine bioconjugation

Article abstract of DOI:10.1126/science.aal3316

Cysteine can be specifically functionalized by a myriad of acid-base conjugation strategies for applications ranging from probing protein function to antibody-drug conjugates and proteomics. In contrast, selective ligation to the other sulfur-containing amino acid, methionine, has been precluded by its intrinsically weaker nucleophilicity. Here, we report a strategy for chemoselective methionine bioconjugation through redox reactivity, using oxaziridine-based reagents to achieve highly selective, rapid, and robust methionine labeling under a range of biocompatible reaction conditions. We highlight the broad utility of this conjugation method to enable precise addition of payloads to proteins, synthesis of antibody-drug conjugates, and identification of hyperreactive methionine residues in whole proteomes.

Full text of DOI:10.1126/science.aal3316

RESEARCH
◥
(Fig. 1A). As such, we sought to pursue redox re-
activity as an alternative strategy for methionine
ligation and now report a method, termed redox-
activated chemical tagging (ReACT), that en-
ables chemoselective methionine bioconjugation
(Fig. 1A) in proteins and proteomes.
RESEARCH ARTICLE
CHEMICAL BIOLOGY
Development of ReACT for chemoselective
methionine bioconjugation
Redox-based reagents
for chemoselective
Inspired by observations of facile autoxidation
of methionine residues to methionine sulfoxides
during mass spectrometry analyses, we reasoned
that an oxidative sulfur imidation reaction (29)
might serve as an attractive starting point for the
ReACT strategy, owing to the flexibility of intro-
ducing various functionalities on the nitrogen
pendant (Fig. 1A and fig. S1). We initiated our
study by screening a variety of sulfur imidation
reactions with methionine derivative S1 as a
model substrate in 1:1 CD3OD/D2O solvent using
proton nuclear magnetic resonance (1H NMR)
analysis of substrate conversion and reaction se-
lectivity between the desired N-transfer product
(NTP, sulfimide) and unwanted O-transfer prod-
uct (OTP, sulfoxide) (Fig. 1B). Surveys of various
transition metal–catalyzed sulfur imidation re-
actions were unfruitful, resulting in either no
conversion or formation of sulfoxide as the only
product (fig. S1). However, a strain-driven sulfur
imidation of methionine using oxaziridine 1 (Ox1)
as the sulfur imidation reagent afforded 95% con-
version of S1 within 2.5 min without additional
catalyst with a NTP:OTP ratio of 5:1 (Fig. 1B).
On the basis of previous reports that oxaziridines
substituted with an electron-withdrawing group
(EWG) favored formation of the OTP (30), a car-
bamate substituted with a weak EWG was ini-
tially examined (Fig. 1C). From this starting point,
altering the linkage of the probe from carbamate
to a weaker electron withdrawing urea (Ox2) re-
sulted in enhanced selectivity (NTP:OTP = 12:1)
with comparable conversion. Further attempts
to tune electronic effects by substitution of the
benzylic hydrogen of Ox2 with an electron with-
drawing CF3 group (Ox3) resulted in much lower
selectivity (NTP:OTP = 2:1) and reaction conver-
sion (58%), likely as a result of increased steric
hindrance. We observed a marked improvement
in NTP:OTP selectivity from 6:1 to 18:1 by increas-
ing the percentage of water in the solvent medium
from 0 to 95% (Fig. 1B). In accord with previously
posited hypotheses (30, 31), this improvement
likely results from increased stabilization of the
transition state leading to intermediate A, which
should be improved by solvation and hydrogen
bonding to the developing alkoxy anion (Fig. 1C).
Together, these data presage the utility of this
ligation reaction in biological environments.
We next evaluated the reactivity of oxaziridine
probes with other biologically relevant amino acid
competitors. In all cases, we did not observe any
conjugation products with any of the other ami-
no acids tested; only methionine gave a ligated
product with the ReACT reagent. Cysteine as well
as selenocysteine were oxidized to their cystine
forms, with no NTP formation observed (fig. S2),
attesting to the high selectivity of the ReACT
methionine bioconjugation
Shixian Lin,1* Xiaoyu Yang,1,8* Shang Jia,1 Amy M. Weeks,6 Michael Hornsby,6
Peter S. Lee,6 Rita V. Nichiporuk,4 Anthony T. Iavarone,4 James A. Wells,6,7
F. Dean Toste,1,5† Christopher J. Chang1,2,3,5
†
Cysteine can be specifically functionalized by a myriad of acid-base conjugation strategies for
applications ranging from probing protein function to antibody-drug conjugates and
proteomics. In contrast, selective ligation to the other sulfur-containing amino acid, methionine,
has been precluded by its intrinsically weaker nucleophilicity. Here, we report a strategy for
chemoselective methionine bioconjugation through redox reactivity, using oxaziridine-based
reagents to achieve highly selective, rapid, and robust methionine labeling under a range of
biocompatible reaction conditions. We highlight the broad utility of this conjugation method to
enable precise addition of payloads to proteins, synthesis of antibody-drug conjugates, and
identification of hyperreactive methionine residues in whole proteomes.
