Global Mapping of Protein–Lipid Interactions by Using Modified Choline-Containing Phospholipids Metabolically Synthesized in Live Cells

Article abstract of DOI:10.1002/anie.201702509

The protein–lipid interaction is an essential metabolic process that mediates cellular signaling and functions. Existing strategies for large-scale mapping studies of the protein–lipid interaction fall short in their incompatibility with metabolic incorporation or inability to remove unwanted interferences from lipidated proteins. By incorporating an alkyne-containing choline head group and a diazirine-modified fatty acid simultaneously into choline-containing phospholipids synthesized from live mammalian cells, protein–phospholipid interactions have been successfully imaged in live cells. Subsequent in situ profiling of the modified Cho phospholipid-crosslinked proteins followed by quantitative proteomics allowed identification of several hundred putative phospholipid-interacting proteins, some of which were further validated.

Full text of DOI:10.1002/anie.201702509

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201702509
German Edition: DOI: 10.1002/ange.201702509
Protein–Lipid Interactions
Global Mapping of Protein–Lipid Interactions by Using Modified
Choline-Containing Phospholipids Metabolically Synthesized in Live
Cells
Danyang Wang, Shubo Du, Amaury Cazenave-Gassiot, Jingyan Ge, Jun-Seok Lee,
Markus R. Wenk, and Shao Q. Yao*
Abstract: The protein–lipid interaction is an essential meta-
bolic process that mediates cellular signaling and functions.
Existing strategies for large-scale mapping studies of the
protein–lipid interaction fall short in their incompatibility
with metabolic incorporation or inability to remove unwanted
interferences from lipidated proteins. By incorporating an
alkyne-containing choline head group and a diazirine-modi-
fied fatty acid simultaneously into choline-containing phos-
pholipids synthesized from live mammalian cells, protein–
phospholipid interactions have been successfully imaged in live
cells. Subsequent in situ profiling of the modified Cho phos-
pholipid-crosslinked proteins followed by quantitative proteo-
mics allowed identification of several hundred putative phos-
pholipid-interacting proteins, some of which were further
validated.
[
1]
Figure 1. A) Structure of a typical phosphatidylcholine (PC) in mamma-
lian cells. B) Metabolically synthesized unnatural bifunctional PC (9)
used in the current study.
L
ipids are highly abundant cellular metabolites. In eukary-
otic cells, cholesterol and phospholipids serve as major
components of biological membranes. Amongst various
phospholipids, choline-containing phospholipids are the
most abundant, with the major form, phosphatidylcholine
needed. Of various techniques used to study protein–lipid
interactions, the use of photoactivatable lipids offers the key
advantage of light-controlled conversion of transient non-
covalent interactions into covalent isolable complexes, thus
enabling large-scale enrichment/identification of lipid-bind-
(
PC; Figure 1), alone comprising 50% of the entire phospho-
lipid pool. Besides being essential cellular building blocks,
lipids participate in various functions including post-transla-
[
2]
tional modification (that is, protein lipidation). By interact-
ing with cellular proteins, lipids modulate their subcellular
[
5–8]
ing proteins by mass spectrometry (MS).
