Small Molecule Interactome Mapping by Photoaffinity Labeling Reveals Binding Site Hotspots for the NSAIDs

Article abstract of DOI:10.1021/jacs.7b11639

Many therapeutics elicit cell-type specific polypharmacology that is executed by a network of molecular recognition events between a small molecule and the whole proteome. However, measurement of the structures that underpin the molecular associations bet

Full text of DOI:10.1021/jacs.7b11639

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Small Molecule Interactome Mapping by Photoaffinity Labeling
Reveals Binding Site Hotspots for the NSAIDs
Jinxu Gao, Adelphe Mfuh, Yuka Amako, and Christina M. Woo*
Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford St., Cambridge, Massachusetts 02138, United States
S
* Supporting Information
ABSTRACT: Many therapeutics elicit cell-type specific poly-
pharmacology that is executed by a network of molecular
recognition events between a small molecule and the whole
proteome. However, measurement of the structures that
underpin the molecular associations between the proteome
and even common therapeutics, such as the nonsteroidal anti-
inflammatory drugs (NSAIDs), is limited by the inability to map
the small molecule interactome. To address this gap, we
developed a platform termed small molecule interactome
mapping by photoaffinity labeling (SIM-PAL) and applied it to
the in cellulo direct characterization of specific NSAID binding
sites. SIM-PAL uses (1) photochemical conjugation of NSAID derivatives in the whole proteome and (2) enrichment and
isotope-recoding of the conjugated peptides for (3) targeted mass spectrometry-based assignment. Using SIM-PAL, we identified
the NSAID interactome consisting of over 1000 significantly enriched proteins and directly characterized nearly 200 conjugated
peptides representing direct binding sites of the photo-NSAIDs with proteins from Jurkat and K562 cells. The enriched proteins
were often identified as parts of complexes, including known targets of NSAID activity (e.g., NF-κB) and novel interactions (e.g.,
AP-2, proteasome). The conjugated peptides revealed direct NSAID binding sites from the cell surface to the nucleus and a
specific binding site hotspot for the three photo-NSAIDs on histones H2A and H2B. NSAID binding stabilized COX-2 and
histone H2A by cellular thermal shift assay. Since small molecule stabilization of protein complexes is a gain of function
regulatory mechanism, it is conceivable that NSAIDs affect biological processes through these broader proteomic interactions.
SIM-PAL enabled characterization of NSAID binding site hotspots and is amenable to map global binding sites for virtually any
molecule of interest.
Global proteomic profiles from small molecule affinity
purification strategies are commonly obtained using mass
spectrometry (MS);8 however, the vast majority of proteomics
studies stop short of obtaining direct structural evidence for the
molecular interaction. Inherent challenges have prevented the
mapping of small molecule binding sites to the whole
proteome. Small molecule interactions occur over a range of
concentrations that require a general mechanism for capture
and enrichment prior to MS analysis. The conjugation
chemistry to capture the binding event must be rapid and
general for unbiased covalent bond formation at the small
molecule binding site. Binding sites of small molecules to
defined protein isolates can be determined by application of
photoaffinity labeling (PAL) to conjugate the small molecule to
the protein prior to MS analysis.9 Yet, the demand for a general
chemical strategy to covalently conjugate a small molecule
locally to the protein interaction site poses great challenges to
spectral assignment by database searching. Database searching
methods are not adapted to the computational complexity
INTRODUCTION
■
Polypharmacology, wherein one drug interacts with multiple
protein targets, is an outcome of molecular recognition events
between small molecules and the whole proteome. Poly-
pharmacology manifests in increased efficacy when properly
exploited or tragic unanticipated off-target effects when not
fully understood. Many pharmaceuticals in diverse therapeutic
areas possess either known or uncharacterized polypharmacol-
ogy,1 such as the nonsteroidal anti-inflammatory drugs
(NSAIDs),2 the immunomodulatory drugs,3 or the opioids.4
Differences in protein networks across cell types form the basis
for molecular interactions that culminate in an observed
phenotype,5 suggesting that a map of the protein−ligand
interaction network may eventually predict these polypharma-
cology outcomes.6,7
A method to directly map the noncovalent small molecule
interactome has the potential to accelerate drug discovery by
providing structural insight and instant validation of the binding
interaction, yet such a characterization is rarely performed.
Common methods to structurally reveal small molecule binding
sites, such as X-ray crystallography or NMR spectroscopy, are
constrained to the measurement of stable interactions between
a single compatible protein and small molecule pair in vitro.
Received: November 2, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
yielded by amino acid residue-agnostic conjugation chemistry
to the whole proteome.
Translation of MS-based binding site identification by PAL
from a single protein to the whole proteome thus requires (1) a
selective chemical workflow to isolate the conjugated peptide
and (2) a targeted MS technique for confident characterization.
PAL covalently conjugates small molecules to the proteome for
stringent enrichment of interacting proteins.10 Application of
cleavable enrichment handles enables recovery of the small
molecule-conjugated peptide.9 Critically, a targeted MS
strategy, wherein unique isotopic markers are installed
specifically to the small molecule conjugated peptide, provides
an orthogonal handle for detection and validation that is
transformative during database assignment of peptides carrying
heterogeneous modifications by MS.11 Recent strides have
enabled the identification of fragment-based small molecule
ligands to the whole proteome.12
The knowledge gap caused by the lack of a small molecule
interactome map extends to common pharmaceuticals like the
NSAIDs. The NSAIDs potently suppress inflammation, pain,
and fever, and have been further explored as potential
treatments for cancer2,13 and Alzheimer’s disease.14 NSAID
mechanisms have been primarily characterized through
inhibition of the enzymes cyclooxygenase-1 and -2 (COX-1,
COX-2, respectively).15 Inhibition of COX-2, the primary
cyclooxygenase involved in inflammation, prevents the
production of prostaglandins, thereby reducing inflammation.
However, a wealth of biomedical evidence points to broader
COX-2-independent mechanisms of the NSAIDs for which a
molecular basis remains poorly defined.16−18 Prior studies
suggest that specific NSAIDs inhibit the nuclear factor-κB (NF-
κB) pathway19 and caspases.20 A detailed understanding of
broader NSAID mechanisms requires a global understanding of
NSAID−protein interactions and their underlying structures.