ulfur occupies a privileged place in biology
owing to its versatile and unique chemistry
(1). Although cysteine and methionine are
the only two sulfur-containing proteinogenic
amino acids, the sulfur center plays a diverse
In contrast to the substantial body of literature
on cysteine bioconjugation, analogous methods
for methionine labeling under physiological con-
ditions remain largely underdeveloped. Despite
a number of compelling motivations for this pur-
suit, previous methods have generally employed
highly electrophilic reagents to convert methio-
nine to sulfonium salts (21, 22) at low pH. Methi-
onine is among the most hydrophobic and the
second rarest amino acid in vertebrates, and taken
together with the fact that the majority of methi-
onine residues are buried within interior protein
cores (1), surface-accessible methionines are lim-
ited and offer a potentially valuable handle for
highly selective protein modification using natu-
rally occurring amino acid side chains. In addi-
tion, posttranslational modifications of methionine,
including by oxidation and/or metal binding (2, 23),
are emerging as critical nodes in signaling path-
ways that control function at the cell and orga-
nism level. For example, reversible oxidation of
specific methionine residues within actin can con-
trol its assembly and disassembly to serve as a
navigational signal (24, 25), and the antioxidant
function of methionine sulfoxide reductase has
been linked to regulation of life span (26). In ad-
dition, recent work suggests that methionine oxi-
dation can also increase binding interactions with
aromatic residues within proteins (27).
S
array of critical roles spanning catalysis to metal
binding to redox regulation and other posttrans-
lational modifications (2–4). In this context, se-
lective protein conjugation methods based on
cysteine modification have enabled a broad range
of fundamental and applied advances (5, 6), from
probes of protein function (4, 7–9) to synthesis
of covalent small-molecule inhibitors (10, 11) and
antibody-drug conjugates (12) to activity- and
reactivity-based protein profiling for functional
cysteine identification (7, 13, 14) and inhibitor de-
velopment (15) (Fig. 1A). Cysteine bioconjugation
strategies typically exploit the intrinsically high
nucleophilicity of the thiol/thiolate side chain, in-
cluding elegant methods based on electrophilic
warheads such as maleimides and alkyl and aryl
halides (16, 17), transition metal–mediated bio-
conjugation (18), and cysteine-to-dehydroalanine
conversion (16, 19, 20).
1Department of Chemistry, University of California, Berkeley, CA,
USA. 2Department of Molecular and Cell Biology, University of
California, Berkeley, CA, USA. 3Howard Hughes Medical Institute,
University of California, Berkeley, CA, USA. 4California Institute
for Quantitative Biosciences, University of California,
Berkeley, CA, USA. 5Chemical Sciences Division, Lawrence
Berkeley National Laboratory, Berkeley, CA, USA.
A major chemical challenge in developing a
selective methionine modification reaction under
pH-neutral physiological conditions is its rela-
tively weak nucleophilicity, which precludes the
traditional approach of identifying an appropri-
ate methionine-specific electrophilic partner for
its acid-base bioconjugation in the presence of
competing, more nucleophilic amino acids such
as cysteine, lysine, tyrosine, or serine (16, 22, 28)
6Department of Pharmaceutical Chemistry, University of
California, San Francisco, CA, USA. 7Department of Cellular
and Molecular Pharmacology, University of California, San
Francisco, CA, USA. 8School of Physical Science and
Technology, ShanghaiTech University, Shanghai, China.
*These authors contributed equally to this work. †Corresponding
author. Email: chrischang@berkeley.edu (C.J.C.); fdtoste@
berkeley.edu (F.D.T.)
Lin et al., Science 355, 597–602 (2017) 10 February 2017
1 of 6

RESEARCH
| RESEARCH ARTICLE
Fig. 1. The ReACT strategy for chemoselective
methionine bioconjugation. (A) (Left) Acid-base
conjugation strategies for cysteine-based protein
functionalization. Iodoacetamide (IAA) and malei-
mide reagents are representative electrophiles for
selective cysteine bioconjugation. (Right) ReACT
strategies for methionine-based protein functional-
ization. Oxaziridine (Ox) compounds serve as oxidant-
mediated reagents for direct functionalization by
converting methionine to the corresponding sulfimide
conjugation product. During this redox process, the
Ox ReACT reagents are reduced to benzaldehyde.
(B) Model redox conjugation reaction with 25 mM of
N-acetyl-L-methionine methyl ester (S1) and 27.5 mM
of various oxaziridine compounds as substrates in
cosolvent (CD3OD/D2O = 1:1). The reactions were
monitored by detecting the chemical shift of the
methionine methyl group with 1H NMR (fig. S1).