The photo-
[3]
localization and activities. Dysregulation of such interac-
crosslinking may be carried out in live mammalian cells that
recapitulate the actual environment where protein–lipid
interactions take place. Thiel et al. first reported the labeling
of mammalian cells with diazirine-containing [3H]-photo-
cholesterol to study cholesterol-binding proteins (Supporting
[
4]
tions leads to numerous pathological conditions. Chemical
tools capable of proteome-wide elucidation of protein–lipid
interactions, especially those that are distinguishable among
different classes of lipids (for example, protein–phospholipid
interactions from other lipid interactions), are urgently
[6]
Information, Figure S2). Armed with a clickable alkyne tag,
analogues of this synthetic sterol were subsequently used to
[
7]
[*] D. Wang, S. Du, Prof. Dr. S. Q. Yao
map protein–cholesterol interactions. Niphakis et al. used
alkyne-containing diazirine-modified fatty acids to carry out
proteome-wide profiling of potential lipid-binding proteins
Department of Chemistry, National University of Singapore
3
Science Drive 3, Singapore 117543 (Singapore)
E-mail: chmyaosq@nus.edu.sg
[8]
and assess their ligandability. By using a minimalist design
Dr. A. Cazenave-Gassiot, Prof. Dr. M. R. Wenk
Department of Biological Sciences
National University of Singapore (Singapore)
principle to minimize structural perturbations in natural
[9]
lipids, these approaches were able to capture a large number
of putative protein–lipid interactions, but they fell short when
compared with metabolic strategies that use lipids made by
the cellsꢀ own machinery. Most metabolic lipids are first
synthesized in endoplasmic reticulum (ER) and Golgi, then
sorted to various subcellular compartments, a process with
Dr. J. Ge
Institute of Bioengineering
Zhejiang University of Technology (China)
Prof. Dr. J.-S. Lee
Department of Biological Chemistry
University of Science & Technology (South Korea)
[
1–3]
which synthetic lipids cannot easily emulate.
While many
photoactivatable metabolites have been developed to char-
Supporting information for this article can be found under:
https://doi.org/10.1002/anie.201702509.
[
10,11]
acterize protein interactions,
few were well-suited for
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
protein–lipid interactions and none could be used for in situ
mapping of protein-phospholipid interactions exclusively.
Finally, biosynthesis of the bifunctional 9 in doubly treated
cells was confirmed by PreIS (m/z 208), shifts in LC retention
time, NL (m/z 28) and extracted ion chromatography (EIC).
On the basis of these results (Figure 2A), we estimated 9 was
present in about 10% of all PC species in the doubly labeled
cells. Individually incorporated PC species (that is, 2- or 6-
modified PC species) were detected in the experiment but
they did not cause interference in downstream experiments
(see below). Surprisingly, we also detected a small amount of
doubly incorporated sphingomyelin (SM) in (2 + 6)-treated
cells (Supporting Information, Figure S10). Previous work
with pacFA showed it was not incorporated into SM
biosynthesis. Since SM occurs naturally in less than 10%
of mammalian phospholipids, we concluded the presence of
these modified SM species in our doubly treated cells should
minimally interfere with our subsequent studies of protein–
PC interactions. Finally, attempts were made to feed cells
simultaneously with 6 and 17-ODYA (that is, 8 in the
Supporting Information, Table S1), which in principle
should enable biosynthesis of bifunctional glycerophospholi-
pids (including PCs) capable of photo-crosslinking/enrich-
ment of lipid-binding proteins (Supporting Information,
Figure S11); even at 100 mm incorporation concentrations,
minimal UV-dependent photo-crosslinked proteins were
detected, indicating few doubly modified glycerophospholi-
pids could be metabolically synthesized this way.
[12]
A bifunctional fatty acid, pacFA (Supporting Information,
Figure S2), was recently reported to be incorporated into
phospholipid biosynthesis, and the resulting lipids were used
to identify more than 200 putative lipid-binding proteins from
[
12]
mammalian cells. This strategy is a step forward compared
to previous strategies using synthetic phospholipids, which
were only compatible with artificially constructed mem-
[
5,13]
branes.
Notwithstanding, pacFA was shown to be meta-
bolically incorporated into different phospholipids as well as
[12]
many lipidated proteins even under optimized conditions.
[12]
Such results are expected as pacFA analogues are substrates
of palmitoylation and have been used to map lipidated
[
14]
proteins and their interactions. In cells, palmitoylation is
dynamic and occurs minutes after proteins are synthe-
[
15,16]
sized.
Realizing the biological importance of PC, and
a lack of reliable metabolic methods to map specific protein–
PC interactions, we report herein a double incorporation
strategy that has enabled the successful biosynthesis of
bifunctional PC (9) in cells, followed by in situ photo-cross-
linking and in vitro large-scale proteomics to identify many
high-confidence PC-binding proteins.