Herein, we report the development of a platform termed
small molecule interactome mapping by photoaffinity labeling
(SIM-PAL) and its application to the NSAIDs. SIM-PAL is
designed to characterize the protein interactions and direct
binding site hotspots of a small molecule in a whole cell
proteome using a PAL-based enrichment strategy coupled to
isotope-targeted MS (Figure 1A). Our platform involves: (1)
photoconjugation of NSAID derivatives in cells, (2) enrichment
and isotopic recoding of NSAID-labeled peptides, and (3)
isotope-targeted assignment of the conjugated peptides. We
show that photo-NSAID derivatives are effective reporters of
NSAID binding sites with recombinant COX-2 and the global
whole cell proteome from Jurkat and K562 cell lines. By virtue
of direct characterization of the conjugated peptide, we
localized the photo-NSAIDs to nearly 200 binding sites from
over 150 proteins, including histones H2A and H2B. Histone
H2A was stabilized by interacting with the NSAIDs by cellular
thermal shift assay. SIM-PAL revealed the precise binding sites
for the photo-NSAIDs via an approach that is readily translated
to broad classes of small molecules.
Figure 1. Strategy to profile NSAID interactome by SIM-PAL. (A)
Photo-NSAIDs are applied in cellulo and photoconjugated to protein
binding partners. Conjugated proteins are tagged by probe 10 using
CuAAC and enriched to separately obtain the protein interactome and
conjugated peptides representing binding site hotspots. Conjugated
peptides are analyzed by isotope-targeted MS. (B) Structures of
NSAIDs naproxen (1), celecoxib (2), and indomethacin (3), their
photo-NSAID analogs 4−6, and the negative controls tag 7, the
orthogonal compound photoglutarimide 8, and the celecoxib analog 9.
COX-2 IC50 by ELISA is shown below. (C) Structure of the cleavable
biotin azide probe 10. Probe 10 is prepared as a 1:3 mixture of stable
12C:13C isotopes (highlighted in red).
RESULTS
■
Development of Photo-NSAIDs as Reporters of NSAID
Binding Sites. Three NSAIDs, naproxen (1), celecoxib (2),
and indomethacin (3), were conjugated to diazirine-based
photoaffinity labels (“photo-NSAIDs”) to serve as reporters for
NSAID binding sites (Figure 1B). These NSAIDs were selected
for their different structure−activity relationship (SAR)
between COX-1 and COX-2.21,22 Naproxen (1) is a non-
B
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
selective COX-1 and COX-2 inhibitor, but is commonly
employed for chronic use due to low rates of gastrointestinal
side effects.23 Celecoxib (2) is a selective COX-2 inhibitor
developed by Pfizer, yet possesses off-target cardiovascular and
gastrointestinal complications.24 Indomethacin (3) is a member
of the indole class of NSAIDs and possesses known COX-2-
independent anti-inflammatory mechanisms.16 The design of
photo-NSAIDs was based upon previous SAR studies25,26 and
the crystal structure between mouse COX-2 and indomethacin
(3).27 During our studies, a crystal structure of a NSAID with
human COX-2 was reported.28 In addition to photo-NSAIDs
4−6, we developed the tag 7,29 a structurally orthogonal
photoglutarimide 8, and a celecoxib analog 9 to assess
selectivity of the binding site identification assay (Scheme S1).
All three photo-NSAIDs maintained COX-2 inhibition by
ELISA (Figure 1B, Figure S1),30 although some variation in
activity was observed. Photocelecoxib (5) was the most potent
(IC50 = 36.6 nM), and photonaproxen (4) was the least potent
(IC50 = 36.0 μM). All photo-NSAIDs possessed antiprolifer-
ative properties within 1.3−1.4-fold of their parent compound
in Jurkat cells (Figure S2). Elimination of the sulfonamide from
celecoxib (2) to give the analog 9 is known to reduce COX-2
inhibition,26 and we found additionally attenuated antiprolifer-
ative activity of the analog 9 in Jurkat cells (Figure S2).
Furthermore, all photo-NSAIDs were competitively displaced
from recombinant COX-2 by the native NSAID (Figure 2A).
COX-2 was separately incubated with each of the photo-
NSAIDs with or without a 100-fold excess of the parent
compound as a competitor.10 The samples were photo-
irradiated, tagged with the fluorophore TAMRA-azide by
copper-mediated azide−alkyne cycloaddition (CuAAC) and
fluorescently visualized to reveal selective and reversible
binding of COX-2 to the photo-NSAIDs. The tag 7 did not
produce observable conjugation to COX-2 by fluorescence.
Photo-NSAIDs Possess Known and Transient Binding
Sites with COX-2. We used recombinant COX-2 to validate
fragmentation patterns of conjugated NSAIDs to a protein and
to determine binding site selectivity for each of the photo-
NSAIDs. Small molecule modification on a peptide may
perturb MS fragmentation pathways in unexpected ways,
rendering the spectra unassignable by database searching. To
evaluate this possibility, photo-NSAIDs (10 μM) were
incubated with recombinant COX-2 (1 μg) for 30 min and
photo-irradiated. The irradiated samples were appended to the
cleavable biotin azide probe 10 to simulate the conjugated
species ultimately observed after enrichment (Figure 1C). The
probe 10 is a multifunctional probe developed to possess a
biotin affinity enrichment handle, an acid-labile diphenylsilane,
and a stable isotope-coded azidoacetate for CuAAC and
isotope-targeted MS. We previously established the compati-
bility of a similar cleavable probe scaffold in targeted MS
experiments.11 Following CuAAC, the samples were trypsin-
digested and the biotin probe was cleaved in situ (2% formic
acid). The resulting peptides were analyzed by LC-MS/MS. MS
data was searched by SEQUEST HT against recombinant
COX-2 with the photo-NSAID as a modification on any amino
acid (Table S1). Due to the nature of photochemical
conjugation, a binding site may be represented by multiple
conjugation events to several amino acid residues on one or
more peptides. All peptide spectral matches (PSMs) assigned
to a conjugated peptide were manually validated.