The reaction time was 10 min in 100% CD3OD
solution due to slow reaction rate and 20 min in
5% CD3OD /D2O solution due to poor solubility of
substrate in aqueous solution. (C) The proposed reac-
tion mechanism between methionine and oxaziridine
compound proceeds by nucleophilic attack of sul-
fide at N atom or O atom of oxaziridine ring, fol-
lowed by N–O bond cleavage to generate reaction
intermediate A or B, respectively.The NTPor OTP is
generated, along with the corresponding aldehyde
or imine as side product, through an intramolecular
rearrangement. (D) Number of unique ReACT-sensitive
Met, Lys, and Cys residues detected in HeLa cell
lysates when treated with 1 mM Ox4 for 10 min.
(E) Yield of conjugation reaction was performed
with 15 mM of BSA carrying four methionines per
protein and 100 mM Ox4 at the indicated time point
as measured by in-gel fluorescence imaging. Error
bars, mean SD from three independent exper-
iments. Representative fluorescent gel is shown in
fig. S4.
reagent for methionine functionalization over its
sulfur congener. As a further demonstration of the
high selectivity of ReACT for methionine conjuga-
tion, we next identified sites of probe labeling within
a whole proteome using liquid chromatography–
tandem mass spectrometry (LC-MS/MS) analysis.
HeLa cell lysates were treated with Ox4, trypsin
digested, and then analyzed by LC-MS/MS for
probe modification on all nucleophilic amino acids
using the X!Tandem program (32). We observed
labeling of 235 methionine residues and a single
lysine residue, with no other modifications de-
tected on cysteine side chains or other nucleo-
philic amino acids (Fig. 1D and Data S1). These
experiments demonstrate the fast kinetics of the
ReACT strategy as well as near-perfect selectivity
for methionine residues from the single-protein
to whole-proteome level under mild biocompat-
ible conditions. Finally, we tested the chemical
stability of the sulfimide methionine conjugation
product, finding that this linkage showed reason-
able stability to acidic and basic conditions, as well
as treatment with a strong protein disulfide re-
ducing agent such as tris(2-carboxyethyl)phosphine
(TCEP) (fig. S3). The product is stable to exposure
to elevated temperatures (80°C) for short times
(1 hour), but prolonged exposure (18 hours) can
lead to substantial decomposition (fig. S3).
protein using a two-step labeling protocol, BSA at a
concentration of 15 mM was first treated with
100 mM oxaziridine probe Ox4 bearing a bio-
orthogonal alkyne group and then subsequently
coupled to Cy3-azide through a copper-catalyzed
azide-alkyne cycloaddition (CuAAC) reaction. The
resulting redox conjugation yield to BSA was an-
alyzed by in-gel fluorescence imaging. ReACT pro-
ceeds rapidly and can be completed with a yield
>95% within 1 to 2 min, with 50% of labeling
occurring within the first 5 s after the addition of
Ox4 to the protein under standard reaction con-
ditions (Fig. 1E and fig. S4). Moreover, the mea-
sured second-order rate constant for reaction of
methionine (at 5- to 40-fold excess) with Ox2 in
phosphate-buffered saline is 18.0 0.6 M−1 s−1,
which is comparable to what is observed for the
ReACT-based bioconjugation to
methionines in model protein substrates
We then evaluated ReACT as a method for site-
selective methionine conjugation of proteins. Start-
ing with bovine serum albumin (BSA) as a model
Lin et al., Science 355, 597–602 (2017) 10 February 2017
2 of 6

RESEARCH
|
RESEARCH ARTICLE
using the antibody fragment to green fluorescent
protein (GFP-Fab) as a starting model. We noted
that although the Fab scaffold possesses one
native methionine residue on its light chain
and two native methionine residues on its heavy
chain, none of these side chains are surface ac-
cessible and thus were not labeled by ReACT,
even with high oxaziridine probe loadings (Fig.
3A and fig. S8). As such, because these native
methionines are buried within the hydrophobic
interior core, ReACT offers a potentially valuable
strategy for Fab bioconjugation because there is
no background labeling of the wild-type Fab and
subsequent engineering of surface-accessible me-
thionine sites can enable precise antibody func-
tionalization at directed locations. Using the
THIOMab platform (34), we demonstrated this
possibility by replacing heavy chain (HC)–A114
or light chain (LC)–V205 residues with methi-
onine and showed efficient labeling with ReACT
(fig. S8). In addition, we established rapid, near-
quantitative, and site-specific C-terminal labeling
with ReACT on a GFP-Fab bearing a C-terminal
methionine (GFP-Fab-CM) (Fig. 3, A and B). The
resulting azide-carrying GFP-Fab (GFP-Fab-N3)
retained similar binding affinity to the GFP lig-
and compared to the wild-type Fab (fig. S9). Click
reactions enabled further functionalization of
GFP-Fab-N3 with biotin, fluorophore, and drug
payloads (Fig. 3B). Moreover, the resulting con-
jugates were compatible with biological environ-
ments. For example, we used a human embryonic
kidney 293T (HEK-293T) cell line with a doxycy-
cline (Dox)–inducible cell surface GFP expression
system, where Dox treatment results in expres-
sion of GFP localized to the cell surface. Upon
pre-addition of Dox followed by incubation with
Cy3-labeled GFP-Fab made by ReACT, we ob-
served excellent colocalization of Cy3 and GFP
signals in live HEK-293T cells. In contrast, no
Cy3 signal was observed in control cells without
Dox addition (Fig. 3C). Furthermore, the intensity
of the Cy3 signal was stable for at least 14 days in
the presence of 100% fetal bovine serum (fig. S10).