To eliminate enrichment of lipidated proteins and their
binders (after photo-crosslinking), pacFA and other bifunc-
tional fatty acids must be avoided. In our strategy (Fig-
ure 1B), they were replaced with a structurally optimized
diazirine-containing fatty acid, 16-diazFA (6). Albeit a sub-
strate for protein lipidation, metabolic products of 6 and their
photo-crosslinked binders would not be detected/enriched in
subsequent imaging/proteomics experiments owing to a lack
of clickable alkyne. Taking advantage of this, and together
with propargylcholine (2, a reported biosynthetic precursor of
With the successful detection of 9 in cellular lipid extracts,
we next subjected the same cells to UV irradiation, followed
by lysis, click chemistry with tetraethylrhodamine azide
(TER-azide), SDS-PAGE protein separation and in-gel
fluorescence scanning (Figure 2B); fluorescently labeled
proteins were detected only in UV-irradiated cells treated
with both 2 and 6 (lane 5), but not with either (lanes 2 and 3),
nor in non-UV-irradiated cells (lane 4), indicating they were
exclusively from PC-interacting proteins and free of lipidated
proteins and/or their binders. These results are thus a signifi-
[
17]
choline phospholipids ), we fed mammalian cells simulta-
neously with (2 + 6) to generate a sufficient pool of bifunc-
tional PC (9). Other diazirine-containing fatty acids (that is, 7
in the Supporting Information, Table S1) gave poorer results.
Choline analogues including diazirine choline (3) and
a bifunctional choline (4) were also tested, but neither was
successfully incorporated into PC biosynthesis (Supporting
Information, Figure S3). We first confirmed that 2 or 6
individually was metabolically incorporated in PC species
[
12,13]
cant improvement over previous methods.
By varying
probe concentrations and incorporation time, we determined
optimal double incorporation conditions as 500 and 100 mm of
2 and 6, respectively, for up to 16 h incorporation time
(Supporting Information, Figure S12). To further ensure the
fluorescently labeled bands were exclusively PC-binding
proteins, we conducted competitive experiments by adding
native choline (1) or palmitic acid (5) to (2 + 6)-treated cells.
A two-fold excess of each competitor was sufficient to
completely abolish the labeled proteins, indicating these
proteins required both major components of PC for positive
binding/photo-crosslinking. Additional competitive experi-
ments further showed the fluorescence intensities of these
labeled proteins were diminished in the presence of native PC
(Supporting Information, Figure S13). Finally, to test whether
our newly developed metabolic strategy could be used to
visualize protein–PC interactions in live cells, HeLa cells were
treated with (2 + 6), followed by UV irradiation, cell fixation,
permealization, and click chemistry, then imaged (Figure 2C;
Supporting Information, Figure S14). Similar to a previous
(
Supporting Information, Figures S4–S9); after feeding A431
cells with each probe, followed by analysis of cellular lipid
extracts for the presence of the corresponding modified PC by
liquid chromatography electrospray ionization tandem mass
spectrometry (LC-ESI-MS/MS), we found nearly 50% of
cellular PC species had been modified in each case. Precursor
ion scans (PreIS; at m/z 184 and 208) readily distinguished 2-
modified PC species from unmodified species. Significant
shifts in the LC retention time of modified PC species were
observed (from 4.0 to 3.6 min). Owing to isobaric interference
between modified PC and major unmodified PC in PreIS (m/z
1
84), for example, 6-modified PC 18:1/diazFA has m/z 786.6
which is the same as unmodified PC 36:2, additional scan of
neutral loss (NL m/z 28 to indicate the presence of diazirine)
was carried out (Supporting Information, Figure S5); these
data together with observed shifts in LC retention time by 6-
modified PC species indicated successful incorporation.