Figure 2. Comparative analysis of NSAIDs and photo-NSAIDs with
recombinant COX-2. A. Fluorescence image of photo-NSAID binding
to COX-2 and competitive displacement by the respective parent
compound. COX-2 (125 ng) was incubated with photo-NSAIDS with
or without the parent molecule, photo-conjugated, and tagged with
TAMRA-azide. B. Docking structure of photonaproxen (4, red) and
naproxen (1, blue) with COX-2. C. Docking structure of photo-
celecoxib (5, green) and celecoxib (2, blue) with COX-2. D. Docking
structure of photoindomethacin (6, yellow) and indomethacin (3,
blue). E. Docking structure of the tag 7 (purple) with COX-2. The
blue box indicates the region of COX-2 that is enlarged in sections B−
D. COX-2 (1 μg) was separately conjugated to each of the photo-
NSAIDs (10 μM) or the tag 7 (10 μM), tryptically digested, and
analyzed by MS on an Orbitrap Elite. Conjugated peptides observed
by MS are highlighted for each photo-NSAID. Docked structures were
either the lowest desolvation energy or highest interface area size
created by Patchdock (October, 2017). Structure of COX-2 from
PDB: 5KIR.
the photo-NSAIDs 4−6, no irregular fragmentation pathways
were observed. As expected, at a consistent dose of 10 μM
across all photo-NSAIDs, photocelecoxib (5) possessed the
highest number of observed conjugated peptides (seven
conjugated peptides from 14 PSMs), including within the
active site of COX-2. Analysis of COX-2 treated by
photonaproxen (4) and photoindomethacin (6) revealed six
Photo-NSAIDs were readily assigned by database searching.
Manual inspection of these PSMs indicated that in the case of
C
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
conjugated peptides across 19 PSMs and eight PSMs,
respectively (Table S1). By contrast, the tag 7 was found
conjugated to one peptide, which was not marked by the
photo-NSAIDs. Within each peptide, the specific conjugation
site localized to a range of 2−4 amino acid residues that reflect
the specificity of the photochemical conjugation event.
were assigned by SEQUEST HT. Two biological replicates
were collected for each of the photo-NSAIDs that displayed
high reproducibility across the enriched proteome (>60%) and
protein abundance (Figure S6A). Using the normalized spectral
abundance factor (NSAF) for PSM-based label free quantifi-
cation,34−36 proteins that were statistically significant (t test, p-
value <0.05) and the PSM ratio was ≥2-fold enriched were
considered significantly enriched by the photo-NSAID relative
to the control tag 7. Following application of the NSAF, a
standard distribution of values was obtained to test for
statistical significance. The majority of proteins that passed
statistical significance were also enriched by the photo-NSAIDs
as compared to the tag 7 (Figure 3A). This analysis yielded
approximately 700 proteins significantly enriched by at least
one of the photo-NSAIDs in Jurkat cells (Table S2). High
proteomic overlap among photo-NSAIDs, but not negative
controls 7 and 8, was observed. Across each of the three
compounds, at least 40% of the identified photo-NSAID-
binding proteins were enriched by all three photo-NSAIDs and
53% of the proteins were enriched by at least two photo-
NSAIDs (Figure 3B). By comparison, 140 proteins that were
significantly enriched by any photo-NSAID were also enriched
by the photoglutarimide 8 (24%).
Jurkat cells were initially dosed with a concentration of 250
μM of each photo-NSAID to maximize downstream observa-
tion of conjugated peptides by MS. Although naproxen (1)
enters blood plasma at concentrations that surpass 250 μM,37
our cell viability data against Jurkat cells showed an IC50 range
of 23−216 μM across the NSAIDs and their derivatives (Figure
S2). We thus examined the NSAID interactome at 50 μM and
found 260 proteins enriched only by the three photo-NSAIDs
and not by the tag 7 (Table S3). Enrichment was determined
by inclusion of proteins only observed in the photo-NSAID
treated samples. At 50 μM, the proteomic overlap between the
photo-NSAIDs was lower, indicative of higher binding
selectivity between the ligands (Figure S6B). A majority of
proteins identified at 50 μM were likewise identified at 250 μM
for each photo-NSAID (86−92%). The celecoxib analog 9 (50
μM) was additionally tested in activated Jurkat cells and
displayed moderate proteomic overlap with photocelecoxib (5)
(Figure S6C).
To determine the generality of these observations, photo-
NSAID interactions were measured against K562 cells, a human
chronic myeloid leukemia cell line. Several lines of evidence
point to NSAID-dependent inhibition and apoptosis of K562
cells.38 A total of 513 proteins were significantly enriched across
two biological replicates from K562 cells, of which 42% of the
proteins were enriched by at least two of the three photo-
NSAIDs (Figures S7 and S8). Significantly enriched proteins
from K562 cells possessed a moderate overlap with proteins
from Jurkat cells (206 proteins), indicating a high degree of
specificity across cell lines (Figure 3C). In sum, a total of 1033
proteins were considered significantly enriched from Jurkat and
K562 cells by photo-NSAIDs (Table S2).
Photo-NSAIDs interacted with proteins distributed through-
out the cell (Figure 3D). Approximately two-thirds of the
NSAID interactome localized to the nucleus and cytoplasm.
Photo-NSAIDs additionally captured proteins annotated as
localized to the mitochondria (10%), endoplasmic reticulum
and Golgi (11%), and membrane or secreted proteins (10%).
These data are a close reflection of the natural distribution of
proteins throughout the cell.39 Only 30% of these proteins were
previously annotated as interacting with a small molecule, let
The photo-NSAIDs, the parent compounds, and the tag 7
were structurally minimized [Gaussian 16, basis set: HF 3-
21g(d)] and individually docked in the crystal structure of
human COX-2 (PDB: 5KIR).28 Docking was performed using
Patchdock, a molecular docking algorithm based on shape
complementarity using rigid structures.31 Structures with the
lowest desolvation energy or highest interface area size docked
the photo-NSAID or parent compound to the same binding
site, although the orientation between each pair of compounds
differed (Figure 2B−D). The docked structure of photo-
naproxen (4, red), photocelecoxib (5, green), and photo-
indomethacin (6, yellow) overlaid with their respective parent
compound (blue). Conjugated amino acid residues within 5 Å
of the docked photo-NSAID are highlighted in Figure 2B−D.
The orientation of each of the photo-NSAIDs positioned the
diazirine in close proximity to one or two specific amino acids
on a conjugated peptide observed by MS. Other marked
residues were located on solvent exposed areas of the protein.
As photochemistry captures dynamic processes and the
photochemical tag is structurally flexible, the additional
conjugation events on recombinant COX-2 may represent
transient interactions with COX-2 in vitro, which may not be
observed in cellulo. The tag 7 was additionally docked to COX-
2 and, in combination with the observed MS data, was shown
to transiently bind to an orthogonal region of COX-2 (Figure
2E). No conjugated peptides from amino acids near to the
canonical binding site were observed with the tag 7.