Taken together, these data demonstrate that
ReACT can enable antibody functionalization
at directed positions with a wide variety of pay-
loads and simultaneously retain their function
for ligand binding.
R
O
NH
O
N
100
S
N
O
N
R
WT protein
MS = 16790 Da
S
H
80
60
40
20
0
R=function group
9 Mets
Ox2
MS+9ΔM2
WT-Calmodulin
Ox4
MS+9ΔM4
MS+8ΔM2+O
Ox5
MS+9ΔM5
Ox2
Ox4
O
Ox2
MS+8ΔM3+O
N
N
N
N
N
H
ΔM3=86
O
O
O
O
Ox6
WT+9ΔM6
O
O
O
Ox4
N
H
MS+8ΔM5+O
MS+8ΔM6+O
ΔM4=96
Ox5
Ox5
O
N
H
ΔM5=140
Ox6
16500
Ox6
N
H
N3
17000
17500
18000
18500
ΔM6=141
Molecular mass (Da)
Fig. 2. The ReACT strategy for protein functionalization. (A) General two-step procedure for
methionine-specific protein functionalization is a combination of ReACT and click reactions. Various pay-
loads (red sphere) can be installed through methionine conjugation at a directed position on a given protein.
(B) Redox conjugation of a CaM model protein (100 mM) with various Ox compounds (1 mM). The chemical
structures of oxaziridine probes are shown with molecular weight changes (DM) listed for the corresponding
modifications.The deconvoluted MS data of full protein peaks are plotted in the same figure.The major peaks
correspond to CaM protein carrying nine sulfimide modifications (DM). For Ox2-labeled protein: expected
mass 17,564 Da, found 17,565 Da; Ox4-labeled protein: expected mass 17,654 Da, found 17,654 Da; Ox5-
labeled protein: expected mass 18,050 Da, found 18,051 Da; Ox6-labeled protein: expected mass 18,059 Da,
found 18,060 Da. The minor peaks correspond to CaM protein bearing eight sulfimide modifications (DM)
and one sulfoxide modification (O).
CuAAC reaction (fig. S5). We then moved on to ap-
ply ReACT to modification of calmodulin (CaM) as
a model protein with redox-sensitive methionines
(33). CaM carries nine redox-active methionine
residues, and upon pretreatment with increasing
concentrations of hydrogen peroxide, we observed
the expected dose-dependent decrease in Ox4 la-
beling as these residues were oxidized from me-
thionine to methionine sulfoxide, the latter of
which is insensitive to ReACT (fig. S6). Only me-
thionine residues were identified carrying the
desired modification by LC-MS/MS analysis of Ox4-
labeled calmodulin, with no probe-generated con-
jugation modification observed on any other ami-
no acids (fig. S6). The results presage the potential
application of ReACT to probe redox-sensitive
methionines by distinguishing them from their
oxidized forms.
alkyne- and azide-containing oxaziridine probes
(1 mM) with a 100-mM CaM model protein (Fig. 2B).
LC-MS/MS results showed that ReACT enabled
near quantitative installation of these bioorthog-
onal handles on all nine native methionine resi-
dues within 10 min of labeling time at room
temperature, with a 25:1 selectivity over the
only other observed minor CaM product bearing
eight sulfimide modifications (NTP) and one
sulfoxide modification (OTP) (Fig. 2B). Further
functionalizations with biotin, fluorophore, and
polyethylene glycol (PEG) payloads proceeded
smoothly (fig. S7). The high methionine reac-
tivity and specificity of ReACT coupled with the
ready availability of click reaction partners
provides a straightforward method for precise
protein functionalization based on this naturally
occurring amino acid (Fig. 2A).