[
17]
report, non-UV-irradiated cells showed strong but deter-
gent-sensitive fluorescence signals throughout various intra-
cellular structures except the nucleus, indicating they were
2
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 2. A) MS analysis of cellular lipid extracts from (2 + 6)-treated A431 cells. Shifts of LC retention time in modified PC species versus
unmodified species were observed (from 4.0 to 3.6 min). PreIS (m/z 184 & 208) identified 2-incorporation (left), and new peaks in EIC of doubly
treated cells identified 6-incorporation (right). B) Global profiling of protein–PC interactions in (2+6)-treated A431 cells (500 and 100 mm of 2
and 6, respectively, for 16 h followed by UV irradiation, cell lysis, click chemistry with TER-azide, SDS-PAGE separation, and then in-gel
fluorescence scanning). The same gel was stained with Coomassie brilliant blue (right). C) Confocal imaging results of protein–PC interactions in
(2+6)-treated HeLa cells. After UV irradiation, cell fixation and permealization, cells were clicked with TER-azide, detergent-washed, and imaged.
Blue: Hoechst stain; red: TER channel. Scale bar: 10 mm.
[
12]
from choline phospholipids (top panels). With UV-irradiated
cells on the other hand, significant fluorescence signals
remained after detergent wash, indicating they were from
covalent, photo-crosslinked protein complexes of the bifunc-
tional PC (9). These fluorescence stains appeared to be
present in all major lipid-containing subcellular structures.
Having confirmed this newly developed metabolic
method was capable of imaging and gel-based profiling of
protein–PC interactions, we next evaluated its ability to map
lipid-interacting proteins. The pacFA method was the only
report capable of mapping protein–lipid interactions by using
(1 h) to minimize contamination of lipidated proteins, our
double incorporation strategy could be carried out with much
longer cell incubation time (16 h), thus providing significantly
improved hit coverage with few contaminating lipidated
proteins for global mapping of protein–lipid interactions.
Large-scale
photo-crosslinking/affinity
enrichment/
LCMS experiments are known to produce significant back-
[
18a,19]
ground proteins.
For example, the highly abundant
housekeeping protein GAPDH, which was identified as
a putative lipid-binding protein from above experiments
and the pacFA strategy, was likely a false positive. In an
effort to improve the mapping of genuine protein–PC
interactions, we combined this newly developed PC-labeling
approach with stable-isotope labeling by amino acids in cell
culture (SILAC) and liquid chromatography–tandem mass
spectrometry (LC-MS/MS; Supporting Information, Figur-
[12]
[
12]
biosynthetic lipids, and thus served as a direct comparison
to our studies. We carried out similar UV-dependent pro-
teome labeling in A431 cells, followed by click chemistry with
Rh-Biotin-N , affinity enrichment, and mass spectrometric
3
analysis (Figure 3A,B; Supporting Information, Figure S15).
The resulting LCMS data were similarly processed and
compared. To cross-check for possible lipidated proteins,
previously reported putative palmitoylated proteins were
[7,8,20]
es S17–S20).
Isotopically heavy cells doubly labeled with
(2 + 6) followed by UV irradiation served as the positive
experiments (+UV), and were treated with 1) doubly labeled
light cells but no UV irradiation (ÀUV; Figure 3D top); or
2) doubly labeled light cells in the presence of native choline
(150 mm) followed by UV irradiation (competitive SILAC;
Figure 3D, bottom). To account for differences in protein
abundance between two isotopically labeled cells, we fur-
thermore carried out pairs of experiments in which (2 + 6)
and SILAC labels were switched (termed forward and reverse
experiments, respectively; Supporting Information, Fig-
[
18]
used.
proteins (exclusively present in UV-irradiated cells), of which
8 were putative palmitoylated proteins. Many of these
We identified a total of 757 putative PC-binding
3
proteins have well-known subcellular localizations in major
lipid-containing compartments and participate in a wide
range of cellular processes (Supporting Information, Fig-
ure S16 and Section S3). Compared to the pacFA strategy,
which required a very short metabolic incorporation time
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 3. A) Workflow summarizing standard LCMS and SILAC experiments to map PC-interacting proteins. B) Numbers of proteins identified
from standard LCMS experiments. C) Subcellular distributions of putative PC-interacting proteins (1353) and high-confidence hits (261) identified
from SILAC experiments. D) SILAC ratio plots of forward (F1; top) and competitive F1 (bottom) experiments from (2+6)-treated A431 cells.