Characterization of the NSAID interactome in Jurkat
and K562 cells. Confident that photo-NSAIDs were
producing defined linkages with COX-2 that recapitulated
NSAID activity, we next sought to characterize global NSAID
interactions within the whole cell proteome. Photo-NSAIDs,
the tag 7, and the photoglutarimide 8 were added to activated
Jurkat T cells as a model system for inflammation. Small
molecule labeling was photo-irradiation-dependent and dose-
dependent up to 250 μM (Figure S3A,B). Competitive
displacement of the tightest COX-2 binder photocelecoxib
(5) by the parent compound occurred at a 1:10 molar ratio
(Figure S3C). Jurkat cells were stimulated with phorbol
myristyl acetate and ionomycin for 18 h prior to photo-
NSAID exposure.32 The stimulated Jurkat cells were exposed to
each compound (250 μM, 1 h) and photo-irradiated in situ to
conjugate the small molecule to the proteome. The resulting
NSAID-conjugated proteins were enriched using the probe 10
in a biotin-dependent manner (Figure S4). COX-2 from Jurkat
cells was enriched by photo-NSAIDs, and this enrichment was
abrogated by competition with the parent compound, as
indicated by Western blot (Figure S5A). Cellular thermal shift
assay, a mechanism to measure stabilization of protein−ligand
interactions,33 showed that COX-2 was stabilized by celecoxib
(2) and photocelecoxib (5) at an apparent aggregation
temperature (Tagg) over 10 °C relative to the tag 7 (Figure
S5B,C).
The enriched proteomes were digested with trypsin, and the
released peptides were analyzed by LC-MS/MS on an Orbitrap
Fusion Tribrid with collision-induced dissociation (CID) and
higher energy CID (HCD) fragmentation modes. MS data
D
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
protein complexes that were enriched include members of the
proteasome and the adaptor protein complex 2 (Figure 3E).
These proteins may have existed as complexes leading to
photo-NSAID conjugation by virtue of proximity or were
enriched due to associative protein interactions. While protein
complexes can be enriched through associative protein−protein
interactions, our use of strong dissociative detergents to prepare
cell lysates (1% Rapigest, sonication) does not typically lead to
observation of protein complexes following enrichment.11
Direct Photo-NSAID Interaction Mapping Reveals
Binding Site Hotspots. Following tryptic digestion of the
photo-NSAID proteome, the probe 10 was acid cleaved to
release and isolate the conjugated peptides from the enrich-
ment media. The conjugated peptides were analyzed by
application of isotope-targeted MS. A unique isotopic signature
was embedded to the probe 10 using a 13C-derived stable
isotope ratio of 1:3 over [M:M + 2] spacing to perform
isotope-targeted MS. During click chemistry, the unique
isotopic signature was exclusively transferred to the photo-
NSAIDs conjugated to the proteome. The isotopic signature is
therefore only found on small molecule-conjugated peptides
and is used during isotope-targeted MS to overcome barriers in
detection of low abundance species and validation of modified
peptides against a background of unmodified peptides.
During data collection, the isotopically recoded species is
immediately recognizable by full scan MS and may be selected
for fragmentation by use of an inclusion list.41,42 This selection
process increases the fraction of isotopically recoded, small
molecule-conjugated peptides selected for fragmentation.11
More critically, the isotope signature played a crucial role in
manual validation of database search assignments due to the
ambiguity of the amino acid modification site. Database
searching was performed against the tryptic protein database
with each of the photo-NSAIDs as a modification on any amino
acid residue. Small molecule conjugation to any amino acid
drastically increases the size of the protein database. For
example, a single modification increases the size of the fully
tryptic human protein database by 60-fold, while two
modifications increase the database size by 1000-fold. This
exponential increase in the tryptic peptide database leads to a
breakdown in false discovery rate (FDR).43 We thus used a
two-tier validation process for confident assignment of
conjugated peptides. First, modified peptides at 5% FDR
were filtered based on visual inspection of the MS2 spectral
assignment. Second, each of the precursor spectra were
individually validated for the isotopic signature in the MS1.
Based on this analysis, 575 PSMs, corresponding to 194
individual conjugated peptides, were characterized across the
photo-NSAIDs, tag 7, and photoglutarimide 8 (Table S4).
Selected frequently observed conjugated peptides are displayed
in Table 1.
Figure 3. Photo-NSAID protein interactome at 250 μM. (A) Fold
change enrichment of the photo-NSAIDs against the tag 7 based on
PSMs. Proteins are ordered in descending fold change ratio. Proteins
that were not statistically significant across the two biological replicates
are not displayed. (B) Overlap across the Jurkat photo-NSAID
significantly enriched proteome. (C) Proteomic overlap between
significantly enriched proteins from Jurkat and K562 cells. (D)
Subcellular localization of the 1033 significantly enriched proteins. (E)
Selected protein interaction networks captured by at least one photo-
NSAID. Color scheme: red = significantly enriched; pink = ≥2-fold
enriched; gray = identified in data; white = member of the protein
complex not identified in data. Dashed border indicates direct
observation of at least one conjugated peptide. Gray lines indicate
the protein interaction network, built using CORUM as a reference
(July, 2017). Half maximal effective concentration (EC50) values for
inhibition of the NF-κB pathway as determined by a NF-κB luciferase
reporter assay shown below.
The individual peptides derived from approximately 150
proteins, of which over 90% of the proteins were ≥2-fold
enriched. Forty-three of these proteins were considered
significantly enriched by at least one of the photo-NSAIDs. A
number of isotopically coded species that were either not
selected or not confidently assigned by SEQUEST were also
observed. Detection of enriched conjugated species using a
pattern searching algorithm42 revealed nearly 1000 isotopically
coded precursor ions in the MS1 spectra across our photo-
NSAID conjugated peptide data.
alone one of the NSAIDs (BindingDB, ChEMBL, DrugBank),
pointing to the broader range of molecular interactions that
remain to be revealed by the reported binding site mapping
approach.