We next moved on to apply ReACT to a thera-
peutic conjugate, Herceptin-Fab (Her-Fab). As
expected, ReACT did not label wild-type Her-Fab,
owing to its lack of surface-accessible methio-
nines (fig. S11). By engineering Her-Fab platforms
carrying one or two methionine residues at the
C terminus of the light chain, ReACT afforded
near quantitative conjugation with one or two
redox modifications, respectively (Fig. 3D). The
data establish that ReACT can enable synthesis
of ADCs with a defined drug-to-antibody ratio
(DAR) in excellent purity, which remains a major
challenge for bioconjugation methods employ-
ing cysteine or lysine ligation. Moreover, the bio-
orthogonal azide or alkyne handle introduced by
methionine conjugation can be readily function-
alized with additional payloads. The ADC synthe-
sized by linking monomethyl auristatin E (MMAE)
With these data in hand, we envisioned that
ReACT could enable installation of various payloads
onto a protein of interest at defined methionine
sites, serving as a method for functionalization
using naturally occurring amino acids (Fig. 2A).
To this end, we evaluated the reactivity of various
Application of ReACT to
antibody bioconjugation
To showcase the utility of ReACT in a pilot bio-
logical application, we turned our attention to
the synthesis of antibody-drug conjugates (ADCs),
Lin et al., Science 355, 597–602 (2017) 10 February 2017
3 of 6

RESEARCH
| RESEARCH ARTICLE
to Her-Fab exhibited a fivefold increase in tox-
icity to Her2-positive BT474 breast cancer cells
compared with either wild-type Her-Fab or a mix-
ture of wild-type Her-Fab and free MMAE (Fig. 3E),
demonstrating its utility in a biological context.
of cells with low, medium, and high levels of
ReACT probe Ox4 (Fig. 4A and fig. S12), we
sought to identify hyperreactive methionines that
should be enriched with low-dose labeling along
with less-reactive methionine sites. By perform-
ing parallel TOP-ABPP (n = 2 for all three groups)
in HeLa cell lysates, we were able to identify 116
(low dose), 458 (medium dose), and 1111 (high
dose) peptides that carry the desired ReACT me-
thionine modification (Fig. 4, A and B, and Data
S2). Compilation of the hyperreactive methionine-
containing target proteins identified in the low-
dose ReACT-treated group spanned many protein
classes, including enzymes, chaperones, and nu-
cleoproteins, as well as many structural proteins
(Data S2). In general, only surface-accessible me-
thionine residues were identified even with the
high-dose probe, indicating that ReACT does not
disrupt or denature proteins under these label-
ing conditions (Fig. 4C and fig. S13).
Of particular interest are the hyperreactive me-
thionine targets, because they can predict sites of
methionine-regulated protein function. This un-
biased ReACT approach not only enables char-
acterization of previously studied redox-sensitive
methionines in whole-proteome settings but more
importantly identifies new functional methionine
sites. As a positive control, we identified three
hyperreactive methionines within actin, includ-
ing Met44 and Met47, whose redox activities
have been previously shown to play a central role
in controlling actin polymerization in living cells
(Fig. 4C) (24, 25). With these data validating the
ReACT method in hand, we moved on to identify
and characterize new targets with methionine-
dependent function. As one representative exam-
ple, we found three hyperreactive methionine
residues on enolase, a central enzyme in the an-
cient and conserved metabolic pathway of gly-
colysis (36), which is important in regulating
Application of ReACT to
chemoproteomic identification
of functional methionines in cells
Finally, we turned our attention to the use of
ReACT as a methionine-targeted warhead for
chemoproteomics applications, owing to its high
specificity and reactivity, as well as the small
warhead size of the oxaziridine group (e.g., the
molecular weight of Ox4 is 202 Da) that allows
access to a broad range of proteins. To this end,
we applied ReACT to probe reactive methionines
in the proteome through tandem orthogonal
proteolysis–activity–based protein profiling (TOP-
ABPP) (35). Through dose-dependent treatment
Fig. 3. ReACT for synthesis of methionine-targeted
antibody conjugates. (A) Crystal structure of
Her-Fab [Protein Data Bank (PDB) 1n8z] with three
native methionine residues shown as green sticks
and sulfur atoms shown as yellow spheres.The sulfur
atoms appear to be buried in the pocket of Fab on the
crystal structure. (B) The ratio of probe/methionine
for Fab labeling is 10, with 10-min labeling time at
room temperature. The deconvoluted MS data of
GFP-Fab constructs with or without ReACT label-
ing. GFP-Fab-CM: expected mass 47158 Da, found
47158 Da; GFP-Fab-N3 (labeled by Ox6): expected
mass 47299 Da, found 47299 Da; biotin-functionalized
GFP-Fab-N3 (GFP-Fab-Biotin): expected mass 48019 Da,
found 48019 Da; Cy3-functionalized GFP-Fab-N3
(GFP-Fab-Cy3): expected mass 48282 Da, found
48282 Da. MMAE-functionalized GFP-Fab-N3 (GFP-Fab-
MMAE): expected mass 48949 Da, found 48949 Da.