Right: Representative peptide MS1 chromatograms of selected targets showing relative labeling in “heavy” and “light” cells. E) Pull-down (PD)/
western blotting (WB) validation of (2+6)-treated cells to confirm protein–PC interactions. F) Confirmation of PC-PDI interaction by (top) UV-
dependent labeling of recombinant PDI with biosynthesized 9 isolated from (2+6)-treated cells, and (bottom) insulin aggregation assay to test
À1
PDI activity in the presence of PC (0 to 20 mgmL ). G) PD/WB to validate lipidation (lanes 1 and 2) and protein-PC interaction (lanes 3 and 4)
[
12,13,18]
from cells fed with 17-ODYA and (2+6), respectively. H) Summary of high-confidence hits and their lipidation/lipid-binding status.
ures S17,S19). All experiments were conducted in duplicate.
We identified a total of 1353 putative PC-binding targets,
defined as UV-enriched proteins (SILAC ratio of ꢀ 2 or ꢁ0.5
in forward/reverse experiments, respectively), of which 261
were deemed high-confidence hits. They were further ranked
based on the number of occurrence as positives in competitive
SILAC experiments (Figure 3D, bottom; Supporting Infor-
mation, Figure S19, Table S2).
LCMS and SILAC experiments, were also successfully
enriched from UV-irradiated cells, indicating they are true
lipid-binding proteins. These two proteins are not previously
known lipid-binding proteins. PDI is a well-known compo-
nent of the microsomal lipid transfer protein (MTP) complex
that catalyzes the transport of a variety of lipids including
[23]
PC.
A direct PDI–lipid interaction has however not
experimentally established. We first carried out a docking
experiment, based on a published PDI X-ray structure to
confirm that PC indeed was able to bind to the ligand-binding
site in PDI (Supporting Information, Figure S22). Next, we
were able to detect direct and specific PDI-PC interaction by
successfully cross-linking recombinant PDI with the bifunc-
tional PC (9) isolated from (2 + 6)-treated A431 cells (Fig-
ure 3F, top); covalent labeling between 9 and PDI was UV-
dependent but abolished in the presence of excessive PC.
Interestingly, the enzymatic activity of PDI was also affected
by the presence of PC (Figure 3F, bottom), indicating this
newly discovered protein–PC interaction might play a func-
To confirm the data quality of our SILAC experiments in
mapping genuine protein–PC interactions, representative
positive hits were further validated (Figures 3E–H; Support-
ing Information, Figures S21,S22). Amongst these hits,
MTCO2 (mitochondrially encoded cytochrome C oxidase
II) and StarD7 (a member of the lipid transport protein
family comprising StarD2/D7/D10) are known PC-interacting
[
7,12,13,21]
proteins.
Correspondingly, in PD/WB validation
experiments with (2 + 6)-treated cells, with or without UV
irradiation, both were successfully enriched (Figures 3E;
Supporting Information, Figure S21). Similarly, StarD2/D10,
despite being absent in our SILAC hits owing to their almost
[
24]
tional role in PDI biology. Four more positively identified
PC-binding proteins from our SILAC experiments were
further validated for their lipid-binding and protein lipidation
properties (Figure 3G,H). They were chosen because con-
flicting reports from previous studies showed their putative
lipid-binding property might have been caused by contami-
nation from the aforementioned protein lipidation problems
[
22]
2
5 kDa molecular weights, were also enriched (Supporting
[12]
Information, Figure S21). GAPDH, identified previously
and in our standard LCMS as a putative lipid-binding protein,
was deemed negative from our SILAC experiments (Fig-
ure 3D). Indeed, it was not successfully enriched in PD/WB
validation experiments (Figure 3E). Interestingly, PDI (pro-
tein disulfide isomerase) and AIF (apoptosis-inducing
factor 1), two positive hits identified from both standard
[
12,18]
(that is, CANX/LYRIC).