In line with evidence for capture of a broader range of
associated proteins, a number of protein complexes were
selectively enriched. Comparison of enriched proteins to
CORUM40 revealed that NF-κB subunits (NFKB1, NFKB2)
are directly interacting with the photo-NSAIDs, confirming
their known inhibition of the NF-κB pathway,2 which was
revalidated for celecoxib (2) and photocelecoxib (5) via a NF-
κB luciferase reporter assay (Figure 3E, Figure S9). Additional
Each ligand produced a unique profile of conjugated
peptides. Photocelecoxib (5) and photonaproxen (4) repre-
E
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 1. Selected Conjugated Peptides Observed with High PSM Frequency Across Photo-NSAIDs, the Tag 7, and the Photo-
a
Glutarimide 8
count of conjugated peptide PSMs
conjugated peptide
AMGIMNSFVNDIFER
protein (GENE)
4
5
6
7
8
1
histone H2B type 1-K (HIST1H2BK)
histone H2A type 2-B (HIST2H2AB)
actin, cytoplasmic 1 (ACTB)
41
8
12
28
14
12
7
1
9
3
5
1
2
VGAGAPVYLAAVLEYLTAEILELAGNAAR
VAPEEHPVLLTEAPLNPK
3
3
4
VGAGAPVYMAAVLEYLTAEILELAGNAAR
MSVQPTVSLGGFEITPPVVLR
NLEALALDLMEPEQAVDLTLPK
IHFPLATYAPVISAEK
histone H2A type 2-A (HIST2H2AA3)
Nucleophosmin (NPM1)
2
5
7
6
X-ray repair cross-complementing 6 (XRCC6)
tubulin α-1A chain (TUBA1A)
7
3
1
2
7
13
8
8
VGAGAPVYLAAVLEYLTAEILELAGNAARDNKK
ISGLIYEETR
histone H2A type 2-B (HIST1H2AB)
histone H4 (HIST1H4A)
5
2
5
4
5
9
10
7
1
10
11
12
13
14
15
16
17
VETGVLKPGMVVTFAPVNVTTEVK
AIGAVPLIQGEYMIPCEK
elongation factor 1-α 1 (EEF1A1)
Cathepsin D (CTSD)
2
3
1
2
AFADAMEVIPSTLAENAGLNPISTVTELR
NASDMPETITSR
T-complex protein 1 subunit delta (CCT4)
minor histocompatibility antigen H13 (HM13)
glutathione S-transferase P (GSTP1)
adenosine deaminase (ADA)
2
4
2
AFLASPEYVNLPINGNGK
4
3
LKNDQANYSLNTDDPLIFK
FGPYYTEPVIAGLDPK
proteasome subunit β type-3 (PSMB3)
ATP synthase β, mitochondrial (ATP5B)
EGNDLYHEMIESGVINLK
a
Presented data are in aggregate across Jurkat and K562 cells. For the full dataset, see Table S4.
sented the bulk of the identified interactions and were found
conjugated to 93 and 85 peptides, respectively. Photo-
indomethacin (6) was conjugated to 34 peptides in total. A
degree of overlap between the photo-NSAID-conjugated
peptides was observed, where 30 peptides were identified by
at least two of the three photo-NSAIDs. Of these, a single
conjugated peptide on Ku70, a member of the DNA repair
pathway, was observed by two photo-NSAIDs, the tag 7 and
photoglutarimide 8 (entry 6, Table 1). The remaining 14
conjugated peptides detected by the tag 7 and nine conjugated
peptides from the photoglutarimide 8 were detected exclusively
by that compound.
Conjugated peptides were mapped to proteins found
throughout the cell. Peptides from nuclear proteins (entries
1, 2, 4, 5, 6, 8, 9, 14) were most frequently observed (Table 1).
Mitochondrial proteins (entry 17) and cell surface or secreted
proteins (entries 11, 12, 13) were also observed. NSAID
interactions were mapped to diverse protein types, including
enzymes (entries 6, 11, 14, 15, 17), parts of the proteasome
(entries 11, 16), and structural proteins (entries 3, 7), which
demonstrates the unbiased discovery power of MS. In many
cases, these interactions were highly specific and observed with
only one of the photo-NSAIDs (entries 7, 14, 15, 16, 17).
We found a number of PSMs assigned to NSAID-conjugated
histone peptides (Figure 4A, Figure S10). All directly observed
histone interactions were highly interconnected by clustering
and imply significantly upregulated conjugation of histone
complexes by the photo-NSAIDs (Figure 4B). These proteins
were enriched in photo-NSAID proteomic data as compared to
the tag 7, but were not considered statistically significant. In
particular, two peptides from histone H2A and histone H2B
were primarily detected. The histone H2B peptide was detected
in a total of 54 PSMs across our data sets, predominantly by
photonaproxen (4, entry 1, Table 1). Histone H2A type 2-A
and type 2-B were detected in a total of 62 and 19 PSMs,
respectively (entries 2, 4, and 8, Table 1), predominantly in
Jurkat cells. These PSMs related to peptides conjugated to
photocelecoxib (5), followed by photonaproxen (4), and to a
lesser degree photoindomethacin (6). By virtue of directly
observing the conjugated peptide, photo-NSAIDs were mapped
to a specific binding site hotspot around these two peptides,
which are in close proximity in structures of the nucleosome
(Figure 4C).44
Validation of SIM-PAL data by competition assay revealed
that binding of all three photo-NSAIDs with histone H2A and
H2B was competitively exchanged with the parent compound
by Western blot (Figure 5A). The interaction of photo-NSAIDs
with NF-κB p65, cathepsin D, and nucleophosmin was
additionally displaced by the parent compound. Selective
photoconjugation and partial competition was observed when
probing for Ku70/XRCC6, in agreement with the conjugated
peptide PSM frequency (entry 6, Table 1). The tag 7 and
photo-NSAIDs labeled elongation factor 1-α 1 and rho GDP-
dissociation inhibitor 2 similarly, although partial competition
was still observed in some cases. Cellular thermal shift assay in
Jurkat cells showed stabilization of histone H2A by all three
NSAIDs and the celecoxib analog 9 relative to the tag 7 by 10−
15 °C (Figure 5B). The Tagg was estimated by densitometry
and fitted to a Boltzmann sigmoid equation. Photonaproxen
(4) and photocelecoxib (5) also stabilized histone H2A, while
photoindomethacin (6) possessed a destabilizing effect. These
data are in line with the PSM frequency in Table 1. Further
characterization of these binding sites for biological function
will be pursued.
DISCUSSION
■
The development and application of SIM-PAL to enable
characterization of the small molecule interactome and direct
binding interactions is described (Figure 1A). SIM-PAL
provides several key advantages, including: (1) instant
validation of the interaction by virtue of direct observation,
(2) measurement of a concentration-dependent range of
interactions, including transient and highly specific binding
events, and (3) structural information about the binding site
and the small molecule structure. These advantages are
highlighted by our analysis of three NSAIDs for their binding
sites with a single protein (COX-2) and against the whole
proteome from Jurkat and K562 cells.
F
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 4. continued
complexes with at least one observed binding site to a photo-NSAID.
High confidence interactions between two clusters made using
STRING. (C) Structure of the nucleosome (PDB: 2CV5). Peptides
from histone H2A (red) and histone H2B (blue) that were conjugated
by all three photo-NSAIDs are highlighted.