(C) Fluorescence colocalization imaging of GFP-Fab-
Cy3 with cell surface–targeted GFP in HEK-293T
cells.The GFP was inducibly expressed on the cell
surface with addition of 1 mM/mL of Dox before add-
ing GFP-Fab-Cy3. Cells without addition of Dox were
used as a control and show no antibody staining. All
images use the same scale bar: 20 mm. (D) The de-
convoluted MS data of Her-Fab constructs with or
without ReACT labeling. Her-Fab carries one C-
terminal methionine (Her-Fab1): expected mass
47544 Da, found 47544 Da; Ox6-labeled Her-Fab1
(Her-Fab1-N3): expected mass 47685 Da, found
47686 Da; Her-Fab carries two C-terminal methi-
onine (Her-Fab2): expected mass 47676 Da, found
47677 Da; Ox6-labeled Her-Fab2 (Her-Fab2-N3): ex-
pected mass 48958 Da, found 47958 Da. (E) In vitro
cytotoxicity of Her-Fab2 [median effective concen-
tration (EC50) = 0.086 0.02 mg/mL], noncovalent
mixture of Her-Fab2 and free MMAE (EC50 = 0.096
0.04 mg/mL), and the ReACT-derived ADC from
Her-Fab2 and MMAE (EC50 = 0.015 0.007 mg/mL).
100
GFP-Fab-CM
GFP-Fab-N3
GFP-Fab-Biotin
GFP-Fab-Cy3
GFP-Fab-MMAE
LC-4
HC-100c
HC-82
80
60
40
20
0
46500
47500
48500
49500
Molecular mass (Da)
C-terminus
+ Dox
GFP
Cy3
Cy3
Merge
BF
BF
- Dox
GFP
Merge
100
Her-Fab2
100
80
60
40
20
0
Her-Fab1
Her-Fab1-N3
Her-Fab2
Her-Fab2-N3
Her-Fab2 + Drug
ADC
80
60
40
Error bars, mean
SD from three independent
20
0
experiments. EC50 values and EC50 SDs were de-
termined using four-parameter logistic fitting.
-3
-2
-1
0
1
2
47000
47500
48000
48500
Molecular mass (Da)
Log (Concentration), µg/mL
Lin et al., Science 355, 597–602 (2017) 10 February 2017
4 of 6

RESEARCH
| RESEARCH ARTICLE
diseases such as cancer via the Warburg effect
(37). In this scenario, cancer cells predominantly
produce more energy compared with nontumor-
igenic cells by manipulating glycolysis enzymes
(e.g., enolases). Among these three methionine
residues in enolase, Met169 residue is highly
conserved from yeast to mammals (correspond-
ing to Met171 on yeast enolase 1) (Fig. 4D and fig.
S14). Moreover, this residue is close to the en-
zyme active site and can be oxidized along with
other methionine residues in the mammalian pro-
tein upon oxidant treatment (Fig. 4D and fig. S15).
To characterize the functional importance of
this oxidation-sensitive methionine in enolase in
more detail, we cloned and purified a yeast homo-
log for in vitro biochemistry studies. Treatment
of the wild-type yeast enolase 1 with hypochlorite
decreased enzymatic activity with concomitant
oxidation of methionine residues, including Met171,
on the protein (Fig. 4E). A similar decrease in
protein activity was observed upon oxidation
of the M371L mutant. In contrast, the activity
of the M171L mutant was unaffected by oxidant
treatment under the same conditions (Fig. 4E),
suggesting that this highly conserved residue is
critical for redox regulation of enolase function.
Kinetics measurements of the wild type and M171L
mutants with and without oxidant treatment
revealed that both the turnover number (Kcat)
and the Michaelis constant (Km) are affected
in the wild type upon oxidation but that these
values for the M171L mutant remain the same
(fig. S15). To show the physiological consequen-
ces of this methionine-based redox regulation at
the cellular level, we generated yeast strains with
a double enolase 1 and enolase 2 knockout back-
ground (38) and reintroduced either wild-type
enolase 1 or the M171L mutant. As depicted in
Fig. 4F, we observed that the strain carrying
the M171L mutation is more resistant to oxida-
tive stress–induced cell death compared to the
strain with wild-type enolase 1, establishing that
Fig. 4. Chemoproteomic methionine profiling
with ReACT. (A) Reactive methionine profiling
with ReACT involves treatment of proteomes with
low, medium, and high doses of Ox4 probe, fol-
lowed by CuAAC-based installation of acid cleav-
able biotin-azide tag, enrichment with streptavidin
magnetic beads, and sequential on-bead trypsin
digestions to afford probe-labeled peptides for
LC-MS/MS analysis. (B) The number of peptides
carrying the desired ReACTmodification on methi-
onine and appearing in two independent runs from
the low-, medium-, and high-dose groups is shown.