By labeling cells with 17-ODYA
[18a]
(a previously reported palmitoylation marker ), or (2 + 6)
4
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
These are not the final page numbers!

Angewandte
Communications
Chemie
with UV irradiation, we confirmed all four proteins were
indeed lipid-binding proteins, of which LETM/CANX were
[4] M. P. Wymann, R. Schneiter, Nat. Rev. Mol. Cell Biol. 2008, 9,
62 – 176.
[5] Y. Xia, L. Peng, Chem. Rev. 2013, 113, 7880 – 7929.
[6] C. Thiele, M. J. Hannah, F. Fahrenholz, W. B. Huttner, Nat. Cell
Biol. 2000, 2, 42 – 49.
1
[
12,18a]
palmitoylated as previously reported,
but LYRIC/
[
6]
MTCO2 were not. Finally, a previously reported 31 kDa
PC-specific protein band was further identified and confirmed
to be likely prohibitin-2 (PHD) and voltage-dependent anion-
selective channel protein 1 (VDAC1; Supporting Informa-
tion, Figure S23).
[
7] J. J. Hulce, A. B. Cognetta, M. J. Niphakis, S. E. Tully, B. F.
Cravatt, Nat. Methods 2013, 10, 259 – 264.
[8] M. J. Niphakis, K. M. Lum, A. B. Cognetta III, B. E. Correia, T.-
A. Ichu, J. Olucha, S. J. Brown, S. Kundu, F. Piscitelli, H. Rosen,
B. F. Cravatt, Cell 2015, 161, 1668.
[6–8,12,13]
Compared to established methods,
our newly
[
9] Z. Li, L. Li, P. Hao, C. Y. J. Tan, X. Cheng, G. Y. J. Chen, S. K.
developed double incorporation strategy for metabolic syn-
thesis of bifunctional PCs offered obvious improvement for
global mapping of genuine protein–lipid interactions in live
mammalian cells. From this work, we have successfully
uncovered many new protein–lipid interactions, some of
which have been further validated. It should be highlighted
that our method is amenable to studies of protein–lipid
interactions in which proteins bind to PC via the fatty acid
Sze, H.-M. Shen, S. Q. Yao, Angew. Chem. Int. Ed. 2013, 52,
8551 – 8556; Angew. Chem. 2013, 125, 8713 – 8718.
[
10] L. Davis, J. W. Chin, Nat. Rev. Mol. Cell Biol. 2012, 13, 168 – 182.
[11] N. D. Pham, R. B. Park, J. J. Kohler, Curr. Opin. Chem. Biol.
2013, 17, 90 – 101.
12] P. Haberkant, R. Raijmakers, M. Wildwater, T. Sachsenheimer,
B. Brꢁgger, K. Maeda, M. Houweling, A.-C. Gavin, C. Schultz,
G. van Meer, A. J. R. Heck, J. C. M. Holthuis, Angew. Chem. Int.
Ed. 2013, 52, 4033 – 4038; Angew. Chem. 2013, 125, 4125 – 4130.
13] J. Gubbens, E. Ruijter, L. E. V. de Fays, J. M. A. Damen, B.
de Kruijff, M. Slijper, D. T. S. Rijkers, R. M. J. Liskamp,
A. I. P. M. de Kroon, Chem. Biol. 2009, 16, 3 – 14.
[
[
25]
tails. Alternative strategies will have to be developed in
future to uncover interactions that involve proteins binding to
the choline head group instead. With an emerging report that
chemically modified inositols are biosynthetic precursors of
[
[14] a) T. Peng, H. C. Hang, J. Am. Chem. Soc. 2015, 137, 556 – 559;
b) For a recent review, see H. C. Hang, Chem. Rev. 2011, 111,
[
26]
phosphoinositols (PIs), our method should be amenable to
the biosynthesis of bifunctional PIs.