We applied SIM-PAL to three NSAIDs, naproxen (1),
celecoxib (2), and indomethacin (3), due to their different
structures yet similar anti-inflammatory effects. Variation in
their off-target toxicity and the affected COX-2-independent
anti-inflammatory pathways has been reported with a limited
molecular basis for these effects. For example, celecoxib was the
result of a medicinal chemistry optimization for COX-2
selectivity relative to COX-1,22 yet additional COX-2-
independent mechanisms have been subsequently reported.18
To illuminate the molecular driving force behind NSAID
biology, we developed a set of photo-NSAIDs (4−6, Figure 1B)
that recapitulated the activity of the parent compounds and
demonstrated their direct binding interaction with COX-2
(Figure 2). Due to the dynamic nature of PAL, we identified
additional transient interactions with COX-2 in vitro and the
broader proteome in cellulo. These data report on access and
molecular recognition of the compounds to the proteome and
afford a structural basis for further functional analysis or protein
degradation strategies.45 A systematic study of the kinetics of
PAL will provide concrete measurement of the range of
transient interactions observable by SIM-PAL.
Application of the photo-NSAIDs 4−6 to stimulated Jurkat
and K562 cells at 250 μM revealed over 1000 significantly
enriched protein interactions that were more homologous
across the three photo-NSAIDs than with the tag 7 or the
photoglutarimide 8. Analysis of the photo-NSAID protein
interactome revealed entire protein complexes of both known
and novel interactions. For example, specific proteins in the
NF-κB pathway are now identified as direct interactors (Figure
3E and Figure S9). Enrichment of several protein complexes
also indicates that the photoconjugation event is dynamic and
highly specific to the local environment of the small molecule at
the time of activation. Novel interactions with the proteasome
and the adaptor protein complex 2 were also notable.
Observation of these proteins may be in part due to the
broader range of interactions captured by photoconjugation as
compared to existing target identification strategies. Compar-
ison of photo-NSAID interactions with proteomics profiles
derived from fragment-based small molecules12 revealed a 60%
overlap. Thus, photo-NSAIDs possess a protein interaction
profile that is specific to the molecular structure and cell type.
The photoconjugation step is highly concentration depend-
ent. Analysis of the photo-NSAIDs at 50 μM yielded 260
enriched proteins across the three compounds. The high
proteomic overlap observed at 50 μM and 250 μM implies that
the protein interactions enriched at 50 μM are a more stringent
subset of the photo-NSAID interactome. Carrying out the
described photoconjugation experiments at functionally
relevant concentrations is important for enrichment of binding
partners that would be physiologically observed. These results
provide insight into the molecular interactions within the
proteome generated by the NSAIDs, which may or may not
possess direct biological function.
Figure 4. Spectral matches of photo-NSAID binding site hotspots to
histone proteins. (A) Precursor pattern distribution (MS1) and
database assignment (MS2) for a histone H2A peptide conjugated to
each of the photo-NSAIDs. (B) Cluster diagram of histone protein
Subsequent evaluation of the conjugated peptides using
isotope-targeted MS revealed nearly 200 conjugated peptides,
G
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
significant by the label-free quantification method.34 The direct
observation of these conjugated peptides provides novel insight
into real molecular interactions that are occurring in the whole
cell proteome. Although the majority of conjugated peptides
were identified by only one of the photo-NSAIDs, 30 peptides
were observed conjugated to two or more of the photo-
NSAIDs (Table 1, Table S4). Selection of proteins for follow
up functional studies is typically based on existing biological
relevance, relative abundance, or dose dependence. SIM-PAL
reports on both the enriched proteome and directly on the
conjugated peptide, which provides a higher resolution picture
to select proteins for further study.
We elected to evaluate a specific conjugation event that was
observed on an interface between histone H2A and H2B in the
nucleosome due to the high frequency of PSMs in our MS data.
The two peptides are within 6 Å of each other in a crystal
structure of the nucleosome, indicative of a binding site hotspot
recognized by the photo-NSAIDs.44 While all three photo-
NSAIDs were able to conjugate these peptides, photonaproxen
(4) conjugation to histone H2B was observed the most
frequently, while photocelecoxib (5) was most frequently
observed conjugated to histone H2A. Comparison of these data
to fragment-based small molecule interaction sites revealed that
a diazirine-tagged coumarin, most structurally similar to
photonaproxen (4), likewise conjugated a similar peptide
from histone H2AZ.12 The interaction of photo-NSAIDs with
histones was competitively displaced by the parent compound
by Western blot (Figure 5A). Although a functional attribution
cannot be inferred simply by observation of an interaction,
celecoxib (2) and photocelecoxib (5) stabilized histone H2A by
cellular thermal shift assay (Figure 5B). Stabilization of protein
complexes leading to downstream signaling changes is
reminiscent of other immunomodulators, including cyclo-
sporin, FK506, and rapamycin,46 and the immunomodulatory
drug lenalidomide.47,48 The interaction may also represent a
more promiscuous binding site hotspot of histones with small
molecules that has never previously been observed, analogous
to the influence of serum albumin on drug pharmacology.49
The photo-NSAIDs reported here are validated probes that
may be applied to additional biological studies. Further studies
may reveal functional relevance and the basis for some of the
poorly understood biological effects of NSAIDs.
While photo-NSAIDs interacted with and stabilized COX-2
in cellulo, COX-2 was only observed in PSMs at a FDR > 1%
that were thus filtered out from the proteomics data set (Table
S2). Furthermore, conjugated peptides from only a fraction of
the enriched proteins were identified. These detection
differences reflect the increased challenge in seeking to perform
site-specific identification of a single conjugated peptide as
opposed to protein identification that may derive from multiple
peptides from the same protein. Deeper analysis of the NSAID
interactome may be obtained by increasing protein inputs,
increasing chromatographic separation, or cellular fractionation
methods to deplete abundant proteins. Alternatively, applica-
tion of additional MS fragmentation methods, use of a second
protease (e.g., chymotrypsin), or alternative spectral assignment
method would provide additional conjugated peptide assign-
ments.
Figure 5. Validation of photo-NSAID binding interactions identified
by SIM-PAL. (A) Competitive displacement of photo-NSAIDs by the
parent compound. Jurkat cell lysates conjugated to the indicated
compound with or without the parent compound were clicked with
the biotin probe 10, captured on streptavidin−agarose, and probed for
the indicated protein. (B) Cellular thermal shift assay performed on
Jurkat cells in the presence of the indicated compound, probed for
histone H2A. Densitometry measurement of the Western blot data
was used to estimate Tagg (°C). Data are representative of two
independent biological replicates.