(C) Reactive methionine map on actin (PDB 3byh),
with hyperreactive methionines colored in red, in-
cluding Met44 and Met47, medium-reactive methi-
onines colored in purple; and less-reactive methionines
colored in blue. The methionines colored in yellow
represent residues identified by LC-MS/MS that do
not carry redox modification.The domains carrying
these yellow-colored methionines are involved in
actin polymerization. Because of this activity, no
desired modification is detected on these residues
even when these are surface accessible on the pro-
tein x-ray crystal structure. (D) Proteinstructure align-
ment of human alpha enolase (yellow; PDB 2psn)
and yeast enolase 1 (red; PDB 2AL1), with con-
served methionine residues shown in stick repre-
sentation. (E) The relative activity of yeast enolase 1
variants with or without treatment of NaClO (100 mM).
Error bars, mean SD from four independent ex-
periments. P values indicated in the figure repre-
sent results of an unpaired t test. (F) Growth curve
of wild-type (WT) (ENO2 null) and ENO1-M171L
(ENO2 null) strains, with or without treatment of
NaClO (100 mM). Knockout and mutation strains
were generated by clustered regularly interspaced
short palindromic repeats (CRISPR)–Cas9–mediated
genome editing. Error bars, mean SD from three
independent experiments.
Lin et al., Science 355, 597–602 (2017) 10 February 2017
5 of 6

RESEARCH
| RESEARCH ARTICLE
this methionine residue serves a functional redox-
active role in vivo.
6. S. Bondalapati, M. Jbara, A. Brik, Nat. Chem. 8, 407–418 (2016).
7. E. Weerapana et al., Nature 468, 790–795 (2010).
8. S. Mizukami, Y. Hori, K. Kikuchi, Acc. Chem. Res. 47, 247–256
(2014).
9. T. F. Brewer, F. J. Garcia, C. S. Onak, K. S. Carroll, C. J. Chang,
Annu. Rev. Biochem. 84, 765–790 (2015).
10. I. M. Serafimova et al., Nat. Chem. Biol. 8, 471–476 (2012).
11. J. M. Ostrem, U. Peters, M. L. Sos, J. A. Wells, K. M. Shokat,
Nature 503, 548–551 (2013).
12. P. Agarwal, C. R. Bertozzi, Bioconjug. Chem. 26, 176–192 (2015).
13. D. K. Nomura, M. M. Dix, B. F. Cravatt, Nat. Rev. Cancer 10,
630–638 (2010).
14. L. E. Sanman, M. Bogyo, Annu. Rev. Biochem. 83, 249–273
(2014).
15. D. Leung, C. Hardouin, D. L. Boger, B. F. Cravatt,
Nat. Biotechnol. 21, 687–691 (2003).
16. O. Boutureira, G. J. L. Bernardes, Chem. Rev. 115, 2174–2195
(2015).
17. C. Zhang et al., Nat. Chem. 8, 120–128 (2016).
18. E. V. Vinogradova, C. Zhang, A. M. Spokoyny, B. L. Pentelute,
S. L. Buchwald, Nature 526, 687–691 (2015).
19. T. H. Wright et al., Science 354, aag1465 (2016).
20. A. Yang et al., Science 354, 623–626 (2016).
21. H. G. Gundlach, S. Moore, W. H. Stein, J. Biol. Chem. 234,
1761–1764 (1959).
22. J. R. Kramer, T. J. Deming, Chem. Commun. (Camb.) 49,
5144–5146 (2013).
23. A. Kaya, B. C. Lee, V. N. Gladyshev, Antioxid. Redox Signal. 23,
814–822 (2015).
24. R.-J. Hung, C. W. Pak, J. R. Terman, Science 334, 1710–1713
(2011).
25. B. C. Lee et al., Mol. Cell 51, 397–404 (2013).
26. J. Moskovitz et al., Proc. Natl. Acad. Sci. U.S.A. 98,
12920–12925 (2001).
27. A. K. Lewis et al., Nat. Chem. Biol. 12, 860–866
(2016).
28. D. A. Shannon, E. Weerapana, Curr. Opin. Chem. Biol. 24,
18–26 (2015).
29. V. Bizet, C. M. M. Hendriks, C. Bolm, Chem. Soc. Rev. 44,
3378–3390 (2015).
33. J. R. Erickson et al., Cell 133, 462–474 (2008).
34. J. R. Junutula et al., Nat. Biotechnol. 26, 925–932
(2008).
35. A. E. Speers, B. F. Cravatt, J. Am. Chem. Soc. 127,
10018–10019 (2005).
36. L. A. Fothergill-Gilmore, P. A. M. Michels, Prog. Biophys. Mol.
Biol. 59, 105–235 (1993).
37. M. G. Vander Heiden, L. C. Cantley, C. B. Thompson, Science
324, 1029–1033 (2009).
38. P. G. Leonard et al., Nat. Chem. Biol. 12, 1053–1058
(2016).
39. E. A. Kabat, T. T. Wu, H. Bilofsky, J. Biol. Chem. 252,
6609–6616 (1977).