6
341 – 6385.
15] K. Liu, P.-Y. Yang, Z. Na, S. Q. Yao, Angew. Chem. Int. Ed. 2011,
0, 6776 – 6781; Angew. Chem. 2011, 123, 6908 – 6913.
[
[
[
[
[
[
5
16] M. M. Zhang, L. K. Tsou, G. Charron, A. S. Raghavan, H. C.
Hang, Proc. Natl. Acad. Sci. USA 2010, 107, 8627 – 8632.
17] C. Y. Jao, M. Roth, R. Welti, A. Salic, Proc. Natl. Acad. Sci. USA
2009, 106, 15332 – 15337.
18] a) B. R. Martin, B. F. Cravatt, Nat. Methods 2009, 6, 135 – 138;
b) http://www.uniprot.org.
19] Z. Li, D. Wang, L. Li, S. Pan, Z. Na, C. Y. J. Tan, S. Q. Yao, J. Am.
Chem. Soc. 2014, 136, 9990 – 9998.
20] R. E. Kleiner, P. Verma, K. R. Molloy, B. T. Chait, T. M. Kapoor,
Nat. Chem. Biol. 2015, 11, 807 – 814.
21] Y. Horibata, H. Sugimoto, J. Biol. Chem. 2010, 285, 7358 – 7365.
22] Enriched proteins were gel-separated prior to tryptic digest/
peptide extraction. The gel slices near the 25 kDa regions, where
avidin contamination was severe, were cut off. As a result,
potential protein hits having MW of 20–30 kDa would be
excluded from our MS results.
Acknowledgements
Funding was provided by the Ministry of Education
(
2
MOE2012-T2-2-051), Singapore, and KIST (2E26632/
E26110).
Conflict of interest
[
[
The authors declare no conflict of interest.
Keywords: click chemistry · phosphatidylcholine ·
photo-crosslinking · protein–lipid interactions ·
quantitative proteomics
[
23] J. R. Wetterau, K. A. Combs, S. N. Spinner, B. J. Joiner, J. Biol.
Chem. 1990, 265, 9800 – 9807.
[
[
24] L. Ellgaard, L. W. Ruddock, EMBO Rep. 2005, 6, 28 – 32.
25] A.-E. Saliba, I. Vonkova, A.-C. Gavin, Nat. Rev. Mol. Cell Biol.
2015, 16, 753 – 761.
[
[
1] M. R. Wenk, Cell 2010, 143, 888 – 895.
2] A. Shevchenko, K. Simons, Nat. Rev. Mol. Cell Biol. 2010, 11,
[26] L. Lu, J. Gao, Z. Guo, Angew. Chem. Int. Ed. 2015, 54, 9679 –
9682; Angew. Chem. 2015, 127, 9815 – 9818.
5
93 – 598.
3] T. Peng, X. Yuan, H. C. Hang, Curr. Opin. Chem. Biol. 2014, 21,
44 – 153.
[
Manuscript received: March 9, 2017
Final Article published: && &&, &&&&
1
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Protein–Lipid Interactions
Double the load: A novel double incor-
poration strategy for metabolic biosyn-
thesis of bifunctional choline-containing
phospholipids (for example phosphati-
dylcholine (PC); see picture) was devel-
oped. This strategy offers significant
improvement for global mapping of gen-
uine protein–lipid interactions from live
mammalian cells.
D. Wang, S. Du, A. Cazenave-Gassiot,
J. Ge, J.-S. Lee, M. R. Wenk,
S. Q. Yao*
&&&& — &&&&
Global Mapping of Protein–Lipid
Interactions by Using Modified Choline-
Containing Phospholipids Metabolically
Synthesized in Live Cells
6
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 6
These are not the final page numbers!