SIM-PAL represents the culmination of advances in chemical
enrichment strategies coupled to MS technology and a
computational pattern searching algorithm to lay the ground-
work for rapid progress in direct structural characterization of
small molecule binding sites within the whole proteome.
representing captured photo-NSAID binding sites. These
conjugated peptides mapped to proteins that were enriched
in the proteomic data, but not always classified as statistically
H
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(10) Mackinnon, A. L.; Taunton, J. Current Protocols in Chemical
Biology; Wiley: Hoboken, NJ, 2009; Vol. 1, pp55−73.10.1002/
9780470559277.ch090167
(11) Woo, C. M.; Iavarone, A. T.; Spiciarich, D. R.; Palaniappan, K.
K.; Bertozzi, C. R. Nat. Methods 2015, 12, 561.
(12) Parker, C. G.; Galmozzi, A.; Wang, Y.; Correia, B. E.; Sasaki, K.;
Joslyn, C. M.; Kim, A. S.; Cavallaro, C. L.; Lawrence, R. M.; Johnson,
S. R.; Narvaiza, I.; Saez, E.; Cravatt, B. F. Cell 2017, 168, 527.
(13) Zha, S.; Yegnasubramanian, V.; Nelson, W. G.; Isaacs, W. B.; De
Marzo, A. M. Cancer Lett. 2004, 215, 1.
Recent work in profiling small molecule modification sites has
begun to drastically expand the number of interactions
measured from the complex proteome and enable a deeper
understanding of the molecular underpinnings of polypharma-
cology. SIM-PAL revealed the global interaction map for the
three NSAIDs profiled and is readily translated to other
clinically relevant agents. For example, the immunomodulatory
drugs have widely established pluripotent activity, and the
mechanism of action is only partly understood.3 Alternatively,
metformin is widely used to treat diabetes with little
understanding of the underlying mechanism.50 SIM-PAL is
poised for broad application to bioactive small molecules for
identification of proteomic interactions and binding site
hotspots using an unbiased whole cell assay.
(14) Stewart, W. F.; Kawas, C.; Corrada, M.; Metter, E. J. Neurology
1997, 48, 626.
(15) Vane, J. R. Nat. New Biol. 1971, 231, 232.
(16) Tegeder, I.; Pfeilschifter, J.; Geisslinger, G. FASEB J. 2001, 15,
2057.
(17) Hanif, R.; Pittas, A.; Feng, Y.; Koutsos, M. I.; Qiao, L.; Staiano-
Coico, L.; Shiff, S. I.; Rigas, B. Biochem. Pharmacol. 1996, 52, 237.
(18) Jones, M. K.; Wang, H. T.; Peskar, B. M.; Levin, E.; Itani, R. M.;
Sarfeh, I. J.; Tarnawski, A. S. Nat. Med. 1999, 5, 1418.
(19) Yin, M.-J.; Yamamoto, Y.; Gaynor, R. B. Nature 1998, 396, 77.
(20) Smith, C. E.; Soti, S.; Jones, T. A.; Nakagawa, A.; Xue, D.; Yin,
H. Cell Chem. Biol. 2017, 24, 281.
(21) Mitchell, J. A.; Akarasereenont, P.; Thiemermann, C.; Flower, R.
J.; Vane, J. R. Proc. Natl. Acad. Sci. U. S. A. 1993, 90, 11693.
(22) Flower, R. J. Nat. Rev. Drug Discovery 2003, 2, 179.
(23) Schnitzer, T. J.; Burmester, G. R.; Mysler, E.; Hochberg, M. C.;
Doherty, M.; Ehrsam, E.; Gitton, X.; Krammer, G.; Mellein, B.;
Matchaba, P.; Gimona, A.; Hawkey, C. J.; Grp, T. S. Lancet 2004, 364,
665.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.7b11639.
Supplementary figures and experimental details. The
mass spectrometry proteomics data have been deposited
to the ProteomeXchange Consortium via the PRIDE
partner repository with the data set identifier
PXD007094 (PDF)
Supplementary tables (XLSX)
(24) McGettigan, P.; Henry, D. JAMA-J. Am. Med. Assoc 2006, 296,
1633.
AUTHOR INFORMATION
■
(25) Kalgutkar, A. S.; Crews, B. C.; Rowlinson, S. W.; Marnett, A. B.;
Kozak, K. R.; Remmel, R. P.; Marnett, L. J. Proc. Natl. Acad. Sci. U. S.
A. 2000, 97, 925.
(26) Chandna, N.; Kumar, S.; Kaushik, P.; Kaushik, D.; Roy, S. K.;
Gupta, G. K.; Jachak, S. M.; Kapoor, J. K.; Sharma, P. K. Bioorg. Med.
Chem. 2013, 21, 4581.
(27) Kurumbail, R. G.; Stevens, A. M.; Gierse, J. K.; McDonald, J. J.;
Stegeman, R. A.; Pak, J. Y.; Gildehaus, D.; iyashiro, J. M.; Penning, T.
D.; Seibert, K.; Isakson, P. C.; Stallings, W. C. Nature 1996, 384, 644.
(28) Orlando, B. J.; Malkowski, M. G. Acta Crystallogr., Sect. F: Struct.
Biol. Commun. 2016, 72, 772.
(29) Li, Z.; Hao, P.; Li, L.; Tan, C. Y. J.; Cheng, X.; Chen, G. Y. J.;
Sze, S. K.; Shen, H.-M.; Yao, S. Q. Angew. Chem., Int. Ed. 2013, 52,
8551.
Corresponding Author
*cwoo@chemistry.harvard.edu
ORCID
Christina M. Woo: 0000-0001-8687-9105
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Dr. Bogdan Budnik (Harvard Proteomics Resource
Laboratory) for acquiring MS data. Financial support from the
Burroughs Wellcome Fund (C.M.W.), the Uehara Memorial
Foundation Postdoctoral Fellowship (Y.A.), Murata Overseas
Scholarship Foundation (Y.A.), and Harvard University is
gratefully acknowledged.
(30) Seibert, K.; Zhang, Y.; Leahy, K.; Hauser, S.; Masferrer, J.;
Perkins, W.; Lee, L.; Isakson, P. Proc. Natl. Acad. Sci. U. S. A. 1994, 91,
12013.
(31) Schneidman-Duhovny, D.; Inbar, Y.; Nussinov, R.; Wolfson, H.