Conclusion
To close, ReACT provides a unique and general
redox-based approach to chemoselective methi-
onine conjugation that complements the wealth
of acid-base conjugation methods for modifica-
tion of more nucleophilic amino acids such as
cysteine, lysine, and serine. The relative rarity of
methionine in surface-accessible forms, such as
in the complementarity-determining region of
antibodies (39) that show highly similar sequence
homology, presages the potential utility of this
method for functionalization of antibodies and
their fragments as well as other proteins using
naturally occurring amino acids that are native
or readily introduced by standard site-directed
mutagenesis. More broadly, the ReACT method
enables installation of various payloads onto ther-
apeutic and imaging proteins at well-defined
positions and with excellent DAR. Finally, the
selectivity of the oxaziridine warhead offers a
chemical platform for identifying and studying
functional methionines in whole proteomes, as
well as a starting point for therapeutic interven-
tions based on reactive methionine activation
and/or inhibition.
ACKNOWLEDGMENTS
We thank the U.S. Department of Energy/Lawrence Berkeley
National Laboratory (101528-002 to C.J.C. and DE-AC02-05CH11231
to F.D.T.) for financial support, as well as NIH (GM79465 to
C.J.C.) for pilot protein labeling studies. C.J.C. is an Investigator of
the Howard Hughes Medical Institute and a Canadian Institute for
Advanced Research Senior Fellow. We also thank NIH for grant
1S10OD020062-01 in financial support of University of California
(UC) Berkeley QB3 mass spectrometry facilities. We thank A. Killilea
(UC Berkeley Tissue Culture Facility) and M. Salemi and B. Phinney
(UC Davis Proteomics Core) for expert technical assistance, as well as
S. Pollock for the HEK-293T-GFP cell line, V. Yu for yeast genome
editing, and D. Nomura for helpful discussions. A.M.W. is a Merck
Fellow of the Helen Hay Whitney Foundation. P.S.L is supported
by a postdoctoral fellowship from NIH (F32CA203152). The
experimental data sets used in this study are available as
supplementary materials. S.L., X.Y., F.D.T., and C.J.C. are listed
as inventors on a patent application describing redox-active
reagents for methionine conjugation.
SUPPLEMENTARY MATERIALS
REFERENCES AND NOTES
www.sciencemag.org/content/355/6325/597/suppl/DC1
Materials and Methods
Figs. S1 to S26
1. M. H. Stipanuk, Annu. Rev. Nutr. 24, 539–577 (2004).
2. R. L. Levine, L. Mosoni, B. S. Berlett, E. R. Stadtman, Proc. Natl.
Acad. Sci. U.S.A. 93, 15036–15040 (1996).
Data S1 and S2
30. J. Vidal et al., Chem. Eur. J. 3, 1691–1709 (1997).
31. A. Armstrong, I. D. Edmonds, M. E. Swarbrick, Tetrahedron
Lett. 44, 5335–5338 (2003).
References (40–68)
3. N. M. Giles et al., Chem. Biol. 10, 677–693 (2003).
4. C. E. Paulsen, K. S. Carroll, Chem. Rev. 113, 4633–4679
(2013).
2 November 2016; accepted 11 January 2017
10.1126/science.aal3316
32. R. Craig, R. C. Beavis, Bioinformatics 20, 1466–1467 (2004).
5. C. D. Spicer, B. G. Davis, Nat. Commun. 5, 4740 (2014).
Lin et al., Science 355, 597–602 (2017) 10 February 2017
6 of 6

Redox-based reagents for chemoselective methionine
bioconjugation
Shixian Lin, Xiaoyu Yang, Shang Jia, Amy M. Weeks, Michael
Hornsby, Peter S. Lee, Rita V. Nichiporuk, Anthony T. Iavarone,
James A. Wells, F. Dean Toste and Christopher J. Chang (February
9, 2017)
Science 355 (6325), 597-602. [doi: 10.1126/science.aal3316]
Editor's Summary
Targeting proteins at the other sulfur
As the only amino acid with a thiol (SH) group, cysteine is easily targeted for site-selective
protein modifications. Hydrophobic methionine also has sulfur in its side chain, but its capping methyl
group has hindered analogous targeting efforts. Lin et al. introduce a complementary protocol to tether
new substituents exclusively to methionine, even in the presence of cysteine. They used an oxaziridine
group as an oxidant to form sulfimide (S=N) linkages. The approach allowed antibody-drug
conjugation and chemoproteomic screening for reactive methionine surface residues.
Science, this issue p. 597
This copy is for your personal, non-commercial use only.
Visit the online version of this article to access the personalization and
article tools:
Article Tools
Permissions
http://science.sciencemag.org/content/355/6325/597
Obtain information about reproducing this article:
http://www.sciencemag.org/about/permissions.dtl
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week
in December, by the American Association for the Advancement of Science, 1200 New York
Avenue NW, Washington, DC 20005. Copyright 2016 by the American Association for the
Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.