J. Nucleic Acids Res. 2005, 33, W363.
REFERENCES
■
́
(32) Iniguez, M. A.; Punzon, C.; Fresno, M. J. Immunol 1999, 163,
̃
(1) Lounkine, E.; Keiser, M. J.; Whitebread, S.; Mikhailov, D.;
Hamon, J.; Jenkins, J. L.; Lavan, P.; Weber, E.; Doak, A. K.; Cote, S.;
Shoichet, B. K.; Urban, L. Nature 2012, 486, 361.
111.
(33) Jafari, R.; Almqvist, H.; Axelsson, H.; Ignatushchenko, M.;
Lundback, T.; Nordlund, P.; Molina, D. M. Nat. Protoc. 2014, 9, 2100.
̈
(2) Jana, N. R. Cell. Mol. Life Sci. 2008, 65, 1295.
(3) Zhu, Y. X.; Kortuem, K. M.; Stewart, A. K. Leuk. Lymphoma
2013, 54, 683.
(4) Pasternak, G. W. Neuropharmacology 2014, 76, 198.
(5) Hopkins, A. L. Nat. Chem. Biol. 2008, 4, 682.
(6) Keiser, M. J.; Setola, V.; Irwin, J. J.; Laggner, C.; Abbas, A. I.;
Hufeisen, S. J.; Jensen, N. H.; Kuijer, M. B.; Matos, R. C.; Tran, T. B.;
Whaley, R.; Glennon, R. A.; Hert, J.; Thomas, K. L. H.; Edwards, D.
D.; Shoichet, B. K.; Roth, B. L. Nature 2009, 462, 175.
(7) Paolini, G. V.; Shapland, R. H. B.; van Hoorn, W. P.; Mason, J. S.;
Hopkins, A. L. Nat. Biotechnol. 2006, 24, 805.
(34) Zybailov, B.; Mosley, A. L.; Sardiu, M. E.; Coleman, M. K.;
Florens, L.; Washburn, M. P. J. Proteome Res. 2006, 5, 2339.
(35) Choi, H.; Fermin, D.; Nesvizhskii, A. I. Mol. Cell. Proteomics
2008, 7, 2373.
(36) Zhang, Y.; Wen, Z.; Washburn, M. P.; Florens, L. Anal. Chem.
2015, 87, 4749.
(37) Zhou, D.; Zhang, Q.; Lu, W.; Xia, Q.; Wei, S. J. Clin. Pharmacol.
1998, 38, 625.
(38) Dharmapuri, G.; Doneti, R.; Philip, G. H.; Kalle, A. M. Leuk. Res.
2015, 39, 696.
(39) Thul, P. J.; Åkesson, L.; Wiking, M.; Mahdessian, D.; Geladaki,
(8) Schenone, M.; Dancik, V.; Wagner, B. K.; Clemons, P. A. Nat.
A.; Ait Blal, H.; Alm, T.; Asplund, A.; Bjork, L.; Breckels, L. M.;
̈
Chem. Biol. 2013, 9, 232.
Backstrom, A.; Danielsson, F.; Fagerberg, L.; Fall, J.; Gatto, L.; Gnann,
(9) Gertsik, N.; am Ende, C. W.; Geoghegan, K. F.; Nguyen, C.;
Mukherjee, P.; Mente, S.; Seneviratne, U.; Johnson, D. S.; Li, Y. M.
Cell Chem. Biol. 2017, 24, 3.
̈
̈
C.; Hober, S.; Hjelmare, M.; Johansson, F.; Lee, S.; Lindskog, C.;
Mulder, J.; Mulvey, C. M.; Nilsson, P.; Oksvold, P.; Rockberg, J.;
I
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Schutten, R.; Schwenk, J. M.; Sivertsson, Å.; Sjostedt, E.; Skogs, M.;
̈
Stadler, C.; Sullivan, D. P.; Tegel, H.; Winsnes, C.; Zhang, C.;
Zwahlen, M.; Mardinoglu, A.; Ponten
́
, F.; von Feilitzen, K.; Lilley, K.
S.; Uhlen, M.; Lundberg, E. Science 2017, 356, eaal3321.
́
(40) Ruepp, A.; Waegele, B.; Lechner, M.; Brauner, B.; Dunger-
Kaltenbach, I.; Fobo, G.; Frishman, G.; Montrone, C.; Mewes, H. W.
Nucleic Acids Res. 2010, 38, D497.
(41) Palaniappan, K. K.; Pitcher, A. A.; Smart, B. P.; Spiciarich, D. R.;
Iavarone, A. T.; Bertozzi, C. R. ACS Chem. Biol. 2011, 6, 829.
(42) Woo, C. M.; Felix, A.; Byrd, W. E.; Zuegel, D. K.; Ishihara, M.;
Azadi, P.; Iavarone, A. T.; Pitteri, S. J.; Bertozzi, C. R. J. Proteome Res.
2017, 16, 1706.
(43) Bern, M.; Kil, Y. J.; Becker, C. Current Protocols in Bioinformatics;
Wiley: Hoboken, NJ, 2012; Chpt 13, pp 13.20.1−13.20.14
(44) Tsunaka, Y.; Kajimura, N.; Tate, S.-i.; Morikawa, K. Nucleic Acids
Res. 2005, 33, 3424.
(45) Lu, J.; Qian, Y.; Altieri, M.; Dong, H.; Wang, J.; Raina, K.; Hines,
J.; Winkler, J. D.; Crew, A. P.; Coleman, K.; Crews, C. M. Chem. Biol.
2015, 22, 755.
(46) Schreiber, S. Science 1991, 251, 283.
(47) Kronke, J.; Udeshi, N. D.; Narla, A.; Grauman, P.; Hurst, S. N.;
̈
McConkey, M.; Svinkina, T.; Heckl, D.; Comer, E.; Li, X.; Ciarlo, C.;
Hartman, E.; Munshi, N.; Schenone, M.; Schreiber, S. L.; Carr, S. A.;
Ebert, B. L. Science 2014, 343, 301.
(48) Lu, G.; Middleton, R. E.; Sun, H.; Naniong, M.; Ott, C. J.;
Mitsiades, C. S.; Wong, K.-K.; Bradner, J. E.; Kaelin, W. G. Science
2014, 343, 305.
(49) Kragh-Hansen, U. Pharmacol Rev. 1981, 33, 17.
(50) Pernicova, I.; Korbonits, M. Nat. Rev. Endocrinol. 2014, 10, 143.
J
DOI: 10.1021/jacs.7b11639
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX